Merge branch 'main' into feature/chromastack

# Description

<!--
> Please provide a summary of the changes made in this PR. Include
details such as:
  > * What issue does this PR address or fix?
  > * What new features or enhancements does this PR introduce?
> * Are there any breaking changes or dependencies that need to be
considered?
-->

Windows traditionally limits many file operations to `MAX_PATH` (260
characters). Deeply nested plugin packages can exceed this limit during
`.whl` extraction, leaving files missing and causing the plugin to fail
to load.

This change adds a shared extended_path helper that prepares paths for
Windows file operations using the extended-length prefixes `\\?\` or
`\\?\UNC\`. The helper is used for ZIP archive I/O and extraction
destinations, including plugin wheel extraction. This avoids relying on
the Windows long-path setting, which changes machine-wide settings.

Paths remain unprefixed outside file operations. Code that bypasses the
shared ZIP helpers may still require separate long-path handling.

See [Microsoft's documentation on path length
limitation](https://learn.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation?tabs=registry)

# Screenshots/Recordings/Graphs

<!--
> Please attach relevant screenshots to showcase the UI changes.
> Please attach images that can help explain the changes.
-->

## Tests

<!--
> Please describe the tests that you have conducted to verify the
changes made in this PR.
-->
Tested on Windows 11 by using a portable installation of OrcaSlicer and
placing it in some really long filepath length e.g.
`C:/somename/<240 characters long folder path>/OrcaSlicer` and then
running and trying to install any plugin locally or on OrcaCloud.

<!--
> A guide for users on how to download the artifacts from this PR.
-->

[How to Download Pull Requests Artifacts for
Testing](https://www.orcaslicer.com/wiki/how_to_download_pr_artifacts)

.claude/skills/orca-wxwidgets/references/platforms.md

@@ -569,12 +569,18 @@ on the **first** call. So:
 #endif
 ```
 
-**Before GTK starts**, only the environment exists. `CLI::run` (`src/OrcaSlicer.cpp`) branches on
-`GDK_BACKEND` (an `x11` prefix means the X11 opt-in). On the default path it forces `GDK_BACKEND=x11`
-when wx lacks EGL and `WAYLAND_DISPLAY` is set. It sets `WEBKIT_DISABLE_COMPOSITING_MODE=1`
-(non-replacing) only when both `DISPLAY` and `WAYLAND_DISPLAY` are set, and calls `XInitThreads()`
+**Before GTK starts**, only the environment exists. `CLI::run` (`src/OrcaSlicer.cpp`) sets the NVIDIA
+DMABUF workaround and then branches on `GDK_BACKEND` (an `x11` prefix means the X11 opt-in). On both
+paths, when `/proc/driver/nvidia/version` exists and the linked WebKitGTK honours the switch (>= 2.44,
+`WebView::SupportsForceSharedMemory()`), it sets `WEBKIT_DMABUF_RENDERER_FORCE_SHM=1` and
+`WEBKIT_FORCE_DMABUF_RENDERER=1` (non-replacing): the proprietary driver cannot back WebKit's hardware
+DMABUF renderer. On the default path it forces `GDK_BACKEND=x11` when wx lacks EGL and `WAYLAND_DISPLAY`
+is set, and sets `WEBKIT_DISABLE_COMPOSITING_MODE=1` (non-replacing) inside that branch when the linked
+WebKitGTK is older than 2.42: a forced X11 backend in a Wayland session is the only default-path case
+that runs through XWayland, which `DISPLAY` and `WAYLAND_DISPLAY` alone do not tell you (a native
+Wayland app inherits both). It calls `XInitThreads()`
 only when `DISPLAY` is set. With neither variable set, the GUI refuses to start ("Neither DISPLAY nor
-WAYLAND_DISPLAY set"). The details and the WebKit rule (c12912e0df) are in
+WAYLAND_DISPLAY set"). The details and the WebKit rules (c12912e0df) are in
 `references/webview-gl-aui-media.md` §WebKitGTK on Linux sessions.
 
 **wx's own detectors**, for reference:

.claude/skills/orca-wxwidgets/references/webview-gl-aui-media.md

@@ -61,7 +61,8 @@ available; GTK2/WebKit1 is an opt-out configuration the GUI does not support.
 11. A web host that re-themes in place handles `EVT_WEBVIEW_RECREATED` without `Skip()`; one that
     needs a reload lets it through. §[Orca wrapper](#orca-webview-wrapper-widgetswebview)
 12. Keep `WebViewWebKit`'s destructor removing the `"wx"` handler. §[Orca wrapper](#orca-webview-wrapper-widgetswebview)
-13. Do not widen `WEBKIT_DISABLE_COMPOSITING_MODE` beyond Orca's default-path XWayland case, and never
+13. Do not widen `WEBKIT_DISABLE_COMPOSITING_MODE` beyond the no-EGL default path where Orca forces
+    `GDK_BACKEND=x11`, for WebKitGTK older than 2.42, and never
     make a page's progress depend only on a C++→JS callback. §[WebKitGTK on Linux](#webkitgtk-on-linux-sessions)
 14. GL attribute lists: legacy `int[]` lists end with `0` and spell out `WX_GL_RGBA` and
     `WX_GL_DOUBLEBUFFER`; `wxGLAttributes`/`wxGLContextAttrs` end with `EndList()`; MSAA is requested
@@ -567,33 +568,69 @@ vue-resize workaround script and sets a persistent SQLite cookie store on the vi
 ## WebKitGTK on Linux sessions
 
 **Mechanism.** `CLI::run` (`src/OrcaSlicer.cpp`, under `#ifdef __WXGTK__`, before GTK initialises)
-branches on `GDK_BACKEND`:
+sets the NVIDIA DMABUF workaround and then branches on `GDK_BACKEND`:
+- **Both paths**, before the split: when `/proc/driver/nvidia/version` exists and the linked WebKitGTK
+  honours the switch (>= 2.44, `WebView::SupportsForceSharedMemory()`), it sets
+  `WEBKIT_DMABUF_RENDERER_FORCE_SHM=1` and `WEBKIT_FORCE_DMABUF_RENDERER=1` non-replacing. The
+  proprietary driver cannot back WebKit's hardware DMABUF renderer, so the shared-memory transport keeps
+  accelerated compositing instead of dropping it; `WEBKIT_FORCE_DMABUF_RENDERER` clears the
+  Debian/Ubuntu/openSUSE `isNVIDIA()` early return, which otherwise returns before the FORCE_SHM check.
 - **Default path** (`GDK_BACKEND` unset or not starting with `x11`): if wx lacks EGL (`!wxHAS_EGL`) and
-  `WAYLAND_DISPLAY` is set, it forces `GDK_BACKEND=x11`; it sets `WEBKIT_DISABLE_COMPOSITING_MODE=1`
-  non-replacing **only** when both `DISPLAY` and `WAYLAND_DISPLAY` are non-empty (a Wayland session with
-  XWayland available — the session, not the backend GTK will pick); `XInitThreads()` runs only when
-  `DISPLAY` is set.
+  `WAYLAND_DISPLAY` is set, it forces `GDK_BACKEND=x11`, and in that same branch it sets
+  `WEBKIT_DISABLE_COMPOSITING_MODE=1` non-replacing when the linked WebKitGTK predates 2.42
+  (`WebView::NeedsCompositingWorkaround()`). That forced X11 backend in a Wayland session is the only
+  default-path case that really runs through XWayland; `DISPLAY` and `WAYLAND_DISPLAY` alone do not
+  identify it, because a native Wayland app inherits both. `XInitThreads()` runs only when `DISPLAY` is
+  set.
 - **`GDK_BACKEND=x11…` opt-in path**: sets `DRI_PRIME=1` (and the NVIDIA PRIME variables when
   `/proc/driver/nvidia/version` exists), calls `XInitThreads()`, and deliberately does **not** disable
-  compositing (the comment: the WebKit2GTK XWayland bug appears fixed in WebKit2GTK ≥ 2.42, and leaving
-  compositing on keeps hardware acceleration for Device / Setup Wizard / login pages).
+  compositing, which keeps hardware acceleration for Device / Setup Wizard / login pages.
 
 Runtime backend detection elsewhere uses Orca's `is_running_on_wayland()` / `is_running_on_x11()`
 (see `references/platforms.md`), not these environment variables.
 
 **Pitfall.**
-- **Rule:** Set `WEBKIT_DISABLE_COMPOSITING_MODE` only on the default path in Wayland sessions,
-  non-replacing (`setenv(..., false)`, so a value the user already exported wins), before GTK
-  initialises; never unconditionally. Keep JS-side fallbacks so a page never
+- **Rule:** On NVIDIA reach for `WEBKIT_DMABUF_RENDERER_FORCE_SHM` before
+  `WEBKIT_DISABLE_COMPOSITING_MODE`. **Why:** the proprietary driver cannot back WebKit's hardware
+  DMABUF renderer, leaving web views blank or slow; FORCE_SHM keeps accelerated compositing and only
+  skips the hardware path, while disabling compositing drops it. Threaded scrolling is a separate,
+  release-dependent default (off in 2.44/2.45, on from 2.46). `WEBKIT_DISABLE_DMABUF_RENDERER` is not a
+  substitute — it is read first, leaves the transport set empty (no compositing from 2.44 on, where the
+  X11/Wayland fallback renderers were removed) and has a reported crash path. On Debian/Ubuntu and
+  openSUSE the `isNVIDIA()` patch returns before the FORCE_SHM check, so `WEBKIT_FORCE_DMABUF_RENDERER=1`
+  must clear it first; that variable exists only in those patches. WebKit reads all of these once per
+  process, before the first WebView.
+  The test is `/proc/driver/nvidia/version`, so a hybrid laptop that renders on its iGPU also takes the
+  shared-memory path, which is a per-frame GPU readback rather than zero-copy; exporting
+  `WEBKIT_DMABUF_RENDERER_FORCE_SHM=0` in the environment restores the hardware path.
+  ```cpp
+  // Wrong: kills compositing and threaded scrolling (and can crash on modern WebKit)
+  ::setenv("WEBKIT_DISABLE_DMABUF_RENDERER", "1", false);
+  // Right: keep compositing, skip only the hardware DMABUF path
+  if (::access("/proc/driver/nvidia/version", F_OK) == 0 && WebView::SupportsForceSharedMemory()) {
+      ::setenv("WEBKIT_DMABUF_RENDERER_FORCE_SHM", "1", /*replace*/ false);
+      ::setenv("WEBKIT_FORCE_DMABUF_RENDERER", "1", /*replace*/ false);
+  }
+  ```
+  Cite: WebKit 2.52.6 `Source/WebKit/UIProcess/gtk/AcceleratedBackingStore.cpp`
+  (`rendererBufferTransportMode()`), Debian `disable-nvidia-dmabuf.patch`, WebKit bug 262607.
+- **Rule:** Set `WEBKIT_DISABLE_COMPOSITING_MODE` only on the default path, in the branch where wx
+  lacks EGL and Orca forces `GDK_BACKEND=x11`, non-replacing (`setenv(..., false)`, so a value the user
+  already exported wins), before GTK
+  initialises; never unconditionally, and only for the WebKitGTK releases that need it (older than
+  2.42, `WebView::NeedsCompositingWorkaround()`). Keep JS-side fallbacks so a page never
   waits forever for a C++→JS call.
-  **Why:** WebKit2GTK accelerated compositing could fail silently under XWayland, leaving web views
-  blank (Setup Wizard stuck on "Loading..."); disabling it everywhere needlessly drops hardware
-  acceleration; without the JS timeout the user is stuck when the signal is lost.
+  **Why:** WebKit2GTK accelerated compositing could fail silently under XWayland before 2.42, leaving
+  web views blank (Setup Wizard stuck on "Loading..."); the variable also turns off threaded scrolling,
+  so applying it on a newer WebKit needlessly drops both GPU compositing and async scrolling (the
+  Device tab loses its smoothness); without the JS timeout the user is stuck when the signal is lost.
   ```cpp
   // Wrong: unconditional, kills HW acceleration everywhere
   ::setenv("WEBKIT_DISABLE_COMPOSITING_MODE", "1", false);
-  // Right (default path only): non-replacing, before GTK init, Wayland session with XWayland
-  if (display && *display && wayland && *wayland)
+  // Right (default path only): non-replacing, before GTK init, WebKitGTK < 2.42,
+  // and only in the branch that forces GDK_BACKEND=x11 (wx without EGL), the
+  // case that runs through XWayland
+  if (WebView::NeedsCompositingWorkaround() && wayland && *wayland)
       ::setenv("WEBKIT_DISABLE_COMPOSITING_MODE", "1", /*replace*/ false);
   ```
   Cite: c12912e0df (`src/OrcaSlicer.cpp`; `resources/web/guide/0/load.js` `OnInit`

CMakeLists.txt

@@ -288,6 +288,19 @@ if (APPLE)
     # Type"); Ninja/-Wall does not, and under -Werror it fails Xcode builds on code CI accepts.
     set(CMAKE_XCODE_ATTRIBUTE_GCC_WARN_64_TO_32_BIT_CONVERSION "NO")
 
+    # The Xcode generator ignores CMAKE_<LANG>_COMPILER_LAUNCHER, so hand the launcher to
+    # Xcode's own Compiler Launcher setting, which wraps every clang compile (C, C++,
+    # Objective-C++ and the precompiled header) with one program: the C++ launcher wins when
+    # both are set. The setting takes a single program path, not a CMake launcher list.
+    # Xcode versions without it ignore it and build uncached.
+    if (CMAKE_GENERATOR STREQUAL "Xcode")
+        if (CMAKE_CXX_COMPILER_LAUNCHER)
+            set(CMAKE_XCODE_ATTRIBUTE_C_COMPILER_LAUNCHER "${CMAKE_CXX_COMPILER_LAUNCHER}")
+        elseif (CMAKE_C_COMPILER_LAUNCHER)
+            set(CMAKE_XCODE_ATTRIBUTE_C_COMPILER_LAUNCHER "${CMAKE_C_COMPILER_LAUNCHER}")
+        endif ()
+    endif ()
+
     message(STATUS "Orca: IS_CROSS_COMPILE: ${IS_CROSS_COMPILE}")
 elseif (CMAKE_SYSTEM_NAME STREQUAL "Linux")
     set(CMAKE_INSTALL_RPATH "$ORIGIN")

build_linux.sh

@@ -22,7 +22,7 @@ function usage() {
     echo "   -g: run the requested build steps inside a Docker/Podman Ubuntu 24.04 container similar to the GitHub Actions Linux runner"
     echo "   -h: prints this help text"
     echo "   -i: build the Orca Slicer AppImage (optional)"
-    echo "   -p: boost ccache hit rate by disabling precompiled headers (default: ON)"
+    echo "   -p: disable precompiled headers (on by default): more ccache hits after header changes, slower misses"
     echo "   -r: skip RAM and disk checks (low RAM compiling)"
     echo "   -s: build the Orca Slicer (optional)"
     echo "   -t: build tests (optional), requires -s flag"
@@ -549,6 +549,20 @@ fi
 if [ -n "${CMAKE_CCACHE}" ] ; then
     echo "${CMAKE_CCACHE} found, enabling compiler caching..."
     export CMAKE_CCACHE_ARGS=(-DCMAKE_C_COMPILER_LAUNCHER="${CMAKE_CCACHE}" -DCMAKE_CXX_COMPILER_LAUNCHER="${CMAKE_CCACHE}")
+    # The same ccache settings as the CI build. Without this sloppiness ccache refuses every
+    # compile that uses the precompiled header, which is most of OrcaSlicer, and skips headers
+    # a checkout has just written. It is added to the sloppiness already configured, and
+    # CCACHE_SLOPPINESS or CCACHE_(NO)DEPEND set in the environment win.
+    if [[ "$(basename "${CMAKE_CCACHE}")" == ccache ]] ; then
+        if [[ -z "${CCACHE_SLOPPINESS+x}" ]] ; then
+            CCACHE_SLOPPINESS=$("${CMAKE_CCACHE}" --get-config sloppiness 2>/dev/null || true)
+            export CCACHE_SLOPPINESS="${CCACHE_SLOPPINESS:+${CCACHE_SLOPPINESS},}pch_defines,time_macros,include_file_mtime,include_file_ctime"
+        fi
+        # Hash the includes the compiler reports instead of preprocessing every miss.
+        if [[ -z "${CCACHE_DEPEND+x}${CCACHE_NODEPEND+x}" ]] ; then
+            export CCACHE_DEPEND=1
+        fi
+    fi
 else
     echo "Note: ccache or sccache are not found. Install either of them for faster rebuilds."
 fi

build_release_macos.sh

@@ -111,6 +111,54 @@ else
   export CMAKE_POLICY_COMPAT=""
 fi
 
+# Compiler cache, as in build_linux.sh. The program comes from CMAKE_CCACHE, else from a
+# launcher already set in the environment (the CI build sets ccache there), else sccache or
+# ccache from PATH; one program serves both languages. Each run applies its own choice to the
+# build directories, replacing a launcher an earlier run or a hand-run configure recorded.
+CMAKE_CCACHE_ARGS=()
+CCACHE_VAR=""
+for var in CMAKE_CCACHE CMAKE_CXX_COMPILER_LAUNCHER CMAKE_C_COMPILER_LAUNCHER; do
+    if [ -n "${!var:-}" ]; then
+        CCACHE_VAR=$var
+        break
+    fi
+done
+if [ -n "$CCACHE_VAR" ]; then
+    CCACHE_PROGRAM=${!CCACHE_VAR}
+    echo "Checking ${CCACHE_VAR} environment variable for compiler cache program..."
+    CMAKE_CCACHE=$(command -v "${CCACHE_PROGRAM}") || {
+        echo "${CCACHE_VAR} environment variable is set to '${CCACHE_PROGRAM}' but it was not found in PATH."
+        CMAKE_CCACHE=""
+    }
+elif command -v sccache >/dev/null 2>&1 ; then
+    CMAKE_CCACHE=$(command -v sccache)
+elif command -v ccache >/dev/null 2>&1 ; then
+    CMAKE_CCACHE=$(command -v ccache)
+fi
+if [ -n "${CMAKE_CCACHE}" ] ; then
+    echo "${CMAKE_CCACHE} found, enabling compiler caching..."
+    CMAKE_CCACHE_ARGS=(-DCMAKE_C_COMPILER_LAUNCHER="${CMAKE_CCACHE}" -DCMAKE_CXX_COMPILER_LAUNCHER="${CMAKE_CCACHE}")
+    # The same ccache settings as the CI build. Without this sloppiness ccache refuses every
+    # compile that uses the precompiled header, which is most of OrcaSlicer, and skips headers
+    # a checkout has just written. It is added to the sloppiness already configured, and
+    # CCACHE_SLOPPINESS or CCACHE_(NO)DEPEND set in the environment win.
+    if [[ "$(basename "${CMAKE_CCACHE}")" == ccache ]] ; then
+        if [[ -z "${CCACHE_SLOPPINESS+x}" ]] ; then
+            CCACHE_SLOPPINESS=$("${CMAKE_CCACHE}" --get-config sloppiness 2>/dev/null || true)
+            export CCACHE_SLOPPINESS="${CCACHE_SLOPPINESS:+${CCACHE_SLOPPINESS},}pch_defines,time_macros,include_file_mtime,include_file_ctime"
+        fi
+        # Hash the includes the compiler reports instead of preprocessing every miss.
+        if [[ -z "${CCACHE_DEPEND+x}${CCACHE_NODEPEND+x}" ]] ; then
+            export CCACHE_DEPEND=1
+        fi
+    fi
+else
+    echo "Note: ccache or sccache are not found. Install either of them for faster rebuilds."
+    # Clear a launcher recorded by an earlier configure. The entries must be set empty rather
+    # than removed: CMake fills a missing launcher entry from the environment.
+    CMAKE_CCACHE_ARGS=(-DCMAKE_C_COMPILER_LAUNCHER:STRING= -DCMAKE_CXX_COMPILER_LAUNCHER:STRING=)
+fi
+
 echo "Build params:"
 echo " - ARCH: $ARCH"
 echo " - BUILD_CONFIG: $BUILD_CONFIG"
@@ -162,6 +210,7 @@ function build_deps() {
                         -DCMAKE_OSX_ARCHITECTURES:STRING="${_ARCH}" \
                         -DCMAKE_OSX_DEPLOYMENT_TARGET="${OSX_DEPLOYMENT_TARGET}" \
                         -DCMAKE_IGNORE_PREFIX_PATH="${CMAKE_IGNORE_PREFIX_PATH}" \
+                        "${CMAKE_CCACHE_ARGS[@]}" \
                         ${CMAKE_POLICY_COMPAT}
                 fi
                 cmake --build . --config "$BUILD_CONFIG" --target deps
@@ -264,6 +313,7 @@ function build_slicer() {
                     -DCMAKE_OSX_ARCHITECTURES="${_ARCH}" \
                     -DCMAKE_OSX_DEPLOYMENT_TARGET="${OSX_DEPLOYMENT_TARGET}" \
                     -DCMAKE_IGNORE_PREFIX_PATH="${CMAKE_IGNORE_PREFIX_PATH}" \
+                    "${CMAKE_CCACHE_ARGS[@]}" \
                     ${CMAKE_POLICY_COMPAT}
             fi
             cmake --build . --config "$BUILD_CONFIG" --target "$SLICER_BUILD_TARGET"

deps/JPEG/JPEG.cmake

@@ -1,22 +1,7 @@
-if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
-    if (JPEG_VERSION STREQUAL "6")
-        message("Using Jpeg Lib 62")
-        set(jpeg_flag "")
-    elseif (JPEG_VERSION STREQUAL "7")
-        message("Using Jpeg Lib 70")
-        set(jpeg_flag "-DWITH_JPEG7=ON")
-    else ()
-        message("Using Jpeg Lib 80")
-        set(jpeg_flag "-DWITH_JPEG8=ON")
-    endif ()
-endif()
-
 orcaslicer_add_cmake_project(JPEG
-    URL https://github.com/libjpeg-turbo/libjpeg-turbo/archive/refs/tags/3.0.1.zip
-    URL_HASH SHA256=d6d99e693366bc03897677650e8b2dfa76b5d6c54e2c9e70c03f0af821b0a52f
-    DEPENDS ${ZLIB_PKG}
+    URL https://github.com/libjpeg-turbo/libjpeg-turbo/archive/refs/tags/3.2.0.zip
+    URL_HASH SHA256=1b4c2cb6cba1ca5e53b9e85eeeab02154cab4b70894c96e97a9c747a26d19d63
     CMAKE_ARGS
         -DENABLE_SHARED=OFF
         -DENABLE_STATIC=ON
-        ${jpeg_flag}
 )

docs/HLSD/chromastack.md

@@ -0,0 +1,239 @@
+# ChromaStack - High Level Design
+
+## Purpose and scope
+
+ChromaStack prints a colour picture on a single face of a model: a few filaments are printed one on
+top of another, the printer swaps filament at fixed layer boundaries, and the *height* of every
+pixel decides how many of those layers reach it. Filament is translucent, so a pixel that stops
+part-way through a run shows a blend of that run's filament and whatever is under it, which is
+where the colours come from - far more of them than there are filaments loaded.
+
+This is that model inside OrcaSlicer, as a mode of a texture displacement layer. The picture and
+the filament stack are the input; the output is a grayscale height map and a set of filament swaps.
+Nothing about it is a new pipeline: the height map is handed to the layer the user painted, so the
+existing projection, preview and bake turn it into real geometry, and the swaps are handed to the
+object's height ranges, which is the slicer's own way of printing a Z band in another filament.
+
+It does not import other filament-painting tools' project files, and it does not choose the filaments or
+their order - that is the user's artistic decision, and the panel exists to make its result
+visible before the print.
+
+## The colour model
+
+`libslic3r/ChromaStack.hpp` holds the whole of it, free of any GUI dependency, which is what lets the
+solve run against plain image bytes and be unit tested.
+
+A **stack** (`ChromaStack`) is a list of **bands** (`ChromaBand`), bottom-up: a filament and
+how many layers it gets. Alongside them it carries the print's layer height, which the colours
+depend on directly, and the **base**: how many layers of solid backing the picture sits on.
+
+The base is not a band. It is a measurement of the model - everything below the painted face, often
+a hundred layers or more on a tall part - while a band is a handful of layers the user chooses. It
+prints in the bottom band's filament, because that is what is loaded when the print starts, so it
+extends that band's height range down to the first layer instead of being a swap of its own. Only
+the bands are quantized into the height map, which is why the 255-layer limit
+(`CHROMA_MAX_RELIEF_LAYERS`) bounds the picture and not the stack.
+
+Each filament carries its **transmission distance** (TD), the usual measure for how deep
+light gets into it: about 1 for an opaque filament, 4 to 7 for a usefully translucent one, 8 and
+above for a nearly clear one. TD lives in the filament preset, as
+`filament_transmission_distance` - it is a property of the material, so it belongs with the
+material and ships with a profile rather than being retyped per project. It has no effect on
+slicing.
+
+`chroma_palette()` turns a stack into the colour it shows at every possible column height, which
+is all the solver ever needs. The model is Beer-Lambert absorption, layer by layer: a layer of
+thickness `h` in a filament of transmission distance `TD` passes
+
+    T = exp(-h / (CHROMA_TD_OPAQUE_FRACTION * TD / 3))
+
+of the light from under it and shows its own colour with the rest, so walking up from an opaque
+base,
+
+    C_k = (1 - T_k) * c_k + T_k * C_(k-1),  C_0 = c_1
+
+The constants say what TD means in physical terms: a filament stops showing what is behind it at
+about a tenth of its TD number (`CHROMA_TD_OPAQUE_FRACTION`), and three e-folding lengths of
+absorption pass 5% of the light, which is where the eye stops seeing through. Those are calibrated
+against how the community's TD step prints behave, not derived from first principles - TD is an
+empirical scale.
+
+How fast that happens is the single most useful fact about a stack, and `chroma_layers_to_opaque()`
+is it: a filament hides what is under it at about a tenth of its TD, so how many *layers* that takes
+depends entirely on how thick they are. TD 4 covers in two layers at 0.2 mm and in five at 0.08 mm.
+Below about two layers a band has no intermediate colours at all - the picture can only use the
+filament or not use it - which is why an otherwise sound stack prints as flat steps at a coarse
+layer height. The panel reports the number under the palette ramp, and warns when it falls below
+two, because nothing else makes it visible: the filaments look fine, the bands look long enough, and
+the result looks like the colour model is wrong when it is the layer height.
+
+How many of a picture's layers actually earn their place is a second question, and
+`chroma_distinct_levels()` answers it: a band eventually hides what is under it, and another layer of
+it then shows the colour the one below already showed. It is measured against the palette rather than
+against TD, because how long a band keeps paying depends on what is under it as much as on the
+filament - a light band over a dark one is finished in four or five layers, while a dark one over a
+light one goes on moving the colour perceptibly well past the depth at which it is nominally opaque,
+since the last few percent of light coming through still shifts a near-black colour. `Trim` shortens
+each band to its last layer that still changes anything.
+
+Beer-Lambert is also the only model that stays self-consistent applied per layer: transmittance
+multiplies, so a run of n layers of one filament passes `exp(-n h / d)` whether it is composited in
+one step or n, and the palette therefore does not depend on how the stack happens to be cut into
+bands. A unit test asserts exactly that. Compositing is done on linear light, because the quantity
+being attenuated is light rather than an sRGB code.
+
+The base is assumed opaque. That is what makes a column's colour depend only on the layers above it,
+and it is why the panel warns when the base is thin.
+
+## The solve
+
+`chroma_solve()` maps each pixel of the picture to the column height whose palette colour is
+closest to it, perceptually: both sides go to CIELAB and the match is a CIE76 distance. A 1-D
+palette of a few dozen entries makes the extra accuracy of CIEDE2000 hard to see and the cost of it
+easy to feel - several hundred million comparisons for a large picture - so the cheaper metric is
+the deliberate choice here, unlike the texture colour quantizer, which matches against a
+two-dimensional palette of filament mixes and uses a precomputed CIEDE2000 cube.
+
+Optional Floyd-Steinberg error diffusion, serpentine, trades the bands a photograph shows across a
+sky or a cheek for fine stipple. Off by default, since stipple is the artifact rather than the cure
+for flat-coloured artwork.
+
+Three things come back: the per-pixel layer count, the same thing as an 8-bit grayscale height map
+(0 at the base, 255 at the top of the stack), and the palette colours chosen, as an RGB image. The
+quantization belongs to the solve rather than to the displacement because the relief's steps have
+to land on layer boundaries, and a grey level is the only channel the displacement reads.
+
+The solve also reports the mean and worst colour difference it had to accept. That is the only
+honest answer to "will this stack reproduce my picture": under about 5 is a good match, over about
+15 means the filaments cannot reach the picture's colours and no number of layers will help.
+
+## How it attaches to a texture displacement layer
+
+A ChromaStack layer is an ordinary `TextureDisplacementLayer` whose height map is computed rather than
+imported:
+
+- `chroma_source` keeps the picture the user chose, so the solve can be redone whenever the stack,
+  the layer height or the picture changes;
+- `chroma` is the stack;
+- `image_data` - the same field every other layer uses - holds whatever the last solve produced.
+
+So everything downstream needs no changes at all and knows nothing about colour stacks. It also
+means the solve writes `depth_mm` (the relief is exactly as tall as the layers above the base) and
+pins `invert`, `midlevel` and `smoothing`: blurring a solved height map moves its steps off the
+layer boundaries they were solved for.
+
+Colour by filament swap and the layer's own `color_enabled` (one filament per triangle, via
+`mmu_segmentation_facets`) are mutually exclusive, and the panel enforces it. They are two
+different ways of colouring the same surface, and painting the relief per triangle would hide the
+very stack the heights were solved against.
+
+## The filament swaps
+
+`chroma_ranges()` turns the bands into Z bands in the object's own coordinates, merging any two
+adjacent bands that share a filament, since nothing is swapped between them. The panel writes them
+onto the object as `layer_config_ranges`, each range carrying `extruder` - which
+`apply_to_print_region_config()` spreads over every feature filament the range does not otherwise
+override, i.e. the whole band prints in that filament - and `layer_height`, both because the solve
+is only valid at the height it was run at and because `layer_height_profile_from_ranges()` reads
+that option without checking it is there.
+
+Height ranges rather than colour-change custom G-code: they are object-relative, so they survive
+the object being moved, they are saved in the project file, and they appear in the object list where
+the user can see and edit them. The cost is that an object can only have one set of them, so
+applying a stack replaces whatever ranges the object had, with a confirmation when it had any.
+
+`ObjectList::rebuild_layer_ranges()` mirrors them in the object tree without touching the
+selection - changing the selection while a gizmo is open tears the gizmo down under the user.
+
+## Geometry the model assumes
+
+A filament swap happens at one Z across the whole plate, and the relief grows along the painted
+surface's normal. The two agree only where that surface is horizontal, so:
+
+- the painted patch's height above the bottom of the object is what the base layer count has to
+  match, and the panel measures it from the patch and offers the number rather than leaving the user
+  to work it out;
+- a patch that is not flat in Z gets a warning, because its colours will only be right where it is
+  level. The relief itself is still computed, since a gently domed plaque is a legitimate thing to
+  print and only the user can judge how much error is acceptable.
+
+The model itself is the user's: the stack needs a slab of the base thickness with a face to paint,
+and nothing here creates or thickens geometry.
+
+## Seeing the colours
+
+The colours are blends of translucent filaments. No single filament makes them, so they cannot be
+expressed as the per-triangle filament assignment the 3D scene paints with, and there is nothing for
+the ordinary view to draw. They exist in two places: the sliced preview, which shows the filament
+bands as they print, and an overlay the gizmo draws over whatever surface is on screen.
+
+That overlay is a shader, `resources/shaders/*/chromastack.*`, and it reads the colour from each
+fragment's own **print Z** rather than from the height map or from anything in the mesh. A filament
+swap happens at one Z across the whole plate, so the layer a point sits in is what decides its
+colour whatever the picture says. Two things follow. A band edge lands exactly on the layer boundary
+however coarse the triangles are - a per-triangle colouring puts it wherever the mesh happens to
+have an edge. And it keeps working after a bake, when the relief is real geometry, the paint has
+been cleared and the height map is attached to nothing on screen; the overlay is built from the live
+preview's displaced mesh before a bake and from the volume's own mesh after one.
+
+The stack's colours reach the shader as a texture of one texel per printed layer, because a base can
+be several hundred layers and no uniform array holds that. The mesh carries nothing about the stack,
+so editing the stack costs a palette upload rather than a mesh rebuild.
+
+On top of the colour the shader tries to look like printed plastic, because a flat shaded colour
+reads as a rendering and the whole point of the overlay is judging a print:
+
+- the terminator is **wrapped** and the grazing angles **glow**, both scaled by the transmission
+  distance of the filament on top - filament translucent enough to be worth stacking scatters light
+  rather than reflecting it, and an opaque stack correctly gets neither;
+- the walls carry **layer lines**, a groove at every layer boundary that also modulates the
+  specular, so a wall glints in stripes as a printed one does. Terraces get none: a terrace is one
+  layer's top surface and has no lines to show.
+
+Both effects are driven by the fragment's height, so they are free of any texture or UV, and both
+fade out through `fwidth` once a pixel covers more than about half a layer - lines finer than a
+pixel are noise, and a colour read from one layer out of the dozen under a pixel flickers as the
+camera moves. The palette lookup slides off the texel centres over the same range so the linear
+filter averages the bands a pixel covers instead of picking one of them.
+
+The overlay is drawn pulled forward and without depth writes, the same pass the painted-colour
+overlay uses, since it is the same surface as whatever is underneath it.
+
+## What the panel shows
+
+The kind of a layer is the layer's own, chosen when it is added and fixed from then on: **Add layer**
+offers *Relief layer* or *ChromaStack picture*, and the card carries its kind as a chip. There is no
+panel-wide mode, so one object can hold a picture on its face and a knurl on its rim, and a card
+never shows a control that cannot do anything on it.
+
+What the two cards do not share:
+
+| Control | Relief card | ChromaStack card |
+| --- | --- | --- |
+| Depth | slider | the solve writes it; the card prints the result |
+| Tile size | repeats the pattern | **Picture size**, the same number named for what it does |
+| Midlevel, Smoothing, Invert, Edge fade | present | absent: each would move the steps off the layer boundaries |
+| Colors, Mix filaments, Denoise | one filament per triangle | absent: swaps are how this card colours |
+| Tile, tile method | present | absent: one copy, never tiled |
+| Blend | four modes | absent: it is the top layer and replaces what is under it |
+| Picture, prediction, stack, palette, match, Dither, Solve, Apply swaps | absent | present |
+
+A ChromaStack card has no grip and cannot be reordered: its colours come from swapping filament at a
+height, so anything printed above it would bury them. An object may hold only one, refused when the
+layer is added rather than when the swaps are applied, because the swaps are the object's own height
+ranges and there is one set of those.
+
+The card itself, top to bottom: what it is in one line, the picture and the predicted print side by
+side at the same size, the match figures, the base with the measurement from the painted face offered
+beside it, a row per band - swatch, filament, layer count, TD - the colours the stack reaches as a
+ramp, the totals, dither with the layer height it is solved against, and the two actions. Base and
+band counts are typed fields rather than sliders: a base is measured in hundreds of layers and a band
+in single digits, and dragging past the end of a slider's range is how a stack silently loses its
+backing.
+
+The ramp is computed from the stack alone, before any solve. It is what answers "can these filaments
+print this picture" without spending a solve to find out they cannot: a short or flat ramp means the
+filaments are too alike, or too opaque, to give the picture anything to work with.
+
+A per-slot flag records that the stack has been edited since the last solve, because a stale height
+map looks exactly like a good one.

resources/images/toolbar_tooltip.svg

@@ -1 +1 @@
-<?xml version="1.0" encoding="UTF-8"?><svg id="a" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 15 15"><rect x="7" y="12" width="1" height="1" style="fill:#fefefe;"/><path d="M4.5,5.5c0-1.71,1.43-3.09,3.16-3,1.62.08,2.84,1.54,2.84,3.17h0c0,.61-.25,1.2-.69,1.62l-2.31,2.24v.97" style="fill:none; stroke:#fefefe; stroke-linecap:round; stroke-linejoin:round;"/></svg>
+<?xml version="1.0" encoding="UTF-8"?><svg id="a" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 15 15"><rect x="1.5" y="3.5" width="12" height="8" rx="1" ry="1" style="fill:none; stroke:#fefefe; stroke-linecap:round; stroke-linejoin:round;"/><path d="M3,5h1v1h-1Z M5,5h1v1h-1Z M7,5h1v1h-1Z M9,5h1v1h-1Z M11,5h1v1h-1Z M4,7h1v1h-1Z M6,7h1v1h-1Z M8,7h1v1h-1Z M10,7h1v1h-1Z M5,9h5v1h-5Z" style="fill:#fefefe;"/></svg>

resources/shaders/110/chromastack.fs

@@ -0,0 +1,92 @@
+#version 110
+
+// What a ChromaStack print looks like, shaded per fragment.
+//
+// The colour is not a vertex attribute and not a texture on the surface: it is read from the
+// fragment's own height. A filament swap happens at one print Z across the whole plate, so the layer
+// a point sits in is what decides its colour - which means a band edge lands exactly on the layer
+// boundary however coarse the mesh is, and the same shader is right before a bake (on the preview's
+// displaced mesh) and after one (on the volume's own).
+//
+// On top of that it tries to look like printed plastic rather than like flat shaded colour: the
+// terminator is wrapped and the grazing angles glow, because filament translucent enough to be worth
+// stacking scatters light rather than reflecting it; and the walls carry layer lines, because that is
+// the first thing the eye uses to read a print as a print.
+
+#define INTENSITY_CORRECTION 0.6
+const vec3 LIGHT_TOP_DIR = vec3(-0.4574957, 0.4574957, 0.7624929);
+#define LIGHT_TOP_DIFFUSE    (0.8 * INTENSITY_CORRECTION)
+#define LIGHT_TOP_SPECULAR   (0.125 * INTENSITY_CORRECTION)
+#define LIGHT_TOP_SHININESS  20.0
+const vec3 LIGHT_FRONT_DIR = vec3(0.6985074, 0.1397015, 0.6985074);
+#define LIGHT_FRONT_DIFFUSE  (0.3 * INTENSITY_CORRECTION)
+#define INTENSITY_AMBIENT    0.3
+
+// The stack's colours, one texel per printed layer, bottom to top. A texture rather than a uniform
+// array because a base can be several hundred layers and no uniform array survives that.
+uniform sampler2D palette;
+uniform float     palette_size;  // texels in `palette`, i.e. the stack's total printed layers
+uniform float     base_z;        // world Z of the bottom of the object
+uniform float     layer_height;  // mm
+// How translucent the stack reads, 0 to 1, from the transmission distance of the filament on top.
+// It drives the wrap and the rim: an opaque stack should not glow.
+uniform float     translucency;
+uniform float     alpha;
+
+varying vec3  clipping_planes_dots;
+varying vec3  eye_normal;
+varying vec3  eye_position;
+varying vec3  world_normal;
+varying float world_z;
+
+void main()
+{
+    if (any(lessThan(clipping_planes_dots, vec3(0.0))))
+        discard;
+
+    float z_rel  = world_z - base_z;
+    float layers = max(z_rel / layer_height, 0.0);
+    // How much of a layer one pixel covers. Everything below is faded out as this grows: layer lines
+    // finer than a pixel are noise, and a colour read from one layer out of the dozen under a pixel
+    // flickers as the camera moves.
+    float px     = fwidth(layers);
+
+    // The layer this fragment sits in, 1-based. Up close that is an exact texel of the palette; once a
+    // pixel spans a layer or more it slides to a continuous coordinate so the linear filter averages
+    // the bands it covers instead of picking one of them at random.
+    float stepped = clamp(ceil(layers - 1.0e-4), 1.0, palette_size) - 0.5;
+    float smooth_ = clamp(layers, 0.5, palette_size - 0.5);
+    float u       = mix(stepped, smooth_, clamp(px - 0.5, 0.0, 1.0)) / palette_size;
+    vec3  albedo  = texture2D(palette, vec2(u, 0.5)).rgb;
+
+    vec3 N = normalize(eye_normal);
+    vec3 V = normalize(-eye_position);
+
+    // Layer lines, on the walls only: a terrace is one layer's top surface and has none to show. The
+    // groove runs round the part at every layer boundary, so it is a function of height alone.
+    float wall   = 1.0 - abs(world_normal.z);
+    float phase  = fract(layers);
+    float groove = smoothstep(0.0, 0.16, phase) * smoothstep(1.0, 0.84, phase);
+    groove       = mix(groove, 0.65, clamp(px * 1.5, 0.0, 1.0)); // fade to flat when too fine to resolve
+    float lines  = mix(1.0, 0.84 + 0.16 * groove, wall * 0.9);
+    albedo      *= lines;
+
+    // Diffuse, with a wrapped terminator: light entering translucent filament comes back out a little
+    // way round the curve, so the shaded side never goes as dark as an opaque surface's would.
+    float wrap    = 0.25 + 0.45 * translucency;
+    float ndotl   = dot(N, LIGHT_TOP_DIR);
+    float diffuse = max((ndotl + wrap) / (1.0 + wrap), 0.0) * LIGHT_TOP_DIFFUSE;
+    diffuse      += max((dot(N, LIGHT_FRONT_DIR) + wrap) / (1.0 + wrap), 0.0) * LIGHT_FRONT_DIFFUSE;
+
+    // Specular, strongest along a layer line, which is what makes a printed wall glint in stripes
+    // rather than evenly.
+    float spec = LIGHT_TOP_SPECULAR * pow(max(dot(V, reflect(-LIGHT_TOP_DIR, N)), 0.0), LIGHT_TOP_SHININESS);
+    spec      *= mix(1.0, 0.35 + 0.65 * groove, wall * 0.9);
+
+    // Scattering out of the edges: a translucent part lights up where it is seen at a grazing angle
+    // and where it is thin. Tinted by the filament, because what comes back out has been through it.
+    float rim = pow(1.0 - clamp(dot(N, V), 0.0, 1.0), 4.0);
+
+    vec3 color = albedo * (INTENSITY_AMBIENT + diffuse) + vec3(spec) + albedo * rim * translucency * 0.55;
+    gl_FragColor = vec4(color, alpha);
+}

resources/shaders/110/chromastack.vs

@@ -0,0 +1,40 @@
+#version 110
+
+uniform mat4 view_model_matrix;
+uniform mat4 projection_matrix;
+uniform mat3 view_normal_matrix;
+uniform mat4 volume_world_matrix;
+// The volume's normal matrix in world space, so the fragment shader can tell a wall from a terrace -
+// which is what decides whether layer lines show at all.
+uniform mat3 world_normal_matrix;
+// Clipping plane, x = min z, y = max z (the FFF/SLA preview's top/bottom cut).
+uniform vec2 z_range;
+// Clipping plane, general orientation (the section view).
+uniform vec4 clipping_plane;
+
+// GLModel's P3N3 layout, which is what init_from(indexed_triangle_set) produces - this shader needs
+// nothing per vertex that the mesh does not already carry, because the colour comes from the
+// fragment's own height rather than from anything baked into the vertices.
+attribute vec3 v_position;
+attribute vec3 v_normal;
+
+varying vec3  clipping_planes_dots;
+varying vec3  eye_normal;
+varying vec3  eye_position;
+varying vec3  world_normal;
+varying float world_z;
+
+void main()
+{
+    vec4 model_pos = vec4(v_position, 1.0);
+    vec4 world_pos = volume_world_matrix * model_pos;
+    world_z        = world_pos.z;
+    world_normal   = normalize(world_normal_matrix * v_normal);
+    eye_normal     = normalize(view_normal_matrix * v_normal);
+
+    vec4 eye_pos = view_model_matrix * model_pos;
+    eye_position = eye_pos.xyz;
+    gl_Position  = projection_matrix * eye_pos;
+
+    clipping_planes_dots = vec3(dot(world_pos, clipping_plane), world_pos.z - z_range.x, z_range.y - world_pos.z);
+}

resources/shaders/110/texture_displacement_shaded.fs

@@ -36,11 +36,6 @@ uniform int       palette_b[64];
 uniform float     prefer_pure_de; // PREFER_PURE_DE: how much better than a single filament a mix must be
 uniform sampler2D color_tex;     // the layer's colour image, sampled at the same uv as the height
 uniform bool      has_color_tex;
-// False means the texture is a single placement, not a repeat. The sampler's wrap mode can express
-// that for the *height* - a zero border reads as no displacement - but not for colour, because a
-// zero border is indistinguishable from the image genuinely being black there. So the uv is tested
-// instead, which is what sample_layer_color() does on the CPU side.
-uniform bool      tile_enabled;
 uniform bool volume_mirrored;
 
 uniform mat4 view_model_matrix;
@@ -50,14 +45,6 @@ uniform sampler2D height_tex;
 uniform vec2      height_tex_texel;
 uniform float      depth_mm;
 uniform float      tiling_scale;
-// Tile extent along v, in mm. Zero or negative means "the same as tiling_scale" - the square tile
-// every layer had before the two axes could differ - which is also what layer_tile_mm() does on the
-// CPU side, so an unset uniform behaves exactly as this shader always did.
-uniform float      tiling_scale_v;
-// Mirror the texture along one axis of the tile. Both default to false, which GLSL guarantees for an
-// uninitialised bool uniform, so a shader whose caller has not set them behaves exactly as before.
-uniform bool       mirror_u;
-uniform bool       mirror_v;
 // Height map width / height. Scales the v axis so a non-square image keeps its proportions
 // instead of being squeezed into a square tile - mirrors libslic3r's apply_uv_transform().
 uniform float      tex_aspect;
@@ -67,12 +54,6 @@ uniform bool       invert;
 uniform float      midlevel;      // the height that means "don't move"; needed by the parallax step
 uniform vec3       eye_model_pos; // camera position in the texture frame (world minus tex_anchor)
 uniform vec3       tex_anchor;    // the volume's origin in world space: the texture frame's origin
-// The placement's rotation and its inverse, so this shader turns a world position into the texture
-// frame exactly as texture_displacement_bake_frame() does. Both are passed rather than one
-// transposed here, because GLSL 110 has no transpose(). Identity leaves the old world-aligned
-// behaviour, which is what an unset uniform gives.
-uniform mat3       tex_rot;     // world -> texture frame
-uniform mat3       tex_rot_inv; // texture frame -> world
 uniform bool       use_vertex_uv;
 // 2x3 affine (lin = (m00, m01, m10, m11), tr = (m02, m12)) applied to the dragged island's uv; see the
 // 140 variant. Identity when nothing is dragged.
@@ -181,29 +162,13 @@ void projection_axes(vec3 p, vec3 n, out vec3 t, out vec3 b, out vec2 units_per_
     }
 }
 
-// Tile extents in mm, u then v, with the v fallback layer_tile_mm() applies.
-vec2 tile_mm()
-{
-    float su = (tiling_scale > 1e-6) ? tiling_scale : 1.0;
-    return vec2(su, (tiling_scale_v > 1e-6) ? tiling_scale_v : su);
-}
-
-// +1 or -1 per axis, matching layer_mirror_sign() on the CPU side.
-vec2 mirror_sign()
-{
-    return vec2(mirror_u ? -1.0 : 1.0, mirror_v ? -1.0 : 1.0);
-}
-
 vec2 project_uv(vec3 p, vec3 n)
 {
     vec2 planar = projection_raw(p, n);
+    planar *= (tiling_scale > 1e-6) ? (1.0 / tiling_scale) : 1.0;
     float cs = cos(rotation_rad);
     float sn = sin(rotation_rad);
     vec2 r = vec2(planar.x * cs - planar.y * sn, planar.x * sn + planar.y * cs);
-    // Extents divide after the rotation, so a tile whose two sides differ stays a rotated rectangle
-    // instead of shearing. With a square tile the two orders agree exactly. The mirror is a sign on
-    // the same division, in the tile's own frame.
-    r *= mirror_sign() / tile_mm();
     // After the rotation, so the rotation stays a rotation rather than becoming a shear.
     r.y *= tex_aspect;
     return r + uv_offset;
@@ -272,12 +237,9 @@ void main()
 
     // World millimetres throughout, like the bake - see the 140 variant.
     vec3 triangle_normal = normalize(cross(dFdx(world_pos.xyz), dFdy(world_pos.xyz)));
-    vec3 tex_pos = tex_rot * (world_pos.xyz - tex_anchor); // the frame the texture is projected in, as the bake does
+    vec3 tex_pos = world_pos.xyz - tex_anchor; // the frame the texture is projected in, as the bake does
     if (volume_mirrored)
         triangle_normal = -triangle_normal;
-    // The same normal in that frame. triangle_normal itself stays in world, because the lighting at
-    // the end of this shader needs it there.
-    vec3 tex_normal = tex_rot * triangle_normal;
 
     // Where the colour is read from. Both branches below already compute the uv this fragment's
     // *height* came from - including the parallax-marched one on the triplanar path - and the colour
@@ -308,14 +270,14 @@ void main()
     } else if (weight > 0.0) {
         vec3 t, b;
         vec2 units_per_mm;
-        projection_axes(tex_pos, tex_normal, t, b, units_per_mm);
+        projection_axes(tex_pos, triangle_normal, t, b, units_per_mm);
 
         // Parallax occlusion mapping: march the view ray through the height shell and shade at the
         // first point where it drops below the displaced surface (see header).
         float amp      = (invert ? -1.0 : 1.0) * depth_mm * clamp(weight, 0.0, 1.0);
-        vec3  view_dir = normalize(tex_rot * eye_model_pos - tex_pos);
-        float v_dot_n  = dot(view_dir, tex_normal);
-        vec2  uv       = project_uv(tex_pos, tex_normal);
+        vec3  view_dir = normalize(eye_model_pos - tex_pos);
+        float v_dot_n  = dot(view_dir, triangle_normal);
+        vec2  uv       = project_uv(tex_pos, triangle_normal);
 
         // The shell the displaced surface lives inside, as signed heights along the normal. Taken from
         // both ends of h in [0, 1] so it stays correct for an inverted layer or a raised midlevel,
@@ -327,18 +289,18 @@ void main()
         // How far, in mm, sweeping the ray across the shell slides the sample point sideways. Below half
         // a texel there is no parallax to find and the march would be pure cost - which is the common
         // case of looking straight down at a surface.
-        float sweep = length(view_dir - tex_normal * v_dot_n) * (h_hi - h_lo) / max(v_dot_n, 1e-4);
-        if (v_dot_n > 0.05 && sweep > 0.5 * min(tile_mm().x, tile_mm().y) * height_tex_texel.x) {
+        float sweep = length(view_dir - triangle_normal * v_dot_n) * (h_hi - h_lo) / max(v_dot_n, 1e-4);
+        if (v_dot_n > 0.05 && sweep > 0.5 * tiling_scale * height_tex_texel.x) {
             // A point at ray parameter s (model_pos + view_dir * s) sits at height s * v_dot_n above the
             // undisplaced surface. Start at the top of the shell, where the ray is outside the surface
             // by construction, and step inward; the crossing is what this pixel actually sees.
             float s        = h_hi / v_dot_n;
             float ds       = (h_hi - h_lo) / (v_dot_n * float(PARALLAX_STEPS));
-            vec2  prev_uv  = project_uv(tex_pos + view_dir * s, tex_normal);
+            vec2  prev_uv  = project_uv(tex_pos + view_dir * s, triangle_normal);
             float prev_gap = h_hi - amp * (H_AT(prev_uv) - midlevel); // >= 0 by construction
             for (int i = 0; i < PARALLAX_STEPS; ++i) {
                 s -= ds;
-                vec2  cur_uv = project_uv(tex_pos + view_dir * s, tex_normal);
+                vec2  cur_uv = project_uv(tex_pos + view_dir * s, triangle_normal);
                 float gap    = s * v_dot_n - amp * (H_AT(cur_uv) - midlevel);
                 if (gap <= 0.0) {
                     // Crossed between the last two samples - interpolating the hit is what stops it
@@ -360,23 +322,18 @@ void main()
         float hU = texture2D(height_tex, uv + vec2(0.0, height_tex_texel.y)).r;
 
         vec2 dh_duv = vec2((hR - hL) / (2.0 * height_tex_texel.x), (hU - hD) / (2.0 * height_tex_texel.y));
-        vec2  tile       = tile_mm();
-        float amplitude  = (invert ? -1.0 : 1.0) * depth_mm * clamp(weight, 0.0, 1.0);
+        float inv_tiling = (tiling_scale > 1e-6) ? (1.0 / tiling_scale) : 1.0;
+        float amplitude  = (invert ? -1.0 : 1.0) * depth_mm * inv_tiling * clamp(weight, 0.0, 1.0);
 
         float cs = cos(rotation_rad);
         float sn = sin(rotation_rad);
-        // One uv unit is tile.x mm along u but tile.y / tex_aspect mm along v, so each component of
-        // the uv-space gradient carries its own factor before being rotated back into t/b.
-        // Each component carries its own extent and its own mirror sign, both of which project_uv()
-        // applied on the way in; without the sign a mirrored layer would light as though it were not.
-        vec2  ms    = mirror_sign();
-        vec2  g     = vec2(dh_duv.x * ms.x / tile.x, dh_duv.y * tex_aspect * ms.y / tile.y);
+        // One uv unit is tiling_scale mm along u but tiling_scale / tex_aspect mm along v, so the v
+        // component of the gradient carries the extra factor before being rotated back into t/b.
+        vec2  g     = vec2(dh_duv.x, dh_duv.y * tex_aspect);
         // ...and back out of raw-projection units into millimetres along t / b; see the 140 variant.
         vec2  slope = amplitude * vec2(g.x * cs + g.y * sn, -g.x * sn + g.y * cs) * units_per_mm;
 
-        // t and b came out of the texture frame, so the gradient is in it too and has to come back
-        // to world before it perturbs the normal the lighting uses.
-        vec3 gradient = tex_rot_inv * (slope.x * t + slope.y * b);
+        vec3 gradient = slope.x * t + slope.y * b;
         gradient -= triangle_normal * dot(triangle_normal, gradient);
         triangle_normal = normalize(triangle_normal - gradient);
     }
@@ -396,9 +353,7 @@ void main()
     // Only the albedo - the specular term (intensity.y) stays white - so a coloured fragment reads as
     // the same material under the same light, and the relief this preview exists to show is unaffected.
     vec3 albedo = uniform_color.rgb;
-    bool uv_covered = tile_enabled ||
-                      (all(greaterThanEqual(color_uv, vec2(0.0))) && all(lessThan(color_uv, vec2(1.0))));
-    if (palette_count > 0 && has_color_tex && have_uv && weight > 0.0 && uv_covered)
+    if (palette_count > 0 && has_color_tex && have_uv && weight > 0.0)
         albedo = palette_rgb[nearest_palette_entry(texture2D(color_tex, color_uv).rgb)];
     gl_FragColor = vec4(vec3(intensity.y) + albedo * intensity.x, uniform_color.a);
 }

resources/shaders/110/texture_displacement_uvcheck.fs

@@ -16,16 +16,7 @@ uniform bool volume_mirrored;
 uniform int   mode;
 uniform float checker_freq;
 uniform float tiling_scale;
-// Tile extent along v, in mm. Zero or negative means "the same as tiling_scale" - see
-// layer_tile_mm() on the CPU side, so an unset uniform behaves exactly as this shader always did.
-uniform float tiling_scale_v;
-// Mirror along one tile axis; false by default, so an unset uniform is the old behaviour.
-uniform bool  mirror_u;
-uniform bool  mirror_v;
 uniform vec3  tex_anchor; // the volume's origin in world space
-// The placement's rotation, so this overlay projects the same frame the bake does. Identity, which
-// is what an unset uniform gives, leaves the old world-aligned behaviour.
-uniform mat3  tex_rot;    // world -> texture frame
 uniform float rotation_rad;
 uniform vec2  uv_offset;
 uniform bool  use_vertex_uv;
@@ -88,30 +79,13 @@ vec2 projection_raw(vec3 p, vec3 n)
     return vec2(box_face_sign(n.z) * p.x, -p.y);
 }
 
-
-// Tile extents in mm, u then v, with the v fallback layer_tile_mm() applies.
-vec2 tile_mm()
-{
-    float su = (tiling_scale > 1e-6) ? tiling_scale : 1.0;
-    return vec2(su, (tiling_scale_v > 1e-6) ? tiling_scale_v : su);
-}
-
-// +1 or -1 per axis, matching layer_mirror_sign() on the CPU side.
-vec2 mirror_sign()
-{
-    return vec2(mirror_u ? -1.0 : 1.0, mirror_v ? -1.0 : 1.0);
-}
-
 vec2 project_uv(vec3 p, vec3 n)
 {
     vec2 planar = projection_raw(p, n);
+    planar *= (tiling_scale > 1e-6) ? (1.0 / tiling_scale) : 1.0;
     float cs = cos(rotation_rad);
     float sn = sin(rotation_rad);
     vec2 r = vec2(planar.x * cs - planar.y * sn, planar.x * sn + planar.y * cs);
-    // Extents divide after the rotation, so a tile whose two sides differ stays a rotated rectangle
-    // instead of shearing. With a square tile the two orders agree exactly. The mirror is a sign on
-    // the same division, in the tile's own frame.
-    r *= mirror_sign() / tile_mm();
     r.y *= tex_aspect; // after the rotation, so the rotation stays a rotation rather than a shear
     return r + uv_offset;
 }
@@ -131,7 +105,7 @@ void main()
 
     // World space anchored at the volume's origin, like the bake and the shaded preview.
     vec3 triangle_normal = normalize(cross(dFdx(world_pos.xyz), dFdy(world_pos.xyz)));
-    vec3 tex_pos = tex_rot * (world_pos.xyz - tex_anchor);
+    vec3 tex_pos = world_pos.xyz - tex_anchor;
     if (volume_mirrored)
         triangle_normal = -triangle_normal;
 
@@ -139,7 +113,7 @@ void main()
     if (mode == 1) {
         base = heatmap(distortion);
     } else {
-        vec2 uv = use_vertex_uv ? vertex_uv : project_uv(tex_pos, tex_rot * triangle_normal);
+        vec2 uv = use_vertex_uv ? vertex_uv : project_uv(tex_pos, triangle_normal);
         vec2 c  = floor(uv * checker_freq);
         float check = mod(c.x + c.y, 2.0);
         base = (check < 0.5) ? vec3(0.22, 0.23, 0.26) : vec3(0.82, 0.83, 0.86);

resources/shaders/140/chromastack.fs

@@ -0,0 +1,94 @@
+#version 140
+
+// What a ChromaStack print looks like, shaded per fragment.
+//
+// The colour is not a vertex attribute and not a texture on the surface: it is read from the
+// fragment's own height. A filament swap happens at one print Z across the whole plate, so the layer
+// a point sits in is what decides its colour - which means a band edge lands exactly on the layer
+// boundary however coarse the mesh is, and the same shader is right before a bake (on the preview's
+// displaced mesh) and after one (on the volume's own).
+//
+// On top of that it tries to look like printed plastic rather than like flat shaded colour: the
+// terminator is wrapped and the grazing angles glow, because filament translucent enough to be worth
+// stacking scatters light rather than reflecting it; and the walls carry layer lines, because that is
+// the first thing the eye uses to read a print as a print.
+
+#define INTENSITY_CORRECTION 0.6
+const vec3 LIGHT_TOP_DIR = vec3(-0.4574957, 0.4574957, 0.7624929);
+#define LIGHT_TOP_DIFFUSE    (0.8 * INTENSITY_CORRECTION)
+#define LIGHT_TOP_SPECULAR   (0.125 * INTENSITY_CORRECTION)
+#define LIGHT_TOP_SHININESS  20.0
+const vec3 LIGHT_FRONT_DIR = vec3(0.6985074, 0.1397015, 0.6985074);
+#define LIGHT_FRONT_DIFFUSE  (0.3 * INTENSITY_CORRECTION)
+#define INTENSITY_AMBIENT    0.3
+
+// The stack's colours, one texel per printed layer, bottom to top. A texture rather than a uniform
+// array because a base can be several hundred layers and no uniform array survives that.
+uniform sampler2D palette;
+uniform float     palette_size;  // texels in `palette`, i.e. the stack's total printed layers
+uniform float     base_z;        // world Z of the bottom of the object
+uniform float     layer_height;  // mm
+// How translucent the stack reads, 0 to 1, from the transmission distance of the filament on top.
+// It drives the wrap and the rim: an opaque stack should not glow.
+uniform float     translucency;
+uniform float     alpha;
+
+in vec3  clipping_planes_dots;
+in vec3  eye_normal;
+in vec3  eye_position;
+in vec3  world_normal;
+in float world_z;
+
+out vec4 out_color;
+
+void main()
+{
+    if (any(lessThan(clipping_planes_dots, vec3(0.0))))
+        discard;
+
+    float z_rel  = world_z - base_z;
+    float layers = max(z_rel / layer_height, 0.0);
+    // How much of a layer one pixel covers. Everything below is faded out as this grows: layer lines
+    // finer than a pixel are noise, and a colour read from one layer out of the dozen under a pixel
+    // flickers as the camera moves.
+    float px     = fwidth(layers);
+
+    // The layer this fragment sits in, 1-based. Up close that is an exact texel of the palette; once a
+    // pixel spans a layer or more it slides to a continuous coordinate so the linear filter averages
+    // the bands it covers instead of picking one of them at random.
+    float stepped = clamp(ceil(layers - 1.0e-4), 1.0, palette_size) - 0.5;
+    float smooth_ = clamp(layers, 0.5, palette_size - 0.5);
+    float u       = mix(stepped, smooth_, clamp(px - 0.5, 0.0, 1.0)) / palette_size;
+    vec3  albedo  = texture(palette, vec2(u, 0.5)).rgb;
+
+    vec3 N = normalize(eye_normal);
+    vec3 V = normalize(-eye_position);
+
+    // Layer lines, on the walls only: a terrace is one layer's top surface and has none to show. The
+    // groove runs round the part at every layer boundary, so it is a function of height alone.
+    float wall   = 1.0 - abs(world_normal.z);
+    float phase  = fract(layers);
+    float groove = smoothstep(0.0, 0.16, phase) * smoothstep(1.0, 0.84, phase);
+    groove       = mix(groove, 0.65, clamp(px * 1.5, 0.0, 1.0)); // fade to flat when too fine to resolve
+    float lines  = mix(1.0, 0.84 + 0.16 * groove, wall * 0.9);
+    albedo      *= lines;
+
+    // Diffuse, with a wrapped terminator: light entering translucent filament comes back out a little
+    // way round the curve, so the shaded side never goes as dark as an opaque surface's would.
+    float wrap    = 0.25 + 0.45 * translucency;
+    float ndotl   = dot(N, LIGHT_TOP_DIR);
+    float diffuse = max((ndotl + wrap) / (1.0 + wrap), 0.0) * LIGHT_TOP_DIFFUSE;
+    diffuse      += max((dot(N, LIGHT_FRONT_DIR) + wrap) / (1.0 + wrap), 0.0) * LIGHT_FRONT_DIFFUSE;
+
+    // Specular, strongest along a layer line, which is what makes a printed wall glint in stripes
+    // rather than evenly.
+    float spec = LIGHT_TOP_SPECULAR * pow(max(dot(V, reflect(-LIGHT_TOP_DIR, N)), 0.0), LIGHT_TOP_SHININESS);
+    spec      *= mix(1.0, 0.35 + 0.65 * groove, wall * 0.9);
+
+    // Scattering out of the edges: a translucent part lights up where it is seen at a grazing angle
+    // and where it is thin. Tinted by the filament, because what comes back out has been through it.
+    float rim = pow(1.0 - clamp(dot(N, V), 0.0, 1.0), 4.0);
+
+    vec3 color = albedo * (INTENSITY_AMBIENT + diffuse) + vec3(spec) + albedo * rim * translucency * 0.55;
+    out_color  = vec4(color, alpha);
+}

resources/shaders/140/chromastack.vs

@@ -0,0 +1,40 @@
+#version 140
+
+uniform mat4 view_model_matrix;
+uniform mat4 projection_matrix;
+uniform mat3 view_normal_matrix;
+uniform mat4 volume_world_matrix;
+// The volume's normal matrix in world space, so the fragment shader can tell a wall from a terrace -
+// which is what decides whether layer lines show at all.
+uniform mat3 world_normal_matrix;
+// Clipping plane, x = min z, y = max z (the FFF/SLA preview's top/bottom cut).
+uniform vec2 z_range;
+// Clipping plane, general orientation (the section view).
+uniform vec4 clipping_plane;
+
+// GLModel's P3N3 layout, which is what init_from(indexed_triangle_set) produces - this shader needs
+// nothing per vertex that the mesh does not already carry, because the colour comes from the
+// fragment's own height rather than from anything baked into the vertices.
+in vec3 v_position;
+in vec3 v_normal;
+
+out vec3  clipping_planes_dots;
+out vec3  eye_normal;
+out vec3  eye_position;
+out vec3  world_normal;
+out float world_z;
+
+void main()
+{
+    vec4 model_pos = vec4(v_position, 1.0);
+    vec4 world_pos = volume_world_matrix * model_pos;
+    world_z        = world_pos.z;
+    world_normal   = normalize(world_normal_matrix * v_normal);
+    eye_normal     = normalize(view_normal_matrix * v_normal);
+
+    vec4 eye_pos = view_model_matrix * model_pos;
+    eye_position = eye_pos.xyz;
+    gl_Position  = projection_matrix * eye_pos;
+
+    clipping_planes_dots = vec3(dot(world_pos, clipping_plane), world_pos.z - z_range.x, z_range.y - world_pos.z);
+}

resources/shaders/140/texture_displacement_shaded.fs

@@ -95,11 +95,6 @@ uniform int       palette_b[64];
 uniform float     prefer_pure_de; // PREFER_PURE_DE: how much better than a single filament a mix must be
 uniform sampler2D color_tex;     // the layer's colour image, sampled at the same uv as the height
 uniform bool      has_color_tex;
-// False means the texture is a single placement, not a repeat. The sampler's wrap mode can express
-// that for the *height* - a zero border reads as no displacement - but not for colour, because a
-// zero border is indistinguishable from the image genuinely being black there. So the uv is tested
-// instead, which is what sample_layer_color() does on the CPU side.
-uniform bool      tile_enabled;
 uniform bool volume_mirrored;
 
 uniform mat4 view_model_matrix;
@@ -109,14 +104,6 @@ uniform sampler2D height_tex;
 uniform vec2      height_tex_texel; // (1/width, 1/height) of height_tex
 uniform float      depth_mm;
 uniform float      tiling_scale;
-// Tile extent along v, in mm. Zero or negative means "the same as tiling_scale" - the square tile
-// every layer had before the two axes could differ - which is also what layer_tile_mm() does on the
-// CPU side, so an unset uniform behaves exactly as this shader always did.
-uniform float      tiling_scale_v;
-// Mirror the texture along one axis of the tile. Both default to false, which GLSL guarantees for an
-// uninitialised bool uniform, so a shader whose caller has not set them behaves exactly as before.
-uniform bool       mirror_u;
-uniform bool       mirror_v;
 // Height map width / height. Scales the v axis so a non-square image keeps its proportions
 // instead of being squeezed into a square tile - mirrors libslic3r's apply_uv_transform().
 uniform float      tex_aspect;
@@ -126,12 +113,6 @@ uniform bool       invert;
 uniform float      midlevel;      // the height that means "don't move"; needed by the parallax step
 uniform vec3       eye_model_pos; // camera position in the texture frame (world minus tex_anchor)
 uniform vec3       tex_anchor;    // the volume's origin in world space: the texture frame's origin
-// The placement's rotation and its inverse, so this shader turns a world position into the texture
-// frame exactly as texture_displacement_bake_frame() does. Both are passed rather than one
-// transposed here, because GLSL 110 has no transpose(). Identity leaves the old world-aligned
-// behaviour, which is what an unset uniform gives.
-uniform mat3       tex_rot;     // world -> texture frame
-uniform mat3       tex_rot_inv; // texture frame -> world
 uniform bool       use_vertex_uv; // true: sample at vertex_uv with a derived tangent frame (LSCM)
 // A 2x3 affine (columns packed as lin = (m00, m01, m10, m11), tr = (m02, m12)) applied to the uv of
 // the island currently being dragged in the UV editor (island_active > 0.5). Identity when nothing is
@@ -247,29 +228,13 @@ void projection_axes(vec3 p, vec3 n, out vec3 t, out vec3 b, out vec2 units_per_
     }
 }
 
-// Tile extents in mm, u then v, with the v fallback layer_tile_mm() applies.
-vec2 tile_mm()
-{
-    float su = (tiling_scale > 1e-6) ? tiling_scale : 1.0;
-    return vec2(su, (tiling_scale_v > 1e-6) ? tiling_scale_v : su);
-}
-
-// +1 or -1 per axis, matching layer_mirror_sign() on the CPU side.
-vec2 mirror_sign()
-{
-    return vec2(mirror_u ? -1.0 : 1.0, mirror_v ? -1.0 : 1.0);
-}
-
 vec2 project_uv(vec3 p, vec3 n)
 {
     vec2 planar = projection_raw(p, n);
+    planar *= (tiling_scale > 1e-6) ? (1.0 / tiling_scale) : 1.0;
     float cs = cos(rotation_rad);
     float sn = sin(rotation_rad);
     vec2 r = vec2(planar.x * cs - planar.y * sn, planar.x * sn + planar.y * cs);
-    // Extents divide after the rotation, so a tile whose two sides differ stays a rotated rectangle
-    // instead of shearing. With a square tile the two orders agree exactly. The mirror is a sign on
-    // the same division, in the tile's own frame.
-    r *= mirror_sign() / tile_mm();
     // After the rotation, so the rotation stays a rotation rather than becoming a shear.
     r.y *= tex_aspect;
     return r + uv_offset;
@@ -340,12 +305,9 @@ void main()
     // printed part whatever the instance's scale or rotation, so the preview has to project from the
     // world position and perturb the world normal.
     vec3 triangle_normal = normalize(cross(dFdx(world_pos.xyz), dFdy(world_pos.xyz)));
-    vec3 tex_pos = tex_rot * (world_pos.xyz - tex_anchor); // the frame the texture is projected in, as the bake does
+    vec3 tex_pos = world_pos.xyz - tex_anchor; // the frame the texture is projected in, as the bake does
     if (volume_mirrored)
         triangle_normal = -triangle_normal;
-    // The same normal in that frame. triangle_normal itself stays in world, because the lighting at
-    // the end of this shader needs it there.
-    vec3 tex_normal = tex_rot * triangle_normal;
 
     // Where the colour is read from. Both branches below already compute the uv this fragment's
     // *height* came from - including the parallax-marched one on the triplanar path - and the colour
@@ -389,14 +351,14 @@ void main()
         // closed-form uv here, unlike the LSCM case.
         vec3 t, b;
         vec2 units_per_mm;
-        projection_axes(tex_pos, tex_normal, t, b, units_per_mm);
+        projection_axes(tex_pos, triangle_normal, t, b, units_per_mm);
 
         // Parallax occlusion mapping: march the view ray through the height shell and shade at the
         // first point where it drops below the displaced surface (see header).
         float amp      = (invert ? -1.0 : 1.0) * depth_mm * clamp(weight, 0.0, 1.0);
-        vec3  view_dir = normalize(tex_rot * eye_model_pos - tex_pos);
-        float v_dot_n  = dot(view_dir, tex_normal);
-        vec2  uv       = project_uv(tex_pos, tex_normal);
+        vec3  view_dir = normalize(eye_model_pos - tex_pos);
+        float v_dot_n  = dot(view_dir, triangle_normal);
+        vec2  uv       = project_uv(tex_pos, triangle_normal);
 
         // The shell the displaced surface lives inside, as signed heights along the normal. Taken from
         // both ends of h in [0, 1] so it stays correct for an inverted layer or a raised midlevel,
@@ -408,18 +370,18 @@ void main()
         // How far, in mm, sweeping the ray across the shell slides the sample point sideways. Below half
         // a texel there is no parallax to find and the march would be pure cost - which is the common
         // case of looking straight down at a surface.
-        float sweep = length(view_dir - tex_normal * v_dot_n) * (h_hi - h_lo) / max(v_dot_n, 1e-4);
-        if (v_dot_n > 0.05 && sweep > 0.5 * min(tile_mm().x, tile_mm().y) * height_tex_texel.x) {
+        float sweep = length(view_dir - triangle_normal * v_dot_n) * (h_hi - h_lo) / max(v_dot_n, 1e-4);
+        if (v_dot_n > 0.05 && sweep > 0.5 * tiling_scale * height_tex_texel.x) {
             // A point at ray parameter s (model_pos + view_dir * s) sits at height s * v_dot_n above the
             // undisplaced surface. Start at the top of the shell, where the ray is outside the surface
             // by construction, and step inward; the crossing is what this pixel actually sees.
             float s        = h_hi / v_dot_n;
             float ds       = (h_hi - h_lo) / (v_dot_n * float(PARALLAX_STEPS));
-            vec2  prev_uv  = project_uv(tex_pos + view_dir * s, tex_normal);
+            vec2  prev_uv  = project_uv(tex_pos + view_dir * s, triangle_normal);
             float prev_gap = h_hi - amp * (H_AT(prev_uv) - midlevel); // >= 0 by construction
             for (int i = 0; i < PARALLAX_STEPS; ++i) {
                 s -= ds;
-                vec2  cur_uv = project_uv(tex_pos + view_dir * s, tex_normal);
+                vec2  cur_uv = project_uv(tex_pos + view_dir * s, triangle_normal);
                 float gap    = s * v_dot_n - amp * (H_AT(cur_uv) - midlevel);
                 if (gap <= 0.0) {
                     // Crossed between the last two samples - interpolating the hit is what stops it
@@ -443,28 +405,23 @@ void main()
         // Central difference, per uv unit (not per texel).
         vec2 dh_duv = vec2((hR - hL) / (2.0 * height_tex_texel.x), (hU - hD) / (2.0 * height_tex_texel.y));
 
-        // uv -> mm is the reciprocal of the tile extents, applied per component below, which is what
-        // turns the uv-space gradient into a real surface slope.
-        vec2  tile       = tile_mm();
-        float amplitude  = (invert ? -1.0 : 1.0) * depth_mm * clamp(weight, 0.0, 1.0);
+        // uv -> mm is 1/tiling_scale for the triplanar projection, so this turns the uv-space
+        // gradient into a real surface slope.
+        float inv_tiling = (tiling_scale > 1e-6) ? (1.0 / tiling_scale) : 1.0;
+        float amplitude  = (invert ? -1.0 : 1.0) * depth_mm * inv_tiling * clamp(weight, 0.0, 1.0);
         // uv was rotated by project_uv() while t/b are the unrotated model axes, so rotate the
         // gradient back into the axes' frame.
         float cs = cos(rotation_rad);
         float sn = sin(rotation_rad);
-        // One uv unit is tile.x mm along u but tile.y / tex_aspect mm along v, so each component of
-        // the uv-space gradient carries its own factor before being rotated back into t/b.
-        // Each component carries its own extent and its own mirror sign, both of which project_uv()
-        // applied on the way in; without the sign a mirrored layer would light as though it were not.
-        vec2  ms    = mirror_sign();
-        vec2  g     = vec2(dh_duv.x * ms.x / tile.x, dh_duv.y * tex_aspect * ms.y / tile.y);
+        // One uv unit is tiling_scale mm along u but tiling_scale / tex_aspect mm along v, so the v
+        // component of the gradient carries the extra factor before being rotated back into t/b.
+        vec2  g     = vec2(dh_duv.x, dh_duv.y * tex_aspect);
         // ...and back out of raw-projection units into millimetres of travel along t / b, which is a
         // no-op except for the spherical longitude (see projection_axes()). After the inverse rotation,
         // because units_per_mm is expressed in the t/b frame rather than in uv.
         vec2  slope = amplitude * vec2(g.x * cs + g.y * sn, -g.x * sn + g.y * cs) * units_per_mm;
 
-        // t and b came out of the texture frame, so the gradient is in it too and has to come back
-        // to world before it perturbs the normal the lighting uses.
-        vec3 gradient = tex_rot_inv * (slope.x * t + slope.y * b);
+        vec3 gradient = slope.x * t + slope.y * b;
         gradient -= triangle_normal * dot(triangle_normal, gradient);
         triangle_normal = normalize(triangle_normal - gradient);
     }
@@ -484,9 +441,7 @@ void main()
     // Only the albedo - the specular term (intensity.y) stays white - so a coloured fragment reads as
     // the same material under the same light, and the relief this preview exists to show is unaffected.
     vec3 albedo = uniform_color.rgb;
-    bool uv_covered = tile_enabled ||
-                      (all(greaterThanEqual(color_uv, vec2(0.0))) && all(lessThan(color_uv, vec2(1.0))));
-    if (palette_count > 0 && has_color_tex && have_uv && weight > 0.0 && uv_covered)
+    if (palette_count > 0 && has_color_tex && have_uv && weight > 0.0)
         albedo = palette_rgb[nearest_palette_entry(texture(color_tex, color_uv).rgb)];
     out_color = vec4(vec3(intensity.y) + albedo * intensity.x, uniform_color.a);
 }

resources/shaders/140/texture_displacement_uvcheck.fs

@@ -25,16 +25,7 @@ uniform bool volume_mirrored;
 uniform int   mode;          // 0 checker, 1 distortion
 uniform float checker_freq;  // checker squares per uv unit (one uv unit == one texture tile)
 uniform float tiling_scale;
-// Tile extent along v, in mm. Zero or negative means "the same as tiling_scale" - see
-// layer_tile_mm() on the CPU side, so an unset uniform behaves exactly as this shader always did.
-uniform float tiling_scale_v;
-// Mirror along one tile axis; false by default, so an unset uniform is the old behaviour.
-uniform bool  mirror_u;
-uniform bool  mirror_v;
 uniform vec3  tex_anchor; // the volume's origin in world space
-// The placement's rotation, so this overlay projects the same frame the bake does. Identity, which
-// is what an unset uniform gives, leaves the old world-aligned behaviour.
-uniform mat3  tex_rot;    // world -> texture frame
 uniform float rotation_rad;
 uniform vec2  uv_offset;
 uniform bool  use_vertex_uv;
@@ -99,30 +90,13 @@ vec2 projection_raw(vec3 p, vec3 n)
     return vec2(box_face_sign(n.z) * p.x, -p.y);
 }
 
-
-// Tile extents in mm, u then v, with the v fallback layer_tile_mm() applies.
-vec2 tile_mm()
-{
-    float su = (tiling_scale > 1e-6) ? tiling_scale : 1.0;
-    return vec2(su, (tiling_scale_v > 1e-6) ? tiling_scale_v : su);
-}
-
-// +1 or -1 per axis, matching layer_mirror_sign() on the CPU side.
-vec2 mirror_sign()
-{
-    return vec2(mirror_u ? -1.0 : 1.0, mirror_v ? -1.0 : 1.0);
-}
-
 vec2 project_uv(vec3 p, vec3 n)
 {
     vec2 planar = projection_raw(p, n);
+    planar *= (tiling_scale > 1e-6) ? (1.0 / tiling_scale) : 1.0;
     float cs = cos(rotation_rad);
     float sn = sin(rotation_rad);
     vec2 r = vec2(planar.x * cs - planar.y * sn, planar.x * sn + planar.y * cs);
-    // Extents divide after the rotation, so a tile whose two sides differ stays a rotated rectangle
-    // instead of shearing. With a square tile the two orders agree exactly. The mirror is a sign on
-    // the same division, in the tile's own frame.
-    r *= mirror_sign() / tile_mm();
     r.y *= tex_aspect; // after the rotation, so the rotation stays a rotation rather than a shear
     return r + uv_offset;
 }
@@ -143,7 +117,7 @@ void main()
 
     // World space anchored at the volume's origin, like the bake and the shaded preview.
     vec3 triangle_normal = normalize(cross(dFdx(world_pos.xyz), dFdy(world_pos.xyz)));
-    vec3 tex_pos = tex_rot * (world_pos.xyz - tex_anchor);
+    vec3 tex_pos = world_pos.xyz - tex_anchor;
     if (volume_mirrored)
         triangle_normal = -triangle_normal;
 
@@ -151,7 +125,7 @@ void main()
     if (mode == 1) {
         base = heatmap(distortion);
     } else {
-        vec2 uv = use_vertex_uv ? vertex_uv : project_uv(tex_pos, tex_rot * triangle_normal);
+        vec2 uv = use_vertex_uv ? vertex_uv : project_uv(tex_pos, triangle_normal);
         vec2 c  = floor(uv * checker_freq);
         float check = mod(c.x + c.y, 2.0);
         // Two distinct greys, plus a faint tint on one set so orientation is readable at a glance.

resources/web/dialog/PluginsDialog/index.html

@@ -18,7 +18,7 @@
   <script src="../js/plugin-config-ui.js"></script>
   <script src="./index.js"></script>
   <script src="./plugin-sort.js"></script>
-  <script src="../js/fuzzy-search.js"></script>
+  <script src="../../js/fuzzy-search.js"></script>
   <script src="./plugin-search.js"></script>
 </head>
 <body onLoad="OnInit()">

resources/web/homepage/js/home.js

@@ -156,12 +156,16 @@ function SetCloudMode( bOn )
 
 	m_CloudMode=bOn;
 	$("body").toggleClass("cloud-mode", bOn);
+
+	if( bOn )
+	{
+		$(".BtnItem").removeClass("BtnItemSelected");
+		$(".BtnShortcut").addClass("BtnItemSelected");
+	}
 }
 
 function GotoCloud()
 {
-	$(".BtnItem").removeClass("BtnItemSelected");
-	$(".BtnShortcut").addClass("BtnItemSelected");
 	SetCloudMode(true);
 	SendSimpleCommand("homepage_open_cloud");
 }

scripts/flatpak/com.orcaslicer.OrcaSlicer.yml

@@ -336,10 +336,10 @@ modules:
         sha256: 3d22d46c566b1761a0e15ea397589b3a5f36ac09b7c785382e6470156c04247f
         dest: external-packages/PNG
 
-      # libjpeg-turbo 3.0.1
+      # libjpeg-turbo 3.2.0
       - type: file
-        url: https://github.com/libjpeg-turbo/libjpeg-turbo/archive/refs/tags/3.0.1.zip
-        sha256: d6d99e693366bc03897677650e8b2dfa76b5d6c54e2c9e70c03f0af821b0a52f
+        url: https://github.com/libjpeg-turbo/libjpeg-turbo/archive/refs/tags/3.2.0.zip
+        sha256: 1b4c2cb6cba1ca5e53b9e85eeeab02154cab4b70894c96e97a9c747a26d19d63
         dest: external-packages/JPEG
 
       # Freetype 2.12.1

scripts/flatpak/entrypoint

@@ -1,7 +1,7 @@
 #!/usr/bin/env sh
 
-# Work-around https://gitlab.gnome.org/GNOME/gnome-build-meta/-/issues/754
-grep -q org.freedesktop.Platform.GL.nvidia /.flatpak-info && export WEBKIT_DISABLE_DMABUF_RENDERER=1
+# WebKit's NVIDIA DMABUF workaround (gnome-build-meta#754) is applied by the
+# binary itself, gated on the runtime WebKitGTK version (src/OrcaSlicer.cpp).
 
 # Work-around https://github.com/bambulab/BambuStudio/issues/3440
 # Use LC_NUMERIC instead of LC_ALL to prevent decimal separator issues

src/OrcaSlicer.cpp

@@ -163,6 +163,7 @@ using json = nlohmann::json;
 #include <X11/Xlib.h>
 #endif
 #include <unistd.h>
+#include "slic3r/GUI/Widgets/WebView.hpp"
 #endif
 
 #ifdef SLIC3R_GUI
@@ -1156,8 +1157,10 @@ int CLI::run(int argc, char **argv)
     // opt-in path. The ~2020-era WebKit2GTK XWayland compositor bug it
     // worked around appears fixed in WebKit2GTK >= 2.42; leaving it
     // unset preserves WebKit hardware acceleration on Device / Setup
-    // Wizard / login / store. The default path still applies it on
-    // XWayland sessions as a conservative fallback for older WebKit.
+    // Wizard / login / store. The default path applies it only where it
+    // forces GDK_BACKEND=x11 below (a build without EGL, so the app really
+    // runs through XWayland), and only when the linked WebKitGTK predates
+    // the fix (< 2.42), as a conservative fallback for older WebKit.
     // ------------------------------------------------------------------
     {
         const char* gdk_backend = ::getenv("GDK_BACKEND");
@@ -1165,6 +1168,14 @@ int CLI::run(int argc, char **argv)
         // honours the first backend in the comma-separated list).
         const bool x11_opt_in = gdk_backend && boost::starts_with(gdk_backend, "x11");
 
+        // NVIDIA's proprietary driver cannot back WebKitGTK's DMABUF renderer.
+        // Force the shared-memory transport, which keeps compositing and threaded
+        // scrolling; the second variable clears Debian/Ubuntu/openSUSE's isNVIDIA().
+        if (::access("/proc/driver/nvidia/version", F_OK) == 0 && WebView::SupportsForceSharedMemory()) {
+            ::setenv("WEBKIT_DMABUF_RENDERER_FORCE_SHM", "1", /* replace */ false);
+            ::setenv("WEBKIT_FORCE_DMABUF_RENDERER", "1", /* replace */ false);
+        }
+
         if (x11_opt_in) {
             // ===== X11 / XWayland (opted in via GDK_BACKEND=x11) =====
             BOOST_LOG_TRIVIAL(info) << "GDK_BACKEND=x11 detected; applying v2.3.2 X11/XWayland workarounds.";
@@ -1204,23 +1215,18 @@ int CLI::run(int argc, char **argv)
                 if (wayland_env && *wayland_env) {
                     BOOST_LOG_TRIVIAL(warning) << "Wayland detected but wxWidgets has no EGL support (wxHAS_EGL is OFF). Forcing X11 backend.";
                     ::setenv("GDK_BACKEND", "x11", true);
+
+                    // GTK will use X11 in a Wayland session here, which is the
+                    // one default-path case that really runs through XWayland.
+                    // WebKit2GTK accelerated compositing could fail silently
+                    // there on releases before 2.42, so force it off. Disabling
+                    // it also turns off threaded scrolling, hence the gate.
+                    if (WebView::NeedsCompositingWorkaround())
+                        ::setenv("WEBKIT_DISABLE_COMPOSITING_MODE", "1", /* replace */ false);
                 }
             }
             #endif
 
-            // WebKit2GTK compositing can fail under XWayland on older
-            // WebKit releases. Disable it only when both DISPLAY and
-            // WAYLAND_DISPLAY are set (i.e. an XWayland session is in
-            // use). On pure X11 or native Wayland, compositing is left
-            // enabled so WebKit retains HW acceleration.
-            {
-                const char* display_env_wk = ::getenv("DISPLAY");
-                const char* wayland_env_wk = ::getenv("WAYLAND_DISPLAY");
-                if (display_env_wk && *display_env_wk && wayland_env_wk && *wayland_env_wk) {
-                    ::setenv("WEBKIT_DISABLE_COMPOSITING_MODE", "1", /* replace */ false);
-                }
-            }
-
             // XInitThreads is needed before GStreamer may use Xlib. On
             // native Wayland without DISPLAY, GStreamer uses waylandsink
             // (no Xlib involved), so the call is skipped.

src/dev-utils/platform/unix/build_linux_image.sh.in

@@ -429,7 +429,8 @@ if [ "\$XDG_SESSION_TYPE" = "wayland" ] && [ "\$ZINK_DISABLE_OVERRIDE" != "1" ];
                 export __EGL_VENDOR_LIBRARY_FILENAMES=/usr/share/glvnd/egl_vendor.d/50_mesa.json
                 export MESA_LOADER_DRIVER_OVERRIDE=zink
                 export GALLIUM_DRIVER=zink
-                export WEBKIT_DISABLE_DMABUF_RENDERER=1
+                # WebKit's NVIDIA DMABUF workaround is applied by the binary itself,
+                # gated on the runtime WebKitGTK version (src/OrcaSlicer.cpp).
             fi
         fi
     fi

src/libslic3r/CMakeLists.txt

@@ -327,6 +327,8 @@ set(lisbslic3r_sources
     Geometry/VoronoiUtils.cpp
     Geometry/VoronoiUtils.hpp
     Geometry/VoronoiVisualUtils.hpp
+    ChromaStack.cpp
+    ChromaStack.hpp
     IMEXArrange.cpp
     IMEXArrange.hpp
     IMEXHelpers.cpp

src/libslic3r/ChromaStack.cpp

@@ -0,0 +1,394 @@
+#include "ChromaStack.hpp"
+
+#include <algorithm>
+#include <array>
+#include <cmath>
+#include <limits>
+#include <cstdlib>
+#include <numeric>
+
+#include <tbb/blocked_range.h>
+#include <tbb/parallel_for.h>
+
+#include <miniz.h>
+
+namespace Slic3r {
+
+// How many e-folding lengths of absorption count as "the colour beneath no longer shows". Three
+// passes 5% of the light, which is about where the eye stops seeing what is under a layer.
+static constexpr float OPAQUE_EFOLDS = 3.f;
+
+// TD is a filament property, so a zero or negative value in a preset is a data error rather than a
+// request for an infinitely opaque filament. Clamp instead: below this the filament is opaque at a
+// single layer anyway.
+static constexpr float TD_MIN_MM = 0.05f;
+static constexpr float TD_MAX_MM = 100.f;
+
+static float srgb_to_linear(float c)
+{
+    return c <= 0.04045f ? c / 12.92f : std::pow((c + 0.055f) / 1.055f, 2.4f);
+}
+
+static float linear_to_srgb(float c)
+{
+    return c <= 0.0031308f ? c * 12.92f : 1.055f * std::pow(c, 1.f / 2.4f) - 0.055f;
+}
+
+// CIELAB under D65, from non-linear sRGB in [0, 1]. Only used for nearest-colour search, so the
+// absolute white point matters far less than both sides using the same one.
+struct Lab
+{
+    float l = 0.f, a = 0.f, b = 0.f;
+};
+
+static Lab srgb_to_lab(float r, float g, float b)
+{
+    const float rl = srgb_to_linear(r), gl = srgb_to_linear(g), bl = srgb_to_linear(b);
+    const float x = (0.4124f * rl + 0.3576f * gl + 0.1805f * bl) / 0.95047f;
+    const float y = (0.2126f * rl + 0.7152f * gl + 0.0722f * bl);
+    const float z = (0.0193f * rl + 0.1192f * gl + 0.9505f * bl) / 1.08883f;
+    auto        f = [](float t) {
+        return t > 0.008856f ? std::cbrt(t) : (7.787f * t + 16.f / 116.f);
+    };
+    const float fx = f(x), fy = f(y), fz = f(z);
+    return Lab{ 116.f * fy - 16.f, 500.f * (fx - fy), 200.f * (fy - fz) };
+}
+
+static float delta_e(const Lab &p, const Lab &q)
+{
+    const float dl = p.l - q.l, da = p.a - q.a, db = p.b - q.b;
+    return std::sqrt(dl * dl + da * da + db * db);
+}
+
+int ChromaStack::relief_layers() const
+{
+    int n = 0;
+    for (const ChromaBand &band : bands)
+        n += std::max(0, band.layers);
+    return n;
+}
+
+int ChromaStack::total_layers() const { return std::max(0, base_layers) + relief_layers(); }
+
+float ChromaStack::depth_mm() const { return float(relief_layers()) * layer_height_mm; }
+
+std::vector<ColorRGB> chroma_palette(const ChromaStack &stack)
+{
+    const int total = stack.total_layers();
+    // No bands means no picture and no filament to print the base in, whatever base_layers says.
+    if (total <= 0 || stack.bands.empty() || stack.layer_height_mm <= 0.f)
+        return {};
+
+    // The filament each printed layer is in, bottom-up: the base first, in the bottom band's filament,
+    // then the bands themselves.
+    std::vector<const ChromaFilament *> per_layer;
+    per_layer.reserve(size_t(total));
+    for (int i = 0; i < std::max(0, stack.base_layers); ++i)
+        per_layer.push_back(&stack.bands.front().filament);
+    for (const ChromaBand &band : stack.bands)
+        for (int i = 0; i < band.layers; ++i)
+            per_layer.push_back(&band.filament);
+
+    std::vector<ColorRGB> palette(size_t(total) + 1);
+
+    // Linear light, walking up from an opaque base: C_0 is the bottom layer's own colour, so a
+    // column one layer tall comes out exactly that colour however translucent the filament is.
+    std::array<float, 3> acc{ srgb_to_linear(per_layer.front()->color.r()), srgb_to_linear(per_layer.front()->color.g()),
+                              srgb_to_linear(per_layer.front()->color.b()) };
+    auto                 store = [&palette](size_t k, const std::array<float, 3> &lin) {
+        palette[k] = ColorRGB(std::clamp(linear_to_srgb(lin[0]), 0.f, 1.f), std::clamp(linear_to_srgb(lin[1]), 0.f, 1.f),
+                              std::clamp(linear_to_srgb(lin[2]), 0.f, 1.f));
+    };
+    store(0, acc);
+
+    for (int k = 1; k <= total; ++k) {
+        const ChromaFilament &f  = *per_layer[size_t(k - 1)];
+        const float             td = std::clamp(f.td_mm, TD_MIN_MM, TD_MAX_MM);
+        // The depth at which this filament hides what is behind it, turned into the e-folding length
+        // of its absorption; see CHROMA_TD_OPAQUE_FRACTION.
+        const float efold = CHROMA_TD_OPAQUE_FRACTION * td / OPAQUE_EFOLDS;
+        const float t     = std::exp(-stack.layer_height_mm / efold);
+        const std::array<float, 3> own{ srgb_to_linear(f.color.r()), srgb_to_linear(f.color.g()),
+                                        srgb_to_linear(f.color.b()) };
+        for (int c = 0; c < 3; ++c)
+            acc[size_t(c)] = (1.f - t) * own[size_t(c)] + t * acc[size_t(c)];
+        store(size_t(k), acc);
+    }
+    return palette;
+}
+
+ChromaSolve chroma_solve(const uint8_t *rgb, int width, int height, const ChromaStack &stack)
+{
+    ChromaSolve solve;
+    const int     total  = stack.total_layers();
+    const int     base   = std::max(0, stack.base_layers);
+    const int     relief = stack.relief_layers();
+    if (rgb == nullptr || width <= 0 || height <= 0 || relief <= 0 || relief > CHROMA_MAX_RELIEF_LAYERS)
+        return solve;
+
+    const std::vector<ColorRGB> palette = chroma_palette(stack);
+    if (int(palette.size()) != total + 1)
+        return solve;
+
+    // Only the heights the solver may choose: base (nothing of the relief printed) up to total.
+    const int               choices = relief + 1;
+    const size_t                        choice_count = size_t(choices);
+    std::vector<Lab>                    choice_lab(choice_count);
+    std::vector<std::array<uint8_t, 3>> choice_rgb(choice_count);
+    for (int i = 0; i < choices; ++i) {
+        const ColorRGB &c = palette[size_t(base + i)];
+        choice_lab[size_t(i)] = srgb_to_lab(c.r(), c.g(), c.b());
+        choice_rgb[size_t(i)] = { uint8_t(std::lround(c.r() * 255.f)), uint8_t(std::lround(c.g() * 255.f)),
+                                  uint8_t(std::lround(c.b() * 255.f)) };
+    }
+
+    const size_t count = size_t(width) * size_t(height);
+    solve.width        = width;
+    solve.height       = height;
+    solve.levels.resize(count);
+    solve.grey.resize(count);
+    solve.preview.resize(count * 3);
+
+    std::vector<float> errors(count, 0.f);
+
+    auto pick = [&](float r, float g, float b, float &out_error) {
+        const Lab target = srgb_to_lab(std::clamp(r, 0.f, 1.f), std::clamp(g, 0.f, 1.f), std::clamp(b, 0.f, 1.f));
+        int       best   = 0;
+        float     best_d = std::numeric_limits<float>::max();
+        for (int i = 0; i < choices; ++i) {
+            const float d = delta_e(target, choice_lab[size_t(i)]);
+            if (d < best_d) {
+                best_d = d;
+                best   = i;
+            }
+        }
+        out_error = best_d;
+        return best;
+    };
+
+    // The grey level a chosen height is written as. Rounded to the nearest of 256 levels, which for
+    // any practical layer budget (a few dozen) maps every height to its own level exactly - the
+    // relief's steps stay on layer boundaries, which is the whole point of quantizing here.
+    auto grey_of = [relief](int choice) {
+        return uint8_t(std::lround(255.0 * double(choice) / double(relief)));
+    };
+
+    auto write = [&](size_t px, int choice, float err) {
+        solve.levels[px]        = uint16_t(base + choice);
+        solve.grey[px]          = grey_of(choice);
+        solve.preview[px * 3]     = choice_rgb[size_t(choice)][0];
+        solve.preview[px * 3 + 1] = choice_rgb[size_t(choice)][1];
+        solve.preview[px * 3 + 2] = choice_rgb[size_t(choice)][2];
+        errors[px]              = err;
+    };
+
+    if (stack.dither) {
+        // Floyd-Steinberg, serpentine so the error does not drift consistently to one side. The
+        // diffusion buffer holds the two rows the kernel reaches, in sRGB - which is where the
+        // artifact being traded away lives, and what every reference implementation diffuses in.
+        const size_t                      row_len = size_t(width);
+        std::vector<std::array<float, 3>> row(row_len), next(row_len);
+        for (int x = 0; x < width; ++x) {
+            const uint8_t *src = rgb + size_t(x) * 3;
+            row[size_t(x)]     = { src[0] / 255.f, src[1] / 255.f, src[2] / 255.f };
+        }
+        for (int y = 0; y < height; ++y) {
+            if (y + 1 < height)
+                for (int x = 0; x < width; ++x) {
+                    const uint8_t *src = rgb + (size_t(y + 1) * size_t(width) + size_t(x)) * 3;
+                    next[size_t(x)]    = { src[0] / 255.f, src[1] / 255.f, src[2] / 255.f };
+                }
+            const bool ltr = (y % 2) == 0;
+            for (int i = 0; i < width; ++i) {
+                const int    x  = ltr ? i : width - 1 - i;
+                const size_t px = size_t(y) * size_t(width) + size_t(x);
+                float        err = 0.f;
+                const int    choice = pick(row[size_t(x)][0], row[size_t(x)][1], row[size_t(x)][2], err);
+                write(px, choice, err);
+
+                std::array<float, 3> residual{};
+                for (int c = 0; c < 3; ++c)
+                    residual[size_t(c)] = row[size_t(x)][size_t(c)] - choice_rgb[size_t(choice)][size_t(c)] / 255.f;
+
+                const int  ahead = ltr ? x + 1 : x - 1;
+                const int  back  = ltr ? x - 1 : x + 1;
+                auto spread = [&](std::vector<std::array<float, 3>> &dst, int at, float w) {
+                    if (at < 0 || at >= width)
+                        return;
+                    for (int c = 0; c < 3; ++c)
+                        dst[size_t(at)][size_t(c)] += residual[size_t(c)] * w;
+                };
+                spread(row, ahead, 7.f / 16.f);
+                if (y + 1 < height) {
+                    spread(next, back, 3.f / 16.f);
+                    spread(next, x, 5.f / 16.f);
+                    spread(next, ahead, 1.f / 16.f);
+                }
+            }
+            row.swap(next);
+        }
+    } else {
+        tbb::parallel_for(tbb::blocked_range<int>(0, height), [&](const tbb::blocked_range<int> &range) {
+            for (int y = range.begin(); y < range.end(); ++y)
+                for (int x = 0; x < width; ++x) {
+                    const size_t   px  = size_t(y) * size_t(width) + size_t(x);
+                    const uint8_t *src = rgb + px * 3;
+                    float          err = 0.f;
+                    write(px, pick(src[0] / 255.f, src[1] / 255.f, src[2] / 255.f, err), err);
+                }
+        });
+    }
+
+    solve.mean_delta_e = count == 0 ? 0. : std::accumulate(errors.begin(), errors.end(), 0.) / double(count);
+    solve.max_delta_e  = errors.empty() ? 0. : double(*std::max_element(errors.begin(), errors.end()));
+    return solve;
+}
+
+float chroma_layers_to_opaque(float td_mm, float layer_height_mm)
+{
+    if (layer_height_mm <= 0.f)
+        return 0.f;
+    return CHROMA_TD_OPAQUE_FRACTION * std::clamp(td_mm, TD_MIN_MM, TD_MAX_MM) / layer_height_mm;
+}
+
+// Below about one dE two colours are the same colour to the eye. Half of that is the threshold for
+// "this layer changed something", so a step has to be comfortably invisible before a layer is called
+// wasted.
+static constexpr float VISIBLE_DE = 0.5f;
+
+static float step_delta_e(const ColorRGB &a, const ColorRGB &b)
+{
+    return delta_e(srgb_to_lab(a.r(), a.g(), a.b()), srgb_to_lab(b.r(), b.g(), b.b()));
+}
+
+int chroma_distinct_levels(const ChromaStack &stack)
+{
+    const std::vector<ColorRGB> palette = chroma_palette(stack);
+    const int                   base    = std::max(0, stack.base_layers);
+    if (int(palette.size()) <= base + 1)
+        return 0;
+    int n = 0;
+    for (size_t i = size_t(base) + 1; i < palette.size(); ++i)
+        if (step_delta_e(palette[i - 1], palette[i]) > VISIBLE_DE)
+            ++n;
+    return n;
+}
+
+void chroma_trim_bands(ChromaStack &stack)
+{
+    // Walked over the stack's own palette rather than over the filaments' TD, because how fast a band
+    // runs out of anything to say depends on what is under it as much as on the filament. A light band
+    // over a dark one is finished in four or five layers; a dark one over a light one goes on paying
+    // for many more, long past the depth at which it is nominally opaque, because the last few percent
+    // of light coming through still moves a near-black colour perceptibly.
+    const std::vector<ColorRGB> palette = chroma_palette(stack);
+    if (palette.empty())
+        return;
+    int layer = std::max(0, stack.base_layers); // palette index of the layer under the band being looked at
+    for (ChromaBand &band : stack.bands) {
+        int keep = 1;
+        for (int i = 1; i <= band.layers; ++i) {
+            const size_t cur = size_t(layer) + size_t(i);
+            if (cur >= palette.size())
+                break;
+            if (step_delta_e(palette[cur - 1], palette[cur]) > VISIBLE_DE)
+                keep = i; // the last layer of this band that still changes the colour
+        }
+        layer += band.layers;
+        band.layers = std::clamp(keep, 1, band.layers);
+    }
+}
+
+int chroma_level_at(const ChromaStack &stack, double z_mm)
+{
+    const int total = stack.total_layers();
+    if (total <= 0 || stack.layer_height_mm <= 0.f)
+        return 0;
+    // Layer k occupies ((k-1)h, k h], so a point inside it rounds up. The epsilon keeps a surface
+    // sitting exactly on a layer boundary in the layer that printed it rather than in the one above,
+    // which is the difference between a terrace showing its own colour and showing the next band's.
+    const int k = int(std::ceil(z_mm / double(stack.layer_height_mm) - 1e-6));
+    return std::clamp(k, 1, total);
+}
+
+std::vector<unsigned char> chroma_encode_height_png(const ChromaSolve &solve)
+{
+    if (solve.empty())
+        return {};
+    size_t      size = 0;
+    void       *png  = tdefl_write_image_to_png_file_in_memory_ex(solve.grey.data(), solve.width, solve.height, 1, &size,
+                                                                  MZ_DEFAULT_LEVEL, MZ_FALSE);
+    if (png == nullptr || size == 0)
+        return {};
+    const unsigned char       *bytes = static_cast<const unsigned char *>(png);
+    std::vector<unsigned char> out(bytes, bytes + size);
+    mz_free(png);
+    return out;
+}
+
+std::vector<ChromaRange> chroma_ranges(const ChromaStack &stack, double z_bottom)
+{
+    std::vector<ChromaRange> out;
+    if (stack.layer_height_mm <= 0.f)
+        return out;
+    // The layer height is stored as a float, like every other dimension of a texture displacement
+    // layer, so multiplying it up leaves noise in the last few digits - 5 layers of 0.1 mm come out as
+    // 0.5000000074. The slicer derives its own layer boundaries from the print config in double
+    // precision and lands on 0.5 exactly, and a boundary that misses it by any amount is a boundary a
+    // layer can fall on the wrong side of. Rounding to the nanometre removes the noise without
+    // disturbing any layer height a profile can express.
+    const auto snap = [](double z) { return std::round(z * 1e6) / 1e6; };
+
+    // The base is printed in the bottom band's filament - it is what is loaded when the print starts -
+    // so it extends that band's range down to the first layer instead of being a swap of its own.
+    int  layer = std::max(0, stack.base_layers);
+    bool first = true;
+    for (const ChromaBand &band : stack.bands) {
+        if (band.layers <= 0)
+            continue;
+        const double z_min = snap(z_bottom + (first ? 0. : double(layer) * double(stack.layer_height_mm)));
+        first = false;
+        layer += band.layers;
+        const double z_max = snap(z_bottom + double(layer) * double(stack.layer_height_mm));
+        if (!out.empty() && out.back().filament == band.filament.filament)
+            out.back().z_max = z_max; // no swap happens here, so it is one range
+        else
+            out.push_back(ChromaRange{ z_min, z_max, band.filament.filament });
+    }
+    return out;
+}
+
+ChromaStack chroma_default_stack(const std::vector<ChromaFilament> &filaments, float layer_height_mm,
+                                     int base_layers, int relief_layers)
+{
+    ChromaStack stack;
+    stack.layer_height_mm = layer_height_mm > 0.f ? layer_height_mm : 0.08f;
+    stack.base_layers     = std::max(0, base_layers);
+    if (filaments.empty() || relief_layers <= 0)
+        return stack;
+
+    std::vector<ChromaFilament> ordered = filaments;
+    std::stable_sort(ordered.begin(), ordered.end(), [](const ChromaFilament &a, const ChromaFilament &b) {
+        // Relative luminance of the linear colour: the perceptual order, so a saturated yellow
+        // counts as light and a saturated blue as dark, which a plain r+g+b sum gets wrong.
+        auto lum = [](const ColorRGB &c) {
+            return 0.2126f * srgb_to_linear(c.r()) + 0.7152f * srgb_to_linear(c.g()) + 0.0722f * srgb_to_linear(c.b());
+        };
+        return lum(a.color) > lum(b.color);
+    });
+
+    const int count = int(ordered.size());
+    int       left  = relief_layers;
+    for (int i = 0; i < count; ++i) {
+        // Spread the remainder over the first bands rather than piling it onto the last one.
+        const int share = left / (count - i) + ((left % (count - i)) > 0 ? 1 : 0);
+        ChromaBand band;
+        band.filament = ordered[size_t(i)];
+        band.layers   = std::max(1, share);
+        left -= share;
+        stack.bands.push_back(band);
+    }
+    return stack;
+}
+
+} // namespace Slic3r

src/libslic3r/ChromaStack.hpp

@@ -0,0 +1,286 @@
+#ifndef slic3r_ChromaStack_hpp_
+#define slic3r_ChromaStack_hpp_
+
+#include <cstddef>
+#include <cstdint>
+#include <string>
+#include <vector>
+
+#include <cereal/cereal.hpp>
+#include <cereal/types/vector.hpp>
+
+#include "Color.hpp"
+
+namespace Slic3r {
+
+// ChromaStack-style colour printing: a picture is reproduced on a single face by stacking thin layers
+// of a few filaments, swapping filament at fixed layer boundaries, and varying the *height* of each
+// pixel's column. Where the column is low, the layers above it are missing and the filament below
+// shows; where it is high, the topmost filament covers everything beneath it. Because filament is
+// translucent, a column that stops part-way through a run shows a blend of that run's filament and
+// what is under it, which is what gives far more colours than there are filaments loaded.
+//
+// So there are only two things to compute, and they are both here:
+//  - the *palette*: the colour the stack shows at every possible column height (chroma_palette());
+//  - the *height map*: per pixel, the column height whose palette colour is closest to the picture
+//    (chroma_solve()).
+//
+// Everything else is machinery that already exists. The height map is handed to a texture
+// displacement layer as its height texture, so the existing projection/preview/bake pipeline turns
+// it into real geometry (see TextureDisplacement.hpp), and the filament swaps are handed to the
+// object's height ranges (ModelObject::layer_config_ranges), which is the slicer's own way of
+// saying "print this Z band with that filament".
+
+// The depth of filament that blocks essentially all the light behind it, as a fraction of the
+// filament's transmission distance (TD).
+//
+// TD's scale is empirical, not physical: its values run from about 1
+// for an opaque filament to 10 and above for a nearly clear one, and the community's calibration
+// prints (the "seashell" step test) read a TD off the step where the colour beneath stops showing.
+// That step is about a tenth of the TD number, which is the only thing this constant encodes - a
+// filament of TD 2 stops hiding what is under it at roughly 0.2 mm, i.e. after two or three layers,
+// while one of TD 8 needs most of a millimetre. The model below turns that into a per-layer
+// transmittance; see chroma_palette().
+static constexpr float CHROMA_TD_OPAQUE_FRACTION = 0.1f;
+
+// What a fresh stack gets above the base. Twenty-four layers is about two millimetres at a fine layer
+// height: enough steps for a continuous tonal ramp, and still a print that finishes in an evening.
+static constexpr int CHROMA_DEFAULT_RELIEF_LAYERS = 24;
+
+// The most picture layers the solve can express. A column's height above the base is stored as one
+// byte per pixel in the 8-bit grayscale height map the displacement reads, so beyond this two
+// different heights would have to share a grey level and the relief's steps would stop landing on
+// layer boundaries - which is the one thing the quantization exists to guarantee.
+//
+// It bounds the bands, not the stack: the base is whatever the model already is under the painted
+// face, often hundreds of layers on a tall part, and nothing about it is quantized into the height
+// map. As a limit on the picture it is not one in practice - 255 layers is 20 mm of relief at
+// 0.08 mm, far past where a picture gains anything.
+static constexpr int CHROMA_MAX_RELIEF_LAYERS = 255;
+
+// The most filament changes one stack may hold. Every band is a swap the printer has to make, so this is
+// a limit on patience rather than on the model; the palette itself is bounded by the layer count.
+static constexpr size_t CHROMA_MAX_BANDS = 16;
+
+// One filament as the colour model sees it: what colour it is, and how far light gets into it.
+struct ChromaFilament
+{
+    // 0-based index into the project's filaments, which is what a height range's `extruder` option
+    // (1-based) is written from. -1 for a filament that is not loaded, which the GUI refuses to
+    // solve with rather than silently printing in the wrong colour.
+    int      filament = -1;
+    ColorRGB color;
+    // The filament preset's own name, so the panel can say which filament this is. Six similar greys
+    // are six identical swatches and six identical numbers otherwise. Stored rather than looked up on
+    // demand because a stack outlives the project's filament list - a preset can be swapped out from
+    // under a saved stack, and then the name is the only record of what it was solved against.
+    std::string name;
+    // Transmission distance, mm, as the filament preset reports it
+    // (`filament_transmission_distance`). Clamped to a sane range when the palette is built: zero
+    // would make a layer infinitely opaque and a negative value is meaningless.
+    float    td_mm = 4.f;
+
+    template<class Archive> void serialize(Archive &ar)
+    {
+        float r = color.r(), g = color.g(), b = color.b();
+        ar(filament, r, g, b, td_mm, name);
+        color = ColorRGB(r, g, b);
+    }
+};
+
+// A run of consecutive layers printed in one filament. The stack is a list of these, bottom-up, so
+// a band boundary is exactly where the print pauses and the filament is changed.
+struct ChromaBand
+{
+    ChromaFilament filament;
+    int              layers = 1;
+
+    template<class Archive> void serialize(Archive &ar) { ar(filament, layers); }
+};
+
+// The whole recipe: which filaments, in which order, how thick a layer is, and how much of the
+// bottom is solid backing rather than picture.
+struct ChromaStack
+{
+    std::vector<ChromaBand> bands; // bottom-up; bands.front() is the one the base is printed in
+
+    // The print's layer height, mm. Taken from the print profile rather than chosen here - a stack
+    // solved against 0.08 mm layers prints as the wrong picture at 0.2 mm, so the GUI re-solves when
+    // the profile's layer height changes.
+    float layer_height_mm = 0.08f;
+
+    // Layers of solid backing under the picture: everything the model already is below the painted
+    // face. Not part of any band - the bands are the picture - but printed in the bottom band's
+    // filament, since that is what is loaded when the print starts.
+    //
+    // Keeping it out of the bands is what lets it be what it really is: a measurement of the model,
+    // often a hundred layers or more, while a band is a handful. Folding the two together put a
+    // three-digit number on a slider meant for single digits and made "how thick is the backing" and
+    // "how many layers does the first filament get" the same control, which they are not.
+    //
+    // The colour model treats the base as opaque. That is what makes a column's colour depend only on
+    // the layers above it, and why the panel warns about a base thin enough to let the plate show
+    // through.
+    int  base_layers = 4;
+
+    // Whether `base_layers` follows the model or was typed in. It follows by default, and it should:
+    // the painted face is wherever the geometry puts it, and a base that disagrees with it shifts
+    // every colour in the picture - the stack ends up anchored at the bottom of the part instead of
+    // under the face, so the picture's whole range is squeezed into the first few millimetres and
+    // everything above it clamps to the top band.
+    //
+    // It cannot be measured once and left: a layer is created, and its picture chosen, before there
+    // is any paint to measure, and the face moves again whenever the part is scaled or the layer
+    // height changes. So the panel re-measures it every frame while this is set.
+    bool base_auto = true;
+
+    // Spread the quantization error into neighbouring pixels (Floyd-Steinberg, serpentine) instead
+    // of rounding each pixel to its nearest palette colour on its own. A palette of twenty or thirty
+    // colours has to cover a photograph's whole range, so plain rounding leaves visible bands across
+    // a sky or a cheek; diffusing the error trades them for fine stipple, which at a tenth of a
+    // millimetre per pixel reads as the intermediate colour. Off for flat-coloured artwork, where
+    // stipple is the artifact rather than the cure.
+    bool dither = false;
+
+    // Total printed layers, base included: what the tallest column is.
+    int  total_layers() const;
+    // Layers the picture is built from, i.e. the bands' own total above the base.
+    int  relief_layers() const;
+    // The relief's height in mm, which is the `depth_mm` of the texture displacement layer the
+    // height map is handed to.
+    float depth_mm() const;
+    bool  empty() const { return bands.empty(); }
+
+    template<class Archive> void serialize(Archive &ar) { ar(bands, layer_height_mm, base_layers, dither, base_auto); }
+};
+
+// The colour the stack shows at every column height: entry k is a column of exactly k printed
+// layers, so the vector has total_layers() + 1 entries.
+//
+// The model is Beer-Lambert absorption, layer by layer. A layer of thickness h in a filament of
+// transmission distance TD passes
+//
+//     T = exp(-h / (CHROMA_TD_OPAQUE_FRACTION * TD / 3))
+//
+// of the light coming from under it and shows its own colour with the rest, so
+//
+//     C_k = (1 - T_k) * c_k + T_k * C_(k-1)
+//
+// walking up from an opaque base, C_0 = c_1. The divisor is what makes "opaque at
+// CHROMA_TD_OPAQUE_FRACTION * TD" true: three e-folding lengths pass 5% of the light, which is
+// where the colour beneath stops being visible to the eye.
+//
+// Beer-Lambert is not just the physically honest choice here, it is the only one that stays
+// consistent when applied per layer: transmittance multiplies, so a run of n layers of one filament
+// passes exp(-n h / d) whether it is composited in one step or n, and the palette does not depend on
+// how the stack happens to be cut into bands.
+//
+// Compositing is done on linear light and the entries are converted back to sRGB, because the
+// quantity being attenuated is light, not an sRGB code.
+std::vector<ColorRGB> chroma_palette(const ChromaStack &stack);
+
+// What chroma_solve() produced.
+struct ChromaSolve
+{
+    int width  = 0;
+    int height = 0;
+
+    // Per pixel, row-major and top-to-bottom like DecodedHeightTexture: the column's printed layer
+    // count, between base_layers and total_layers(). Sixteen bits because it counts from the build
+    // plate, and the base alone can be several hundred layers on a tall part.
+    std::vector<uint16_t> levels;
+    // The same thing as the 8-bit grayscale height map the displacement samples: 0 at base_layers,
+    // 255 at total_layers(). This is what gets encoded as a PNG into the layer's `image_data`, which
+    // is why the solve has to be the thing that owns the quantization - the relief's steps have to
+    // land on layer boundaries, and a grey level is the only channel the displacement reads.
+    std::vector<uint8_t> grey;
+    // Per pixel RGB: the palette colour the solve chose, i.e. what the print is predicted to look
+    // like. Shown in the panel next to the source picture, since the 3D view can only show the
+    // relief - the colours do not exist until the print swaps filament.
+    std::vector<uint8_t> preview;
+
+    // How far the prediction landed from the picture, as a mean and worst-case CIE76 dE over all
+    // pixels. The panel reports it because it is the only honest answer to "will this stack
+    // reproduce my image": under about 5 is a good match, over about 15 means the filaments cannot
+    // reach the picture's colours and no amount of layers will fix it.
+    double mean_delta_e = 0.;
+    double max_delta_e  = 0.;
+
+    bool empty() const { return width <= 0 || height <= 0 || levels.empty(); }
+};
+
+// Solves `rgb` (width * height * 3 bytes, row-major, top-to-bottom) against the stack's palette.
+// An empty stack, a zero-sized image or a stack with no relief layers gives an empty result.
+ChromaSolve chroma_solve(const uint8_t *rgb, int width, int height, const ChromaStack &stack);
+
+// How many layers of this filament it takes to hide what is under it, at this layer height. The
+// single most useful number about a stack and the one nothing else makes visible: it is what decides
+// whether a band is a gradient or a step.
+//
+// A filament hides what is behind it at about CHROMA_TD_OPAQUE_FRACTION of its TD (see
+// chroma_palette()), so this is that depth in layers. Below two the band has no intermediate colours
+// at all - the picture can only use the filament or not use it - which is what a coarse layer height
+// does to an otherwise good stack: TD 4 at 0.2 mm covers in two layers, the same filament at 0.08 mm
+// in five.
+float chroma_layers_to_opaque(float td_mm, float layer_height_mm);
+
+// How many of the picture's layers actually change the colour.
+//
+// A band eventually stops saying anything: once it hides what is under it, another layer of it shows
+// the colour the one below already showed. Those layers still cost print time and still take up
+// relief height, and nothing about the stack makes them visible - the band looks like ten layers of
+// gradient and is, say, five of gradient and five of flat.
+//
+// Measured against the palette rather than against TD, because how long a band keeps paying depends
+// on what is under it as much as on the filament: a light band over a dark one is finished in four or
+// five layers, while a dark one over a light one goes on moving the colour perceptibly well past the
+// depth at which it is nominally opaque.
+int chroma_distinct_levels(const ChromaStack &stack);
+
+// Shortens every band to its last layer that still changes the colour. Bands already shorter than
+// that are left alone: they are still inside their fade, where every layer counts.
+void chroma_trim_bands(ChromaStack &stack);
+
+// The printed layer a point `z_mm` above the bottom of the object falls in: 1-based, clamped to the
+// stack, 0 for a stack that prints nothing. Pair it with chroma_palette() to get the colour shown
+// there.
+//
+// This, rather than the height map, is what decides the colour a surface shows. A filament swap
+// happens at one print Z across the whole plate, so every surface crossing that height changes colour
+// whatever the picture says - the terraces of the relief, the walls between them, and the sides of the
+// model alike. Colouring a preview this way is therefore not an approximation of the print; it is what
+// the print does, and it keeps working after the relief has been baked into the mesh and the height
+// map is no longer attached to anything on screen.
+int chroma_level_at(const ChromaStack &stack, double z_mm);
+
+// The solve's height map as an 8-bit grayscale PNG, ready to be stored in a
+// TextureDisplacementLayer's `image_data`.
+std::vector<unsigned char> chroma_encode_height_png(const ChromaSolve &solve);
+
+// One filament swap as an object height range: the Z band, in the object's own coordinates (0 at its
+// bottom), that prints in `filament` (0-based).
+struct ChromaRange
+{
+    double z_min = 0.;
+    double z_max = 0.;
+    int    filament = -1;
+};
+
+// The stack's bands as height ranges, in object coordinates, starting at `z_bottom` (which is 0 for
+// a stack whose base starts at the object's first layer). Bands printed in the same filament as the
+// band below are merged, so a stack that repeats a filament does not produce a pointless swap.
+std::vector<ChromaRange> chroma_ranges(const ChromaStack &stack, double z_bottom = 0.);
+
+// A starting point for a stack: the given filaments, lightest at the bottom, each with an equal
+// share of `relief_layers` on top of `base_layers`.
+//
+// Light-to-dark is the convention filament-painting guides teach, and it is not arbitrary: a dark
+// filament over a light one reads as a midtone at one or two layers and as the dark colour at four,
+// so putting the dark filaments last is what buys the long tonal ramp. The reverse order gives a few
+// washed-out steps and then nothing.
+ChromaStack chroma_default_stack(const std::vector<ChromaFilament> &filaments, float layer_height_mm,
+                                     int base_layers, int relief_layers);
+
+} // namespace Slic3r
+
+#endif // slic3r_ChromaStack_hpp_

src/libslic3r/GCode.cpp

@@ -432,27 +432,36 @@ static std::vector<Vec2d> get_path_of_change_filament(const Print& print)
 
     std::string OozePrevention::pre_toolchange(GCode& gcodegen)
     {
-        std::string gcode;
-
         unsigned int extruder_id = gcodegen.writer().filament()->id();
-        const auto& filament_idle_temp = gcodegen.config().idle_temperature;
-        if (filament_idle_temp.get_at(extruder_id) == 0) {
-            // There is no idle temperature defined in filament settings.
-            // Use the delta value from print config.
-            if (gcodegen.config().standby_temperature_delta.value != 0) {
-                // we assume that heating is always slower than cooling, so no need to block
-                gcode += gcodegen.writer().set_temperature
-                (this->_get_temp(gcodegen) + gcodegen.config().standby_temperature_delta.value, false, extruder_id);
-                gcode.pop_back();
-                gcode += " ;cooldown\n"; // this is a marker for GCodeProcessor, so it can supress the commands when needed
-            }
-        } else {
-            // Use the value from filament settings. That one is absolute, not delta.
-            gcode += gcodegen.writer().set_temperature(filament_idle_temp.get_at(extruder_id), false, extruder_id);
-            gcode.pop_back();
-            gcode += " ;cooldown\n"; // this is a marker for GCodeProcessor, so it can supress the commands when needed
+
+        // Check if this tool will ever be used again in this print.
+        bool is_last_use = false;
+        if (gcodegen.m_print != nullptr && !gcodegen.m_print->tool_ordering().empty()) {
+            coordf_t current_z        = gcodegen.layer() ? gcodegen.layer()->print_z : gcodegen.writer().get_position().z();
+            const auto& tool_ordering = gcodegen.m_print->tool_ordering();
+            const auto& cfg           = gcodegen.config();
+            const int heater          = imex_physical_heater_for(cfg.is_imex.value, cfg.physical_extruder_map, int(extruder_id));
+            is_last_use               = tool_ordering.is_last_extrusion_layer(current_z, extruder_id);
+            for (unsigned int f : tool_ordering.all_extruders())
+                if (is_last_use && imex_physical_heater_for(cfg.is_imex.value, cfg.physical_extruder_map, int(f)) == heater)
+                    is_last_use = tool_ordering.is_last_extrusion_layer(current_z, f);
         }
 
+        int temp = 0;
+        if (!is_last_use) {
+            const auto& filament_idle_temp = gcodegen.config().idle_temperature;
+            if (filament_idle_temp.get_at(extruder_id) == 0) {
+                if (gcodegen.config().standby_temperature_delta.value == 0)
+                    return std::string();
+                temp = this->_get_temp(gcodegen) + gcodegen.config().standby_temperature_delta.value;
+            } else {
+                temp = filament_idle_temp.get_at(extruder_id);
+            }
+        }
+
+        std::string gcode = gcodegen.writer().set_temperature(temp, false, extruder_id);
+        gcode.pop_back();
+        gcode += " ;cooldown\n"; // this is a marker for GCodeProcessor, so it can supress the commands when needed
         return gcode;
     }
 

src/libslic3r/GCode.hpp

@@ -1004,6 +1004,7 @@ protected:
         size_t                                                  num_objects,
         size_t                                                  num_islands);
 
+    friend class OozePrevention;
     friend class Wipe;
     friend class WipeTowerIntegration;
     friend class PressureEqualizer;

src/libslic3r/GCode/ToolOrdering.cpp

@@ -1276,6 +1276,13 @@ void ToolOrdering::collect_extruder_statistics(bool prime_multi_material)
         sort_remove_duplicates(m_all_printing_extruders);
     }
 
+    m_last_layer_per_extruder.clear();
+    for (const LayerTools& lt : m_layer_tools) {
+        for (unsigned int ext : lt.extruders) {
+            m_last_layer_per_extruder[ext] = lt.print_z;
+        }
+    }
+
     if (prime_multi_material && ! m_all_printing_extruders.empty()) {
         // Reorder m_all_printing_extruders in the sequence they will be primed, the last one will be m_first_printing_extruder.
         // Then set m_first_printing_extruder to the 1st extruder primed.
@@ -3664,5 +3671,11 @@ int WipingExtrusions::get_support_interface_extruder_overrides(const PrintObject
     return -1;
 }
 
-
+bool ToolOrdering::is_last_extrusion_layer(coordf_t print_z, unsigned int extruder_id) const
+{
+    auto it = m_last_layer_per_extruder.find(extruder_id);
+    if (it == m_last_layer_per_extruder.end())
+        return true;
+    return print_z >= it->second - EPSILON;
+}
 } // namespace Slic3r

src/libslic3r/GCode/ToolOrdering.hpp

@@ -370,6 +370,7 @@ public:
                                                   std::vector<int> & initial_filaments);
 
     bool                has_non_support_filament(const PrintConfig &config);
+    bool                is_last_extrusion_layer(coordf_t print_z, unsigned int extruder_id) const;
 
 private:
     void				initialize_layers(std::vector<coordf_t> &zs);
@@ -413,6 +414,8 @@ private:
     // which keeps every legacy per-object code path bit-identical (loops over an empty map are
     // no-ops; downstream emission falls through to the per-object branch).
     std::map<unsigned int, std::map<LayerTools::MixedSubLayerGroup::VolumeKey, std::vector<size_t>>> m_gradient_volume_layers;
+    // Map: extruder_id -> print_z (z coordinate) of the last layer it is used on
+    std::map<unsigned int, coordf_t> m_last_layer_per_extruder;
     const PrintObject*         m_print_object_ptr = nullptr;
     Print*                     m_print;
     bool                       m_sorted = false;

src/libslic3r/PNGReadWrite.cpp

@@ -211,6 +211,35 @@ bool decode_colored_png(const ReadBuf &in_buf, ImageColorscale &out_img)
     return decode_colored_png(stream, out_img);
 }
 
+bool is_decodable_8bit_png(const ReadBuf &pngbuf)
+{
+    static const constexpr int PNG_SIG_BYTES = 8;
+
+    struct ReadBufStream stream{pngbuf};
+    std::vector<png_byte> sig(PNG_SIG_BYTES, 0);
+    stream.read(sig.data(), PNG_SIG_BYTES);
+    if (!png_check_sig(sig.data(), PNG_SIG_BYTES))
+        return false;
+
+    PNGDescr dsc;
+    dsc.png = png_create_read_struct(PNG_LIBPNG_VER_STRING, nullptr, nullptr, nullptr);
+    if (!dsc.png)
+        return false;
+    dsc.info = png_create_info_struct(dsc.png);
+    if (!dsc.info)
+        return false;
+
+    if (!png_read_header_guarded(dsc.png, dsc.info, &stream, PNG_SIG_BYTES))
+        return false;
+
+    const size_t color_type = png_get_color_type(dsc.png, dsc.info);
+    const size_t bit_depth  = png_get_bit_depth(dsc.png, dsc.info);
+    const bool   grey       = color_type == PNG_COLOR_TYPE_GRAY;
+    const bool   colour     = color_type == PNG_COLOR_TYPE_RGB || color_type == PNG_COLOR_TYPE_RGB_ALPHA;
+    return png_get_image_width(dsc.png, dsc.info) > 0 && png_get_image_height(dsc.png, dsc.info) > 0 &&
+           bit_depth == 8 && (grey || colour);
+}
+
 
 // Down to earth function to store a packed RGB image to file. Mostly useful for debugging purposes.
 // Based on https://www.lemoda.net/c/write-png/

src/libslic3r/PNGReadWrite.hpp

@@ -50,6 +50,9 @@ struct ReadBuf { const void *buf = nullptr; const size_t sz = 0; };
 
 bool is_png(const ReadBuf &pngbuf);
 
+// True if decode_png() or decode_colored_png() would accept these bytes, judged from the header alone.
+bool is_decodable_8bit_png(const ReadBuf &pngbuf);
+
 struct ReadBufStream: public IStream {
     const ReadBuf &rbuf_ref;
     size_t pos = 0;

src/libslic3r/Preset.cpp

@@ -1489,7 +1489,7 @@ static std::vector<std::string> s_Preset_print_options{
 static std::vector<std::string> s_Preset_filament_options {/*"filament_colour", */ "default_filament_colour", "required_nozzle_HRC", "filament_diameter", "pellet_flow_coefficient", "volumetric_speed_coefficients", "filament_type",
                                                           "filament_soluble", "filament_is_support", "filament_printable", "filament_extruder_compatibility",
     "filament_max_volumetric_speed", "filament_adaptive_volumetric_speed",
-    "filament_flow_ratio", "filament_density", "filament_adhesiveness_category", "filament_cost", "filament_minimal_purge_on_wipe_tower",
+    "filament_flow_ratio", "filament_density", "filament_transmission_distance", "filament_adhesiveness_category", "filament_cost", "filament_minimal_purge_on_wipe_tower",
     "filament_tower_interface_pre_extrusion_dist", "filament_tower_interface_pre_extrusion_length", "filament_tower_ironing_area", "filament_tower_interface_purge_volume",
     "filament_tower_interface_print_temp",
     "nozzle_temperature", "nozzle_temperature_initial_layer",

src/libslic3r/Print.cpp

@@ -228,6 +228,7 @@ bool Print::invalidate_state_by_config_options(const ConfigOptionResolver & /* n
         "filament_diameter",
          "volumetric_speed_coefficients",
         "filament_density",
+        "filament_transmission_distance",
         "filament_cost",
         "filament_notes",
         "outer_wall_acceleration",

src/libslic3r/PrintConfig.cpp

@@ -3440,6 +3440,20 @@ void PrintConfigDef::init_fff_params()
     def->mode = comAdvanced;
     def->set_default_value(new ConfigOptionFloats { 0. });
 
+    def = this->add("filament_transmission_distance", coFloats);
+    def->label = L("Transmission distance");
+    def->tooltip = L("How deep light gets into this filament, in millimetres, also called its "
+                     "transmission distance or TD. About 1 for an opaque filament, 4 to 7 for a usefully "
+                     "translucent one, 8 and above for a nearly clear one. Used only to predict the "
+                     "colours of a ChromaStack in the texture displacement tool; it has no effect on "
+                     "slicing. Measure it with a step test print, or take a published value for the "
+                     "filament.");
+    def->sidetext = L("mm");
+    def->min = 0;
+    def->max = 100;
+    def->mode = comAdvanced;
+    def->set_default_value(new ConfigOptionFloats { 4. });
+
     def = this->add("filament_type", coStrings);
     def->label = L("Type");
     def->tooltip = L("Filament material type");

src/libslic3r/PrintConfig.hpp

@@ -1709,6 +1709,9 @@ PRINT_CONFIG_CLASS_DEFINE(
     ((ConfigOptionStrings,             volumetric_speed_coefficients))
     ((ConfigOptionInts,              filament_adhesiveness_category))
     ((ConfigOptionFloats,              filament_density))
+    // Transmission distance: how deep light gets into this filament. Not used by
+    // slicing - it is the colour model of a ChromaStack stack (see ChromaStack.hpp).
+    ((ConfigOptionFloats,              filament_transmission_distance))
     ((ConfigOptionStrings,             filament_type))
     ((ConfigOptionBools,               filament_soluble))
     ((ConfigOptionStrings,             filament_ids))

src/libslic3r/TextureDisplacement.cpp

@@ -1,7 +1,6 @@
 #include "TextureDisplacement.hpp"
 
 #include <Eigen/Core>
-#include <Eigen/SVD>
 #include <Eigen/Geometry>
 #include <algorithm>
 #include <array>
@@ -284,7 +283,7 @@ bool height_texture_has_color(const TextureDisplacementLayer &layer)
     return decode_height_texture(layer).has_color();
 }
 
-DecodedHeightTexture decode_height_texture(const TextureDisplacementLayer &layer)
+DecodedHeightTexture decode_height_texture(const TextureDisplacementLayer &layer, bool keep_decoded)
 {
     DecodedHeightTexture result;
     if (layer.empty())
@@ -352,13 +351,15 @@ DecodedHeightTexture decode_height_texture(const TextureDisplacementLayer &layer
                 }
         }
 
-        std::lock_guard<std::mutex> lock(g_decoded_texture_cache.mutex);
-        // Opportunistically drop entries for image_data that no longer exists anywhere, so the
-        // cache doesn't grow without bound across many add/remove-texture cycles in a long session.
-        auto &entries = g_decoded_texture_cache.entries;
-        for (auto it = entries.begin(); it != entries.end();)
-            it = it->second.first.expired() ? entries.erase(it) : std::next(it);
-        entries[key] = { std::weak_ptr<const std::vector<unsigned char>>(layer.image_data), result };
+        if (keep_decoded) {
+            std::lock_guard<std::mutex> lock(g_decoded_texture_cache.mutex);
+            // Opportunistically drop entries for image_data that no longer exists anywhere, so the
+            // cache doesn't grow without bound across many add/remove-texture cycles in a long session.
+            auto &entries = g_decoded_texture_cache.entries;
+            for (auto it = entries.begin(); it != entries.end();)
+                it = it->second.first.expired() ? entries.erase(it) : std::next(it);
+            entries[key] = { std::weak_ptr<const std::vector<unsigned char>>(layer.image_data), result };
+        }
     }
 
     // The image-adjustment panel's live edits, on top of the decode and before the smoothing. Before,
@@ -723,10 +724,7 @@ V2Resolution recommend_v2_resolution(const indexed_triangle_set
         const DecodedHeightTexture &tex = decode_height_texture(layer);
         if (tex.width <= 0)
             continue;
-        // Texels are square in both axes (the aspect correction is what keeps them so), so the finer
-        // of the two tile extents sets the finest texel the refinement has to resolve.
-        const Vec2f tile  = layer_tile_mm(layer);
-        const float texel = std::min(tile.x(), tile.y()) / float(tex.width);
+        const float texel = layer.tiling_scale / float(tex.width);
         if (out.texel_mm <= 0.f || texel < out.texel_mm) {
             out.texel_mm        = texel;
             out.pixels_per_edge = analyze_texture_detail(layer).pixels_per_edge;
@@ -2168,7 +2166,7 @@ namespace {
 // of them was most of the per-sample arithmetic.
 struct UVTransform
 {
-    float inv_u, inv_v, cs, sn, aspect;
+    float scale, cs, sn, aspect;
     Vec2f offset;
 
     // `normalized_uv` drops the two steps that only mean something for a coordinate measured on the
@@ -2176,18 +2174,7 @@ struct UVTransform
     // rotation and offset. See apply_uv_transform(); TextureProjectionMethod::ModelUV is what needs it.
     UVTransform(const TextureDisplacementLayer &layer, float aspect_, bool normalized_uv = false)
     {
-        // The mirror folds into the reciprocal: it is a negation of the same coordinate, so there is
-        // nothing to do per sample beyond carrying the sign here. A normalized uv has no millimetre
-        // tiling to divide by, but a mirror is still a mirror, so the sign applies either way.
-        const Vec2f sign = layer_mirror_sign(layer);
-        if (normalized_uv) {
-            inv_u = sign.x();
-            inv_v = sign.y();
-        } else {
-            const Vec2f tile = layer_tile_mm(layer);
-            inv_u = sign.x() / tile.x();
-            inv_v = sign.y() / tile.y();
-        }
+        scale = (normalized_uv || layer.tiling_scale <= 1e-6f) ? 1.f : (1.f / layer.tiling_scale);
         const float rad = layer.rotation_deg * float(M_PI) / 180.f;
         cs     = std::cos(rad);
         sn     = std::sin(rad);
@@ -2196,11 +2183,9 @@ struct UVTransform
     }
     Vec2f operator()(const Vec2f &planar) const
     {
-        // See apply_uv_transform() for why both the tile extents and the aspect correction follow
-        // the rotation rather than preceding it.
-        Vec2f rotated(planar.x() * cs - planar.y() * sn, planar.x() * sn + planar.y() * cs);
-        rotated.x() *= inv_u;
-        rotated.y() *= inv_v;
+        const Vec2f scaled = planar * scale;
+        Vec2f       rotated(scaled.x() * cs - scaled.y() * sn, scaled.x() * sn + scaled.y() * cs);
+        // See apply_uv_transform() for why the aspect correction follows the rotation.
         if (aspect > 0.f && aspect != 1.f)
             rotated.y() *= aspect;
         return rotated + offset;
@@ -2220,43 +2205,26 @@ inline Vec3f triplanar_weights(const Vec3f &normal)
 
 Vec2f apply_uv_transform(const Vec2f &planar, const TextureDisplacementLayer &layer, float aspect, bool normalized_uv)
 {
+    // A uv that is already normalized has no millimetre scale to divide by and no aspect to correct -
+    // see the header. UVTransform does the same, and the two must stay in step.
+    const float scale = (normalized_uv || layer.tiling_scale <= 1e-6f) ? 1.f : (1.f / layer.tiling_scale);
+    const Vec2f scaled = planar * scale;
+
     const float rad = layer.rotation_deg * float(M_PI) / 180.f;
     const float cs  = std::cos(rad);
     const float sn  = std::sin(rad);
-    Vec2f       rotated(planar.x() * cs - planar.y() * sn, planar.x() * sn + planar.y() * cs);
-
-    // The tile extents. Dividing here rather than before the rotation is what makes a non-square
-    // tile a rotated *rectangle* on the surface: scaling the two axes of an already-rotated
-    // coordinate by different amounts would be a shear, so the rotation has to come first and the
-    // extents divide in the tile's own frame. With a square tile the two orders agree exactly,
-    // R being linear - which is why this is not a change in behaviour for any existing layer.
-    //
-    // The mirror rides along as a sign on the same division. It is in the tile's own frame too, so
-    // it mirrors along the tile's axes rather than the world's, and because the offset is what
-    // positions the tile, mirroring about the tile's middle is a matter of the offset the gizmo
-    // writes alongside it, not of anything here.
-    // A uv that is already normalized has no millimetre scale to divide by - see the header - but it
-    // can still be mirrored, so the sign is applied in both cases and only the extents are skipped.
-    // UVTransform does the same, and the two must stay in step.
-    const Vec2f sign = layer_mirror_sign(layer);
-    if (normalized_uv) {
-        rotated.x() *= sign.x();
-        rotated.y() *= sign.y();
-    } else {
-        const Vec2f tile = layer_tile_mm(layer);
-        rotated.x() *= sign.x() / tile.x();
-        rotated.y() *= sign.y() / tile.y();
-    }
+    Vec2f       rotated(scaled.x() * cs - scaled.y() * sn, scaled.x() * sn + scaled.y() * cs);
 
     // Non-square textures. Without this the [0,1] square of uv covers the whole image whatever its
-    // proportions, so a 2:1 image is squeezed into one tile and every feature in it comes out half
-    // as wide as it should be. The tile is `layer_tile_mm().y() * height / width` mm along v, which
-    // is exactly what keeps texels square - so dividing v by that extent is the same as multiplying
-    // it by width / height. A square texture has aspect 1 and is untouched, which is why this
-    // changes nothing for the shipped library.
+    // proportions, so a 2:1 image is squeezed into a square tile and every feature in it comes out
+    // half as wide as it should be. `tiling_scale` is the tile's size along u; the tile is
+    // `tiling_scale * height / width` mm along v, which is exactly what keeps texels square - so
+    // dividing v by that extent is the same as multiplying it by width / height. A square texture has
+    // aspect 1 and is untouched, which is why this changes nothing for the shipped library.
     //
-    // Applied after the rotation, like the extents, and for the same reason: scaling one axis of an
-    // already-rotated coordinate before turning it would be a shear.
+    // Applied after the rotation, not before: scaling one axis of an already-rotated coordinate is a
+    // shear, and doing it the other way round would make "Rotation" skew the pattern instead of
+    // turning it.
     if (!normalized_uv && aspect > 0.f && aspect != 1.f)
         rotated.y() *= aspect;
 
@@ -3600,36 +3568,13 @@ void texture_displacement_patch_frame(const indexed_triangle_set &patch, const s
            (an.y() <= an.x() && an.y() <= an.z()) ? Vec3f::UnitY() : Vec3f::UnitZ();
 }
 
-Matrix3d texture_displacement_frame_rotation(const Transform3d &volume_to_world)
-{
-    // Separated out by polar decomposition rather than by assuming the linear part is a clean
-    // rotation times a scale: a sheared or mirrored placement is legal here. A reflection is left
-    // behind with the scale on purpose, so a mirrored instance still makes the bake frame's
-    // determinant negative and the bake still knows to flip the normals it derives.
-    const Matrix3d                   linear = volume_to_world.linear();
-    Eigen::JacobiSVD<Matrix3d>       svd(linear, Eigen::ComputeFullU | Eigen::ComputeFullV);
-    Matrix3d                         rotation = svd.matrixU() * svd.matrixV().transpose();
-    if (rotation.determinant() < 0.0) {
-        Matrix3d u = svd.matrixU();
-        u.col(2) *= -1.0;
-        rotation = u * svd.matrixV().transpose();
-    }
-    return rotation.transpose(); // world -> texture frame
-}
-
 Transform3d texture_displacement_bake_frame(const Transform3d &volume_to_world)
 {
-    // The model's own orientation, world scale, and the origin moved to where the volume's origin
-    // sits. So the texture rides with the model however it is moved *and* however it is turned,
-    // while a tile is still `tiling_scale` printed millimetres whatever the instance's scale, and a
-    // single untiled stamp starts on the model rather than at the plate's corner.
-    //
-    // Orientation used to come from the world too, which meant turning a part on the plate slid the
-    // decoration across its surface. Turning a part is a print-time decision - clearance, supports,
-    // warping - and it should not be a design one.
-    Transform3d unrotate = Transform3d::Identity();
-    unrotate.linear()    = texture_displacement_frame_rotation(volume_to_world);
-    return unrotate * Eigen::Translation3d(-volume_to_world.translation()) * volume_to_world;
+    // World orientation and scale, but the origin moved to where the volume's own origin sits: the
+    // texture then rides with the model when it is moved about the plate (and a single, untiled stamp
+    // starts on the model rather than at the plate's corner), while a tile is still `tiling_scale`
+    // printed millimetres whatever the instance's scale.
+    return Eigen::Translation3d(-volume_to_world.translation()) * volume_to_world;
 }
 
 Transform3d texture_displacement_volume_to_world(const ModelVolume &volume)
@@ -5840,13 +5785,7 @@ std::string texture_displacement_layers_to_json(const std::vector<TextureDisplac
 
         j["depth_mm"]     = l.depth_mm;
         j["tiling_scale"] = l.tiling_scale;
-        // The v extent and the mirroring. Written unconditionally, including the <= 0 "same as u"
-        // sentinel, so a project says plainly that the tile was square rather than leaving a reader
-        // to infer it from a missing key.
-        j["tiling_scale_v"] = l.tiling_scale_v;
-        j["mirror_u"]       = l.mirror_u;
-        j["mirror_v"]       = l.mirror_v;
-        j["rotation_deg"]   = l.rotation_deg;
+        j["rotation_deg"] = l.rotation_deg;
         j["offset"]       = vec2_to_json(l.offset);
         j["invert"]       = l.invert;
         j["midlevel"]     = l.midlevel;
@@ -5954,12 +5893,6 @@ bool texture_displacement_layers_from_json(const std::string
 
                 read(j, "depth_mm", l.depth_mm);
                 read(j, "tiling_scale", l.tiling_scale);
-                // Absent in a project written before the tile could be rectangular or mirrored, and
-                // then each keeps its default: -1 for the v extent, which layer_tile_mm() reads as
-                // "square", and false for the mirrors. So an older project loads as what it was.
-                read(j, "tiling_scale_v", l.tiling_scale_v);
-                read(j, "mirror_u", l.mirror_u);
-                read(j, "mirror_v", l.mirror_v);
                 read(j, "rotation_deg", l.rotation_deg);
                 if (const auto o = j.find("offset"); o != j.end())
                     l.offset = vec2_from_json(*o, l.offset);

src/libslic3r/TextureDisplacement.hpp

@@ -18,6 +18,7 @@
 
 #include <array>
 
+#include "ChromaStack.hpp"
 #include "Point.hpp"
 #include "TriangleSelector.hpp"
 
@@ -225,17 +226,7 @@ struct TextureDisplacementLayer
     std::shared_ptr<std::vector<unsigned char>> image_data;
 
     float depth_mm     = 0.4f; // maximum displacement along the surface normal, in mm
-    float tiling_scale = 12.5f; // size of one texture tile along u, in mm
-    // Size of one tile along v, in mm. Anything <= 0 means "whatever tiling_scale is", which is both
-    // the default and what a layer created before the tile could be non-square meant - so an older
-    // undo snapshot still comes back as the square tile it was.
-    float tiling_scale_v = -1.f;
-    // Mirror the texture along one axis of the tile. Set by dragging a width or height handle back
-    // through the tile's middle, which is how a side handle behaves in an image editor: the extent
-    // cannot go negative, so the sign lives here instead. Both default to off, which is what every
-    // layer written before this meant.
-    bool  mirror_u = false;
-    bool  mirror_v = false;
+    float tiling_scale = 12.5f; // size of one texture tile, in mm
     float rotation_deg = 0.f;
     Vec2f offset       = Vec2f::Zero();
     bool  invert       = false;
@@ -363,17 +354,33 @@ struct TextureDisplacementLayer
     // for the boundary to be drawn on.
     bool color_enabled = false;
 
+    // ChromaStack (see ChromaStack.hpp): when set, `image_data` is not a texture the user picked but the
+    // height map chroma_solve() computed, so that the whole displacement pipeline - preview, UV
+    // editor, bake - goes on reading one grayscale height map and needs to know nothing about colour
+    // stacks. The picture it was solved from is kept here instead, because the solve has to be redone
+    // whenever the stack, the print's layer height or the picture itself changes.
+    //
+    // A ChromaStack layer gets its colour from filament swaps at fixed Z, which is what `chroma`
+    // records and what the object's height ranges carry out. That is a different mechanism from
+    // `color_enabled` (per-triangle filament assignment), and the two are mutually exclusive: the
+    // panel turns colour off on a ChromaStack layer, since painting the relief in one filament per
+    // triangle would hide the very stack the heights were solved against.
+    bool                                        chroma_enabled = false;
+    ChromaStack                               chroma;
+    std::shared_ptr<std::vector<unsigned char>> chroma_source;
+
     bool empty() const { return !image_data || image_data->empty(); }
 
     template<class Archive> void save(Archive &ar) const
     {
         std::string blob = image_data ? std::string(image_data->begin(), image_data->end()) : std::string();
+        std::string source = chroma_source ? std::string(chroma_source->begin(), chroma_source->end()) : std::string();
         ar(slot, name, path, path_in_3mf, blob, depth_mm, tiling_scale, rotation_deg, offset, invert, tile_enabled,
            static_cast<int>(tile_method), static_cast<int>(projection_method), lscm_seam_angle_deg, islands,
            static_cast<int>(blend_mode), midlevel, island_padding_mm, lscm_seam_edges, view_project_right,
            view_project_up, smoothing, edge_smoothing, edge_smoothing_amount, auto_connect_islands, island_groups,
-           lscm_uv_overrides, view_project_projective, view_project_matrix, color_enabled, tiling_scale_v,
-           mirror_u, mirror_v);
+           lscm_uv_overrides, view_project_projective, view_project_matrix, color_enabled, chroma_enabled, chroma,
+           source);
     }
     template<class Archive> void load(Archive &ar)
     {
@@ -381,12 +388,14 @@ struct TextureDisplacementLayer
         int         tile_method_int       = 0;
         int         projection_method_int = 0;
         int         blend_mode_int        = 0;
+        std::string source;
         ar(slot, name, path, path_in_3mf, blob, depth_mm, tiling_scale, rotation_deg, offset, invert, tile_enabled,
            tile_method_int, projection_method_int, lscm_seam_angle_deg, islands, blend_mode_int, midlevel,
            island_padding_mm, lscm_seam_edges, view_project_right, view_project_up, smoothing, edge_smoothing,
            edge_smoothing_amount, auto_connect_islands, island_groups, lscm_uv_overrides, view_project_projective,
-           view_project_matrix, color_enabled, tiling_scale_v, mirror_u, mirror_v);
+           view_project_matrix, color_enabled, chroma_enabled, chroma, source);
         image_data        = blob.empty() ? nullptr : std::make_shared<std::vector<unsigned char>>(blob.begin(), blob.end());
+        chroma_source   = source.empty() ? nullptr : std::make_shared<std::vector<unsigned char>>(source.begin(), source.end());
         tile_method       = static_cast<TextureTileMethod>(tile_method_int);
         projection_method = static_cast<TextureProjectionMethod>(projection_method_int);
         blend_mode        = static_cast<TextureBlendMode>(blend_mode_int);
@@ -394,25 +403,6 @@ struct TextureDisplacementLayer
 };
 
 
-// The layer's tile extents in mm, u first. The v extent falls back to the u one whenever
-// tiling_scale_v is unset, so "unset means square" is written down once and every consumer - the
-// bake, the preview shaders, the UV pane and the placement gizmo - agrees on what a layer with no
-// v extent means.
-inline Vec2f layer_tile_mm(const TextureDisplacementLayer &layer)
-{
-    const float su = (layer.tiling_scale > 1e-6f) ? layer.tiling_scale : 1.f;
-    return Vec2f(su, (layer.tiling_scale_v > 1e-6f) ? layer.tiling_scale_v : su);
-}
-
-// +1 or -1 per axis, for the layer's mirroring. A separate factor from the extents rather than a
-// signed extent: the extents are a size the panel shows and the refinement measures texels against,
-// and letting one of them go negative would mean every one of those places had to remember to take
-// the magnitude.
-inline Vec2f layer_mirror_sign(const TextureDisplacementLayer &layer)
-{
-    return Vec2f(layer.mirror_u ? -1.f : 1.f, layer.mirror_v ? -1.f : 1.f);
-}
-
 // Settings that apply to the whole layer stack rather than to one layer, held per ModelVolume next
 // to texture_displacement_layers and consumed by build_texture_displacement().
 struct TextureDisplacementOptions
@@ -629,8 +619,9 @@ DecodedHeightTexture downscale_height_texture(const DecodedHeightTexture &src, i
 // colour PNGs are accepted; for a colour source the height is its luminance, using the same
 // coefficients wxImage::ConvertToGreyscale() uses, so a texture imported as colour displaces exactly
 // as it did when the importer flattened it to grey on the way in. Returns an empty
-// DecodedHeightTexture if image_data is empty or is not a PNG at all.
-DecodedHeightTexture decode_height_texture(const TextureDisplacementLayer &layer);
+// DecodedHeightTexture if image_data is empty or is not a PNG at all. keep_decoded = false leaves the
+// result out of the decode cache, for one-off reads such as a picker thumbnail that must not hold full-size pixels.
+DecodedHeightTexture decode_height_texture(const TextureDisplacementLayer &layer, bool keep_decoded = true);
 
 // decode_height_texture(layer).has_color(), answered from the decode cache rather than from a copy of the
 // texture - cheap enough to ask every frame. Smoothing does not change it, so the raw decode is what is read.
@@ -698,19 +689,14 @@ Vec2f project_planar(const Vec3f &position, const Vec3f &normal);
 // projection method, without going through project_texture_displacement_uv()'s own dispatch
 // (which only knows how to compute the *analytic* methods from a single vertex + normal).
 // `aspect` is the height map's width / height. It scales the v axis so a non-square image is not
-// squeezed into a square tile: the tile is `layer_tile_mm().y() * height / width` mm along v, which
-// keeps texels square. 1 (the default) is the square case and leaves the coordinate exactly as it
-// always was.
-//
-// The tile extents divide *after* the rotation, not before. With one extent that made no difference
-// (R is linear, so R(p / s) == R(p) / s), but with two it is what keeps a non-square tile a rotated
-// rectangle on the surface instead of a sheared parallelogram. The layer's mirroring is a sign on
-// that same division.
+// squeezed into a square tile: `tiling_scale` is the tile's size along u, and the tile is
+// `tiling_scale * height / width` mm along v, which keeps texels square. 1 (the default) is the
+// square case and leaves the coordinate exactly as it always was.
 //
 // `normalized_uv` says the coordinate is already a uv rather than a millimetre measurement on the
-// model - which is what TextureProjectionMethod::ModelUV hands in. Then neither the tile extents (mm
+// model - which is what TextureProjectionMethod::ModelUV hands in. Then neither the tiling scale (mm
 // per tile is meaningless for it) nor the aspect correction (the artist's layout already accounts for
-// the image's proportions) applies, and only the rotation, the mirroring and the offset are left.
+// the image's proportions) applies, and only the rotation and offset are left.
 Vec2f apply_uv_transform(const Vec2f &planar, const TextureDisplacementLayer &layer, float aspect = 1.f,
                          bool normalized_uv = false);
 
@@ -1082,10 +1068,6 @@ indexed_triangle_set build_texture_displacement(const indexed_triangle_set
 Transform3d texture_displacement_volume_to_world(const ModelVolume &volume);
 // The frame the bake and the previews project the texture in: `volume_to_world` with its translation
 // removed, i.e. world orientation and scale about the volume's own origin. See build_texture_displacement().
-// The rotation that carries a world direction into the texture's frame, i.e. the inverse of the
-// placement's own rotation. Exposed because the preview shaders rebuild the frame themselves from a
-// world position and have to turn it the same way the bake does, or the two disagree.
-Matrix3d texture_displacement_frame_rotation(const Transform3d &volume_to_world);
 Transform3d texture_displacement_bake_frame(const Transform3d &volume_to_world);
 
 // Area-weighted vertex normals of `its` - the directions the bake both projects and displaces along.

src/libslic3r/Utils.hpp

@@ -842,6 +842,13 @@ void load_string_file(const boost::filesystem::path& p, std::string& str);
 
 bool check_layer_id_pattern(const std::string& pattern, int layer_id);
 
+// On Windows, return an equivalent path carrying the extended-length ("\\?\" / "\\?\UNC\")
+// prefix so the Win32 APIs accept it past MAX_PATH even when LongPathsEnabled is off. The path
+// is made absolute and lexically normalized first, because the prefix disables normalisation.
+// A path that is already extended (or a device path) is returned unchanged; on other platforms,
+// or when an absolute path cannot be formed, the input is returned unchanged.
+boost::filesystem::path extended_path(const boost::filesystem::path& path);
+
 } // namespace Slic3r
 
 #if WIN32

src/libslic3r/miniz_extension.cpp

@@ -17,6 +17,7 @@
 
 #if defined(_MSC_VER) || defined(__MINGW64__)
 #include "boost/nowide/cstdio.hpp"
+#include "boost/nowide/convert.hpp"
 #endif
 
 #include "I18N.hpp"
@@ -62,7 +63,11 @@ bool open_zip(mz_zip_archive *zip, const char *fname, bool isread)
 
     FILE *f = nullptr;
 #if defined(_MSC_VER) || defined(__MINGW64__)
-    f = boost::nowide::fopen(fname, mode);
+    // fname is UTF-8 (the nowide contract). Route it through the shared extended-path helper so an
+    // archive stored past MAX_PATH opens even with LongPathsEnabled off.
+    const std::wstring wide_path = boost::nowide::widen(fname);
+    const std::string  extended  = boost::nowide::narrow(Slic3r::extended_path(boost::filesystem::path(wide_path)).wstring());
+    f = boost::nowide::fopen(extended.c_str(), mode);
 #elif defined(__GNUC__) && defined(_LARGEFILE64_SOURCE)
     f = fopen64(fname, mode);
 #else
@@ -101,6 +106,14 @@ bool close_zip(mz_zip_archive *zip, bool isread)
     }
     return ret;
 }
+
+#ifdef WIN32
+// mz_file_write_func writing an extracted entry into an already-open FILE*.
+size_t mz_write_to_cfile(void *opaque, mz_uint64 /*file_ofs*/, const void *buffer, size_t size)
+{
+    return std::fwrite(buffer, 1, size, static_cast<FILE *>(opaque));
+}
+#endif
 }
 
 bool open_zip_reader(mz_zip_archive *zip, const std::string &fname)
@@ -116,6 +129,36 @@ bool open_zip_writer(mz_zip_archive *zip, const std::string &fname)
 bool close_zip_reader(mz_zip_archive *zip) { return close_zip(zip, true); }
 bool close_zip_writer(mz_zip_archive *zip) { return close_zip(zip, false); }
 
+bool extract_zip_entry_to_file(mz_zip_archive *zip, mz_uint file_index, const boost::filesystem::path &dest_path)
+{
+    if (!zip)
+        return false;
+#ifdef WIN32
+    // Open the destination ourselves with a Windows extended-length path: miniz's narrow
+    // extract_to_file cannot be relied on past MAX_PATH with LongPathsEnabled off, and its wide
+    // variant takes a plain (unprefixed) path and a text-mode stream. This keeps the write binary.
+    FILE *file = ::_wfopen(Slic3r::extended_path(dest_path).wstring().c_str(), L"wb");
+    if (!file) {
+        zip->m_last_error = MZ_ZIP_FILE_OPEN_FAILED;
+        return false;
+    }
+
+    const bool written = mz_zip_reader_extract_to_callback(zip, file_index, mz_write_to_cfile, file, 0) != 0;
+    const bool closed  = std::fclose(file) == 0;
+    if (!written || !closed) {
+        if (written && !closed)
+            zip->m_last_error = MZ_ZIP_FILE_CLOSE_FAILED;
+        // Drop the partial entry rather than leave a truncated file behind.
+        boost::system::error_code ec;
+        boost::filesystem::remove(Slic3r::extended_path(dest_path), ec);
+        return false;
+    }
+    return true;
+#else
+    return mz_zip_reader_extract_to_file(zip, file_index, dest_path.string().c_str(), 0) != 0;
+#endif
+}
+
 std::string decode_archive_entry_path(mz_zip_archive *zip, const mz_zip_archive_file_stat &stat)
 {
     if (stat.m_is_utf8)
@@ -156,10 +199,13 @@ bool extract_archive_confined(const std::string &zip_path_utf8, const std::strin
         }
         const std::string dest_file = dest_dir + "/" + stat.m_filename;
         try {
+            // Use the extended-length form for every filesystem call so entries under a destination
+            // past MAX_PATH are created even with LongPathsEnabled off.
+            const boost::filesystem::path dest_path(dest_file);
+            const boost::filesystem::path io_dest = extended_path(dest_path);
             if (stat.m_is_directory) {
-                const boost::filesystem::path dest_path(dest_file);
-                if (!boost::filesystem::exists(dest_path))
-                    boost::filesystem::create_directories(dest_path);
+                if (!boost::filesystem::exists(io_dest))
+                    boost::filesystem::create_directories(io_dest);
                 continue;
             }
             if (stat.m_uncomp_size == 0) {
@@ -167,10 +213,9 @@ bool extract_archive_confined(const std::string &zip_path_utf8, const std::strin
                 continue;
             }
             // Replace a symlink at the destination rather than writing through it.
-            const boost::filesystem::path dest_path(dest_file);
-            if (boost::filesystem::is_symlink(boost::filesystem::symlink_status(dest_path)))
-                boost::filesystem::remove(dest_path);
-            if (!mz_zip_reader_extract_to_file(&archive, stat.m_file_index, dest_file.c_str(), 0)) {
+            if (boost::filesystem::is_symlink(boost::filesystem::symlink_status(io_dest)))
+                boost::filesystem::remove(io_dest);
+            if (!extract_zip_entry_to_file(&archive, stat.m_file_index, dest_path)) {
                 BOOST_LOG_TRIVIAL(error) << "Unzip: extract file " << stat.m_filename << " to dest " << dest_file << " failed";
                 close_zip_reader(&archive);
                 return false;

src/libslic3r/miniz_extension.hpp

@@ -1,6 +1,7 @@
 #ifndef MINIZ_EXTENSION_HPP
 #define MINIZ_EXTENSION_HPP
 
+#include <boost/filesystem/path.hpp>
 #include <string>
 #include <miniz.h>
 
@@ -13,6 +14,10 @@ bool close_zip_writer(mz_zip_archive *zip);
 std::string decode_archive_entry_path(mz_zip_archive *zip, const mz_zip_archive_file_stat &stat);
 // Extracts every entry of the archive under dest_dir. Nothing is written if any entry would resolve outside dest_dir.
 bool extract_archive_confined(const std::string &zip_path_utf8, const std::string &dest_dir);
+// Extracts one entry into dest_path, binary-safe and overwriting. On Windows the destination is
+// opened through an extended-length path so it works past MAX_PATH even with LongPathsEnabled off;
+// a partial file is removed if the write fails. False for a directory/unsupported entry or on error.
+bool extract_zip_entry_to_file(mz_zip_archive *zip, mz_uint file_index, const boost::filesystem::path &dest_path);
 
 class MZ_Archive {
 public:

src/libslic3r/utils.cpp

@@ -2254,5 +2254,45 @@ bool check_layer_id_pattern(const std::string& pattern, int layer_id){
     return mod >= 0 && mod < count;
 }
 
+boost::filesystem::path extended_path(const boost::filesystem::path& path) {
+#ifdef WIN32
+    namespace fs = boost::filesystem;
+
+    // Already extended, or an explicit device path (which is not subject to MAX_PATH and must not
+    // be absolutized or normalized): leave it alone.
+    {
+        fs::path prefix_candidate = path;
+        const std::wstring native = prefix_candidate.make_preferred().wstring();
+        if (native.rfind(L"\\\\?\\", 0) == 0 || native.rfind(L"\\\\.\\", 0) == 0)
+            return fs::path(native);
+    }
+
+    fs::path absolute = path;
+    if (!absolute.is_absolute()) {
+        boost::system::error_code ec;
+        absolute = fs::absolute(absolute, ec);
+        if (ec)
+            return path; // can't form a valid prefixed path from a relative one
+    }
+
+    // A trailing separator is redundant to the Win32 APIs, but lexically_normal() would rewrite it
+    // as a "." element, which the "\\?\" prefix (normalisation off) takes literally. Drop it first.
+    while (absolute.has_relative_path() && absolute.filename() == ".")
+        absolute = absolute.parent_path();
+
+    // The "\\?\" prefix turns off Win32 normalisation, so do it here first.
+    absolute = absolute.lexically_normal();
+
+    // Extended paths only recognize backslash separators.
+    const std::wstring native = absolute.make_preferred().wstring();
+    if (native.empty())
+        return path;
+    if (native.rfind(L"\\\\", 0) == 0)
+        return fs::path(L"\\\\?\\UNC\\" + native.substr(2)); // \\server\share\... -> \\?\UNC\server\share\...
+    return fs::path(L"\\\\?\\" + native);                    // C:\dir\... -> \\?\C:\dir\...
+#else
+    return path;
+#endif
+}
 
 }; // namespace Slic3r

src/slic3r/CMakeLists.txt

@@ -222,6 +222,7 @@ set(SLIC3R_GUI_SOURCES
     GUI/Gizmos/GLGizmoUtils.hpp
     GUI/GLCanvas3D.cpp
     GUI/GLCanvas3D.hpp
+    GUI/SequentialPrintClearance.hpp
     GUI/GLModel.cpp
     GUI/GLModel.hpp
     GUI/GLSelectionRectangle.cpp
@@ -689,6 +690,8 @@ set(SLIC3R_GUI_SOURCES
     plugin/host/PluginHostUi.hpp
     plugin/host/PluginPages.cpp
     plugin/host/PluginPages.hpp
+    plugin/host/PluginResourceTextures.cpp
+    plugin/host/PluginResourceTextures.hpp
     plugin/CloudPluginService.cpp
     plugin/CloudPluginService.hpp
     plugin/PluginFsUtils.cpp
@@ -712,6 +715,9 @@ set(SLIC3R_GUI_SOURCES
     plugin/pluginTypes/pages/PagesPluginCapability.hpp
     plugin/pluginTypes/pages/PagesPluginCapability.cpp
     plugin/pluginTypes/pages/PagesPluginCapabilityTrampoline.hpp
+    plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapability.hpp
+    plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapability.cpp
+    plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapabilityTrampoline.hpp
     plugin/pluginTypes/script/ScriptPluginCapability.hpp
     plugin/pluginTypes/script/ScriptPluginCapability.cpp
     plugin/pluginTypes/script/ScriptPluginCapabilityTrampoline.hpp

src/slic3r/GUI/AMSMaterialsSetting.cpp

@@ -85,6 +85,96 @@ static std::string float_to_string_with_precision(float value, int precision = 3
     return stream.str();
 }
 
+static std::string colour_to_ams_string(const wxColour& color)
+{
+    char col_buf[10];
+    sprintf(col_buf, "%02X%02X%02X%02X", (int)color.Red(), (int)color.Green(), (int)color.Blue(), (int)color.Alpha());
+    return col_buf;
+}
+
+static wxColour mix_colour(const wxColour& left, const wxColour& right, double ratio)
+{
+    ratio = std::max(0.0, std::min(1.0, ratio));
+    auto mix_channel = [ratio](unsigned char l, unsigned char r) {
+        return static_cast<unsigned char>(std::round(l + (r - l) * ratio));
+    };
+    return wxColour(mix_channel(left.Red(), right.Red()),
+                    mix_channel(left.Green(), right.Green()),
+                    mix_channel(left.Blue(), right.Blue()),
+                    mix_channel(left.Alpha(), right.Alpha()));
+}
+
+static std::vector<ColorPickerPopup::ColorItem> collect_ams_color_items(DevFilaSystem* fila_system)
+{
+    std::vector<ColorPickerPopup::ColorItem> items;
+    if (!fila_system) return items;
+
+    for (const auto& ams_pair : fila_system->GetAmsList()) {
+        DevAms* ams = ams_pair.second;
+        if (!ams) continue;
+
+        for (const auto& tray_pair : ams->GetTrays()) {
+            DevAmsTray* tray = tray_pair.second;
+            if (!tray || !tray->is_tray_info_ready()) continue;
+
+            ColorPickerPopup::ColorItem item;
+            for (const std::string& col : tray->cols) {
+                item.colors.emplace_back(DevAmsTray::decode_color(col));
+            }
+            if (item.colors.empty() && !tray->color.empty()) {
+                item.colors.emplace_back(DevAmsTray::decode_color(tray->color));
+            }
+            if (item.colors.empty()) continue;
+
+            item.ctype = item.colors.size() > 1 ? tray->ctype : 2;
+            items.emplace_back(std::move(item));
+        }
+    }
+
+    return items;
+}
+
+static std::vector<ColorPickerPopup::ColorItem> collect_preset_color_items(const std::string& filament_id)
+{
+    std::vector<ColorPickerPopup::ColorItem> items;
+    if (filament_id.empty()) return items;
+
+    auto* clr_query = GUI::wxGetApp().get_filament_color_code_query();
+    if (!clr_query) return items;
+
+    auto* agent = wxGetApp().getAgent();
+    const std::string catalog_id = agent ? agent->from_orca_filament_id(filament_id) : filament_id;
+    FilamentColorCodes* color_codes = clr_query->GetFilaInfoMap(wxString::FromUTF8(catalog_id));
+    if (!color_codes || !color_codes->GetFilamentColor2CodeMap()) return items;
+
+    std::set<std::string> seen;
+    for (const auto& color_pair : *color_codes->GetFilamentColor2CodeMap()) {
+        const FilamentColor& fila_color = color_pair.first;
+        FilamentColorCode* color_code = color_pair.second;
+        if (!color_code || fila_color.GetColors().empty()) continue;
+
+        ColorPickerPopup::ColorItem item;
+        item.ctype = fila_color.ColorCount() < 2 ? 2 :
+            (fila_color.m_color_type == FilamentColor::ColorType::GRADIENT_CLR ? 0 : 1);
+        item.name = color_code->GetFilaColorName();
+
+        std::string key = std::to_string(item.ctype);
+        for (const wxColour& color : fila_color.GetColors()) {
+            item.colors.emplace_back(color);
+            char rgba[9];
+            std::snprintf(rgba, sizeof(rgba), "%02X%02X%02X%02X",
+                static_cast<unsigned>(color.Red()), static_cast<unsigned>(color.Green()),
+                static_cast<unsigned>(color.Blue()), static_cast<unsigned>(color.Alpha()));
+            key += "|" + std::string(rgba);
+        }
+
+        if (item.colors.empty() || !seen.insert(key).second) continue;
+        items.emplace_back(std::move(item));
+    }
+
+    return items;
+}
+
 AMSMaterialsSetting::AMSMaterialsSetting(wxWindow *parent, wxWindowID id)
     : DPIDialog(parent, id, _L("AMS Materials Setting"), wxDefaultPosition, wxDefaultSize, wxCAPTION | wxCLOSE_BOX)
     , m_color_picker_popup(ColorPickerPopup(this))
@@ -740,8 +830,19 @@ void AMSMaterialsSetting::on_select_ok(wxCommandEvent &event)
     nozzle_temp_min.ToLong(&nozzle_temp_min_int);
     nozzle_temp_max.ToLong(&nozzle_temp_max_int);
     wxColour color = m_clr_picker->m_colour;
-    char col_buf[10];
-    sprintf(col_buf, "%02X%02X%02X%02X", (int)color.Red(), (int)color.Green(), (int)color.Blue(), (int)color.Alpha());
+    std::string tray_color = colour_to_ams_string(color);
+    std::vector<std::string> tray_colors;
+    int tray_ctype = 2;
+    // Only firmware advertising fun2[23] receives the manual multi-color fields.
+    if (obj->is_support_filament_manual_multi_color) {
+        for (const wxColour& selected_color : m_clr_picker->m_cols) {
+            tray_colors.emplace_back(colour_to_ams_string(selected_color));
+        }
+        if (tray_colors.empty()) {
+            tray_colors.emplace_back(tray_color);
+        }
+        tray_ctype = tray_colors.size() > 1 ? m_clr_picker->ctype : 2;
+    }
 
     if (ams_filament_id.empty() || nozzle_temp_min.empty() || nozzle_temp_max.empty() || m_filament_type.empty()) {
         BOOST_LOG_TRIVIAL(trace) << "Invalid Setting id";
@@ -760,13 +861,33 @@ void AMSMaterialsSetting::on_select_ok(wxCommandEvent &event)
                             << ", tray_info_idx (filament_id) = " << ams_filament_id
                             << ", setting_id = " << ams_setting_id
                             << ", tray_type = " << m_filament_type
-                            << ", tray_color = " << col_buf
+                            << ", tray_color = " << tray_color
                             << ", nozzle_temp_min = " << nozzle_temp_min_int
                             << ", nozzle_temp_max = " << nozzle_temp_max_int;
 
     // set filament
     if (m_is_third) {
-        obj->command_ams_filament_settings(ams_id, slot_id, ams_filament_id, ams_setting_id, std::string(col_buf), m_filament_type, nozzle_temp_min_int, nozzle_temp_max_int);
+        obj->command_ams_filament_settings(ams_id, slot_id, ams_filament_id, ams_setting_id, tray_color, m_filament_type, nozzle_temp_min_int, nozzle_temp_max_int,
+                                           tray_colors, tray_ctype);
+
+        // Refresh the edited tray immediately while its MQTT acknowledgement is pending.
+        DevAmsTray* tray = nullptr;
+        if (is_virtual_tray()) {
+            for (auto& virtual_tray : obj->vt_slot) {
+                if (virtual_tray.id == std::to_string(ams_id)) {
+                    tray = &virtual_tray;
+                    break;
+                }
+            }
+        } else {
+            tray = obj->GetFilaSystem()->GetAmsTray(std::to_string(ams_id), std::to_string(slot_id));
+        }
+        if (tray) {
+            tray->color = tray_color;
+            tray->cols = tray_colors;
+            tray->ctype = tray_ctype;
+            tray->set_hold_count();
+        }
     }
 
     //reset param
@@ -923,8 +1044,15 @@ void AMSMaterialsSetting::set_ctype(int ctype)
 
 void AMSMaterialsSetting::on_picker_color(wxCommandEvent& event)
 {
-    unsigned int color_num  = event.GetInt();
-    set_color(wxColour(color_num>>24&0xFF, color_num>>16&0xFF, color_num>>8&0xFF, color_num&0xFF));
+    std::vector<wxColour> colors = m_color_picker_popup.get_selected_colours();
+    if (colors.empty()) {
+        unsigned int color_num = event.GetInt();
+        colors.emplace_back(color_num >> 24 & 0xFF, color_num >> 16 & 0xFF, color_num >> 8 & 0xFF, color_num & 0xFF);
+    }
+
+    set_ctype(m_color_picker_popup.get_selected_ctype());
+    set_color(colors.front());
+    set_colors(colors);
 }
 
 void AMSMaterialsSetting::on_clr_picker(wxMouseEvent &event)
@@ -939,14 +1067,12 @@ void AMSMaterialsSetting::on_clr_picker(wxMouseEvent &event)
         }
     }
 
-    std::vector<wxColour> ams_colors;
-    obj->GetFilaSystem()->CollectAmsColors(ams_colors);
-
+    m_color_picker_popup.set_ams_colours(collect_ams_color_items(obj->GetFilaSystem().get()));
+    m_color_picker_popup.set_preset_colours(collect_preset_color_items(ams_filament_id));
+    m_color_picker_popup.set_def_colour(m_clr_picker->m_colour, m_clr_picker->m_cols, m_clr_picker->ctype);
     wxPoint img_pos = m_clr_picker->ClientToScreen(wxPoint(0, 0));
     wxPoint popup_pos(img_pos.x - m_color_picker_popup.GetSize().x - FromDIP(95), img_pos.y - FromDIP(65));
     m_color_picker_popup.Position(popup_pos, wxSize(0, 0));
-    m_color_picker_popup.set_ams_colours(ams_colors);
-    m_color_picker_popup.set_def_colour(m_clr_picker->m_colour);
     m_color_picker_popup.Popup();
 }
 
@@ -1735,7 +1861,46 @@ void ColorPicker::doRender(wxDC& dc)
     auto radius = m_show_full ? size.x / 2 - FromDIP(1) : size.x / 2;
     if (m_selected) radius -= FromDIP(1);
 
-    if (alpha == 0) {
+    if (m_show_full && m_cols.size() > 1) {
+        if (ctype == 0) {
+            const double center_x = size.x / 2.0;
+            const double center_y = size.y / 2.0;
+            const double draw_radius = radius;
+            dc.SetPen(*wxTRANSPARENT_PEN);
+            for (int x = 0; x < size.x; ++x) {
+                const double dx = x + 0.5 - center_x;
+                if (std::abs(dx) > draw_radius) continue;
+
+                const double half_height = std::sqrt(std::max(0.0, draw_radius * draw_radius - dx * dx));
+                const int top = std::max(0, static_cast<int>(std::ceil(center_y - half_height)));
+                const int bottom = std::min(size.y - 1, static_cast<int>(std::floor(center_y + half_height)));
+                const double pos = size.x > 1 ? static_cast<double>(x) / (size.x - 1) : 0.0;
+                const double scaled = pos * (m_cols.size() - 1);
+                const size_t idx = std::min(static_cast<size_t>(scaled), m_cols.size() - 2);
+                const double ratio = scaled - idx;
+
+                if (top <= bottom) {
+                    dc.SetBrush(wxBrush(mix_colour(m_cols[idx], m_cols[idx + 1], ratio)));
+                    dc.DrawRectangle(x, top, 1, bottom - top + 1);
+                }
+            }
+        }
+        else {
+            float ev_angle = 360.0 / m_cols.size();
+            const float overlap = 2.0f;
+            wxPoint center(size.x / 2, size.y / 2);
+            dc.SetPen(*wxTRANSPARENT_PEN);
+            dc.SetBrush(wxBrush(m_cols.front()));
+            dc.DrawCircle(center.x, center.y, radius);
+            for (int i = 0; i < m_cols.size(); i++) {
+                dc.SetBrush(m_cols[i]);
+                float startAngle = 270.0f + i * ev_angle;
+                float endAngle = startAngle + ev_angle;
+                dc.DrawEllipticArc(center.x - radius, center.y - radius, 2 * radius, 2 * radius, startAngle - overlap, endAngle + overlap);
+            }
+        }
+    }
+    else if (alpha == 0) {
         wxSize bmp_size = m_bitmap_transparent_def.GetBmpSize();
         int center_x = (size.x - bmp_size.x) / 2;
         int center_y = (size.y - bmp_size.y) / 2;
@@ -1777,52 +1942,6 @@ void ColorPicker::doRender(wxDC& dc)
         dc.SetPen(wxPen(wxColour("#6B6B6B")));
         dc.SetBrush(*wxTRANSPARENT_BRUSH);
         dc.DrawCircle(size.x / 2, size.y / 2, radius);
-
-        if (m_cols.size() > 1) {
-            if (ctype == 0) {
-                int left = FromDIP(0);
-                float total_width = size.x;
-                int gwidth = std::round(total_width / (m_cols.size() - 1));
-
-                for (int i = 0; i < m_cols.size() - 1; i++) {
-
-                    if ((left + gwidth) > (size.x)) {
-                        gwidth = size.x - left;
-                    }
-
-                    auto rect = wxRect(left, 0, gwidth, size.y);
-                    dc.GradientFillLinear(rect, m_cols[i], m_cols[i + 1], wxEAST);
-                    left += gwidth;
-                }
-                if (wxGetApp().dark_mode()) {
-                    dc.DrawBitmap(m_bitmap_border_dark, wxPoint(0, 0));
-                }
-                else {
-                    dc.DrawBitmap(m_bitmap_border, wxPoint(0, 0));
-                }
-            }
-            else {
-                float ev_angle = 360.0 / m_cols.size();
-                float startAngle = 270.0;
-                float endAngle = 270.0;
-                dc.SetPen(*wxTRANSPARENT_PEN);
-                for (int i = 0; i < m_cols.size(); i++) {
-                    dc.SetBrush(m_cols[i]);
-                    endAngle += ev_angle;
-                    endAngle = endAngle > 360.0 ? endAngle - 360.0 : endAngle;
-                    wxPoint center(size.x / 2, size.y / 2);
-                    dc.DrawEllipticArc(center.x - radius, center.y - radius, 2 * radius, 2 * radius, startAngle, endAngle);
-                    startAngle += ev_angle;
-                    startAngle = startAngle > 360.0 ? startAngle - 360.0 : startAngle;
-                }
-                if (wxGetApp().dark_mode()) {
-                    dc.DrawBitmap(m_bitmap_border_dark, wxPoint(0, 0));
-                }
-                else {
-                    dc.DrawBitmap(m_bitmap_border, wxPoint(0, 0));
-                }
-            }
-        }
     }
 
     if (m_is_empty) {
@@ -1891,19 +2010,21 @@ ColorPickerPopup::ColorPickerPopup(wxWindow* parent)
     m_ams_fg_sizer->SetNonFlexibleGrowMode(wxFLEX_GROWMODE_SPECIFIED);
 
     //other
-    wxFlexGridSizer* fg_sizer;
-    fg_sizer = new wxFlexGridSizer(0, 8, 0, 0);
-    fg_sizer->SetFlexibleDirection(wxBOTH);
-    fg_sizer->SetNonFlexibleGrowMode(wxFLEX_GROWMODE_SPECIFIED);
+    m_other_fg_sizer = new wxFlexGridSizer(0, 8, 0, 0);
+    m_other_fg_sizer->SetFlexibleDirection(wxBOTH);
+    m_other_fg_sizer->SetNonFlexibleGrowMode(wxFLEX_GROWMODE_SPECIFIED);
 
 
     for (wxColour col : m_def_colors) {
         auto cp = new ColorPicker(m_def_color_box, wxID_ANY, wxDefaultPosition, wxDefaultSize);
         cp->set_color(col);
+        cp->set_colors({col});
+        cp->ctype = 2;
         cp->set_selected(false);
         cp->SetBackgroundColour(StateColor::darkModeColorFor(wxColour(238,238,238)));
         m_color_pickers.push_back(cp);
-        fg_sizer->Add(cp, 0, wxALL, FromDIP(3));
+        m_default_color_pickers.push_back(cp);
+        m_other_fg_sizer->Add(cp, 0, wxALL, FromDIP(3));
         cp->Bind(wxEVT_LEFT_DOWN, [this, cp](auto& e) {
             set_def_colour(cp->m_colour);
 
@@ -1970,7 +2091,7 @@ ColorPickerPopup::ColorPickerPopup(wxWindow* parent)
     m_sizer_box->Add(m_sizer_ams, 1, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(10));
     m_sizer_box->Add(m_ams_fg_sizer, 0, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(10));
     m_sizer_box->Add(m_sizer_other, 1, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(10));
-    m_sizer_box->Add(fg_sizer, 0, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(10));
+    m_sizer_box->Add(m_other_fg_sizer, 0, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(10));
     m_sizer_box->Add(m_sizer_custom, 0, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(10));
     m_sizer_box->Add(m_custom_cp, 0, wxEXPAND|wxLEFT|wxRIGHT, FromDIP(16));
     m_sizer_box->Add(0, 0, 0, wxTOP, FromDIP(10));
@@ -1981,7 +2102,7 @@ ColorPickerPopup::ColorPickerPopup(wxWindow* parent)
     m_def_color_box->Fit();
 
     m_sizer_main->Add(m_def_color_box, 0, wxALL | wxEXPAND, 10);
-    SetSizer(m_sizer_main);
+    SetSizerAndFit(m_sizer_main);
     Layout();
     Fit();
 
@@ -2019,13 +2140,15 @@ void ColorPickerPopup::on_custom_clr_picker(wxMouseEvent& /*event*/)
     wxPostEvent(GetParent(), evt);
 }
 
-void ColorPickerPopup::set_ams_colours(std::vector<wxColour> ams)
+void ColorPickerPopup::set_ams_colours(const std::vector<ColorItem>& ams)
 {
+    m_ams_fg_sizer->Clear(false);
     if (m_ams_color_pickers.size() > 0) {
         for (ColorPicker* col_pick:m_ams_color_pickers) {
 
             std::vector<ColorPicker*>::iterator iter = find(m_color_pickers.begin(), m_color_pickers.end(), col_pick);
             if (iter != m_color_pickers.end()) {
+                col_pick->Hide();
                 col_pick->Destroy();
                 m_color_pickers.erase(iter);
             }
@@ -2035,17 +2158,22 @@ void ColorPickerPopup::set_ams_colours(std::vector<wxColour> ams)
     }
 
 
-    m_ams_colors = ams;
-    for (wxColour col : m_ams_colors) {
+    m_ams_color_items = ams;
+    for (const ColorItem& item : m_ams_color_items) {
+        if (item.colors.empty()) continue;
+
         auto cp = new ColorPicker(m_def_color_box, wxID_ANY, wxDefaultPosition, wxDefaultSize);
-        cp->set_color(col);
+        cp->set_color(item.colors.front());
+        cp->set_colors(item.colors);
+        cp->ctype = item.colors.size() > 1 ? item.ctype : 2;
+        cp->set_show_full(true);
         cp->set_selected(false);
         cp->SetBackgroundColour(StateColor::darkModeColorFor(wxColour(238,238,238)));
         m_color_pickers.push_back(cp);
         m_ams_color_pickers.push_back(cp);
         m_ams_fg_sizer->Add(cp, 0, wxALL, FromDIP(3));
         cp->Bind(wxEVT_LEFT_DOWN, [this, cp](auto& e) {
-            set_def_colour(cp->m_colour);
+            set_def_colour(cp->m_colour, cp->m_cols, cp->ctype);
 
             wxCommandEvent evt(EVT_SELECTED_COLOR);
             unsigned long g_col = ((cp->m_colour.Red() & 0xff) << 24) + ((cp->m_colour.Green() & 0xff) << 16) + ((cp->m_colour.Blue() & 0xff) << 8) + (cp->m_colour.Alpha() & 0xff);
@@ -2058,9 +2186,68 @@ void ColorPickerPopup::set_ams_colours(std::vector<wxColour> ams)
     Fit();
 }
 
-void ColorPickerPopup::set_def_colour(wxColour col)
+void ColorPickerPopup::set_preset_colours(const std::vector<ColorItem>& preset_colors)
+{
+    m_other_fg_sizer->Clear(false);
+    if (!m_preset_color_pickers.empty()) {
+        for (ColorPicker* col_pick : m_preset_color_pickers) {
+            auto iter = std::find(m_color_pickers.begin(), m_color_pickers.end(), col_pick);
+            if (iter != m_color_pickers.end()) {
+                col_pick->Hide();
+                col_pick->Destroy();
+                m_color_pickers.erase(iter);
+            }
+        }
+        m_preset_color_pickers.clear();
+    }
+
+    for (const ColorItem& item : preset_colors) {
+        if (item.colors.empty()) continue;
+
+        auto cp = new ColorPicker(m_def_color_box, wxID_ANY, wxDefaultPosition, wxDefaultSize);
+        cp->set_color(item.colors.front());
+        cp->set_colors(item.colors);
+        cp->ctype = item.colors.size() > 1 ? item.ctype : 2;
+        cp->set_show_full(true);
+        cp->set_selected(false);
+        cp->SetBackgroundColour(StateColor::darkModeColorFor(wxColour(238,238,238)));
+        if (!item.name.empty()) {
+            cp->SetToolTip(item.name);
+        }
+
+        m_color_pickers.push_back(cp);
+        m_preset_color_pickers.push_back(cp);
+        cp->Bind(wxEVT_LEFT_DOWN, [this, cp](auto& e) {
+            set_def_colour(cp->m_colour, cp->m_cols, cp->ctype);
+
+            wxCommandEvent evt(EVT_SELECTED_COLOR);
+            unsigned long g_col = ((cp->m_colour.Red() & 0xff) << 24) + ((cp->m_colour.Green() & 0xff) << 16) + ((cp->m_colour.Blue() & 0xff) << 8) + (cp->m_colour.Alpha() & 0xff);
+            evt.SetInt(g_col);
+            wxPostEvent(GetParent(), evt);
+        });
+    }
+
+    const bool show_default_colours = m_preset_color_pickers.empty();
+    for (ColorPicker* cp : m_default_color_pickers) {
+        cp->Show(show_default_colours);
+    }
+    for (ColorPicker* cp : (show_default_colours ? m_default_color_pickers : m_preset_color_pickers)) {
+        m_other_fg_sizer->Add(cp, 0, wxALL, FromDIP(3));
+    }
+
+    m_other_fg_sizer->Layout();
+    Layout();
+    Fit();
+}
+
+void ColorPickerPopup::set_def_colour(wxColour col, std::vector<wxColour> cols, int ctype)
 {
     m_def_col = col;
+    if (cols.empty()) {
+        cols.push_back(col);
+    }
+    m_def_cols = cols;
+    m_def_ctype = m_def_cols.size() > 1 ? ctype : 2;
 
     for (ColorPicker* cp : m_color_pickers) {
         if (cp->m_selected) {
@@ -2069,7 +2256,8 @@ void ColorPickerPopup::set_def_colour(wxColour col)
     }
 
     for (ColorPicker* cp : m_color_pickers) {
-        if (cp->m_colour == m_def_col) {
+        if (!cp->IsShown()) continue;
+        if (cp->ctype == m_def_ctype && cp->m_cols == m_def_cols) {
             cp->set_selected(true);
             break;
         }

src/slic3r/GUI/AMSMaterialsSetting.hpp

@@ -82,14 +82,22 @@ public:
 class ColorPickerPopup : public PopupWindow
 {
 public:
+    struct ColorItem
+    {
+        std::vector<wxColour> colors;
+        int                   ctype = 2;
+        wxString              name;
+    };
+
     ScalableBitmap m_ts_bitmap_custom;
     wxStaticBitmap* m_ts_stbitmap_custom;
     StaticBox* m_custom_cp;
     StaticBox* m_def_color_box;
     wxFlexGridSizer* m_ams_fg_sizer;
+    wxFlexGridSizer* m_other_fg_sizer;
     wxColour m_def_col;
     std::vector<wxColour> m_def_colors;
-    std::vector<wxColour> m_ams_colors;
+    std::vector<ColorItem> m_ams_color_items;
     std::vector<ColorPicker*> m_color_pickers;
     std::vector<ColorPicker*> m_ams_color_pickers;
 
@@ -97,13 +105,20 @@ public:
     ColorPickerPopup(wxWindow* parent);
     ~ColorPickerPopup() {};
     void on_custom_clr_picker(wxMouseEvent& event);
-    void set_ams_colours(std::vector<wxColour> ams);
-    void set_def_colour(wxColour col);
+    void set_ams_colours(const std::vector<ColorItem>& ams);
+    void set_preset_colours(const std::vector<ColorItem>& preset_colors);
+    void set_def_colour(wxColour col, std::vector<wxColour> cols = {}, int ctype = 2);
+    const std::vector<wxColour>& get_selected_colours() const { return m_def_cols; }
+    int get_selected_ctype() const { return m_def_ctype; }
     void paintEvent(wxPaintEvent& evt);
     virtual void OnDismiss() wxOVERRIDE;
     virtual bool ProcessLeftDown(wxMouseEvent& event) wxOVERRIDE;
 
-public:
+private:
+    std::vector<ColorPicker*> m_default_color_pickers;
+    std::vector<ColorPicker*> m_preset_color_pickers;
+    std::vector<wxColour> m_def_cols;
+    int m_def_ctype = 2;
 };
 
 

src/slic3r/GUI/BackgroundSlicingProcess.cpp

@@ -747,7 +747,9 @@ Print::ApplyStatus BackgroundSlicingProcess::apply(const Model& model, const Dyn
 
     // Orca: prevent resetting under gcode viewer mode
     if (invalidated != PrintBase::APPLY_STATUS_UNCHANGED) {
-        const auto plater = GUI::wxGetApp().mainframe->m_plater;
+        // Use the app’s plater accessor instead of going through mainframe, otherwise anything during startup that reaches apply() 
+        // (loading presets, config or input files, e.g. from the CLI) therefore dereferences a null `mainframe` and crashes.
+        const auto plater = GUI::wxGetApp().plater();
         if (plater && plater->only_gcode_mode()) {
             invalidated = PrintBase::APPLY_STATUS_UNCHANGED;
         }

src/slic3r/GUI/DeviceManager.cpp

@@ -1798,7 +1798,8 @@ int MachineObject::command_ams_calibrate(int ams_id)
     return command_with_dialog(m_agent->command_ams_calibrate(get_dev_id(), ams_id, MachineObject::m_sequence_id++, is_lan_mode_printer()));
 }
 
-int MachineObject::command_ams_filament_settings(int ams_id, int slot_id, std::string filament_id, std::string setting_id, std::string tray_color, std::string tray_type, int nozzle_temp_min, int nozzle_temp_max)
+int MachineObject::command_ams_filament_settings(int ams_id, int slot_id, std::string filament_id, std::string setting_id, std::string tray_color, std::string tray_type,
+                                                 int nozzle_temp_min, int nozzle_temp_max, const std::vector<std::string>& tray_colors, int tray_ctype)
 {
     int tag_tray_id = 0;
     int tag_ams_id  = ams_id;
@@ -1827,6 +1828,10 @@ int MachineObject::command_ams_filament_settings(int ams_id, int slot_id, std::s
     j["print"]["nozzle_temp_min"]   = nozzle_temp_min;
     j["print"]["nozzle_temp_max"]   = nozzle_temp_max;
     j["print"]["tray_type"]         = tray_type;
+    if (!tray_colors.empty()) {
+        j["print"]["cols"]  = tray_colors;
+        j["print"]["ctype"] = tray_ctype;
+    }
 
     return this->publish_json(j);
 }
@@ -5478,6 +5483,7 @@ void MachineObject::parse_new_info(json print)
         is_support_pa_mode = (get_flag_bits_no_border(fun2, 3) == 1);
         is_support_remote_dry = (get_flag_bits_no_border(fun2, 5) == 1);
         is_support_check_track_switch_match_slice_printer = get_flag_bits_no_border(fun2, 19) == 1;
+        is_support_filament_manual_multi_color = (get_flag_bits_no_border(fun2, 23) == 1);
 
         if (DevPrinterConfigUtil::support_print_check_firmware_for_tpu_left(printer_type)) {
             m_firmware_support_print_tpu_left = get_flag_bits_no_border(fun2, 7) == 1;

src/slic3r/GUI/DeviceManager.hpp

@@ -674,6 +674,7 @@ public:
     bool is_support_remote_dry = false;
     bool is_support_check_track_switch_match_slice_printer{false};
     bool is_support_pa_mode{false};
+    bool is_support_filament_manual_multi_color{false};
     std::optional<bool> m_firmware_support_print_tpu_left;
 
     bool installed_upgrade_kit{false};
@@ -804,7 +805,7 @@ public:
     int command_ams_switch_filament(bool switch_filament);
     int command_ams_air_print_detect(bool air_print_detect);
     int command_ams_calibrate(int ams_id);
-    int command_ams_filament_settings(int ams_id, int slot_id, std::string filament_id, std::string setting_id, std::string tray_color, std::string tray_type, int nozzle_temp_min, int nozzle_temp_max);
+    int command_ams_filament_settings(int ams_id, int slot_id, std::string filament_id, std::string setting_id, std::string tray_color, std::string tray_type, int nozzle_temp_min, int nozzle_temp_max, const std::vector<std::string>& tray_colors = {}, int tray_ctype = 2);
     int command_ams_select_tray(std::string tray_id);
     int command_ams_refresh_rfid(int ams_id, int slot_id);
     int command_ams_control(std::string action);

src/slic3r/GUI/EncodedFilament.cpp

@@ -96,7 +96,7 @@ void FilamentColorCodeQuery::LoadFromLocal()
                     const auto& fila_color_strs = json_data_item["fila_color"].get<std::vector<wxString>>();
                     for (const auto& color_str : fila_color_strs) {
                         if (color_str.size() > 3) /* Skip the value like "#0"*/{
-                            fila_color.m_colors.emplace(wxColour(color_str));
+                            fila_color.AddColor(wxColour(color_str));
                         }
                     }
                 }

src/slic3r/GUI/EncodedFilament.hpp

@@ -87,6 +87,17 @@ struct FilamentColor
 
 public:
     size_t ColorCount() const noexcept { return m_colors.size(); }
+    // Preserve catalog endpoint order for display while retaining set-based lookup identity.
+    const std::vector<wxColour>& GetColors() const noexcept { return m_color_list; }
+    void AddColor(const wxColour& color)
+    {
+        if (m_colors.insert(color).second) m_color_list.push_back(color);
+    }
+
+private:
+    std::vector<wxColour> m_color_list;
+
+public:
 
     void EndSet(int ctype)
     {

src/slic3r/GUI/GLCanvas3D.cpp

@@ -65,6 +65,7 @@
 #include <string_view>
 #include "slic3r/GUI/Selection.hpp"
 #include "GLCanvas3D.hpp"
+#include "SequentialPrintClearance.hpp"
 #ifdef SLIC3R_CAD
 #include "slic3r/GUI/CAD/DesignSketchTool.hpp"   // Design tab: interactive 2D sketch tool
 #endif
@@ -6397,17 +6398,16 @@ void GLCanvas3D::update_sequential_clearance()
     //BBS: add the height logic
     PartPlate* plate = wxGetApp().plater()->get_partplate_list().get_curr_plate();
     Polygons polygons;
-    std::vector<std::pair<Polygon, float>> height_polygons;
     polygons.reserve(instances_count);
-    height_polygons.reserve(instances_count);
-    std::vector<struct height_info> convex_and_bounding_boxes;
-    struct height_info
-    {
-        double         instance_height;
-        BoundingBox    bounding_box;
-        Polygon        hull_polygon;
-    };
+    std::vector<SequentialClearanceInstance> convex_and_bounding_boxes;
+    convex_and_bounding_boxes.reserve(instances_count);
+    std::map<ObjectID, int> print_order;
+    for (const PrintObject* print_object : fff_print()->objects())
+        for (const PrintInstance& instance : print_object->instances())
+            print_order.emplace(instance.model_instance->id(), instance.model_instance->arrange_order);
+
     for (size_t i = 0; i < instance_transforms.size(); ++i) {
+        const size_t first_instance = convex_and_bounding_boxes.size();
         const auto& instances = instance_transforms[i];
         double rotation_z0 = instances.front()->get_rotation().z();
         int index = 0;
@@ -6427,96 +6427,20 @@ void GLCanvas3D::update_sequential_clearance()
             Polygon convex_hull(std::move(inst_pts));
             BoundingBox bouding_box = convex_hull.bounding_box();
             BoundingBox plate_bb = plate->get_bounding_box_crd();
+            const ObjectID instance_id = m_model->objects[i]->instances[index]->id();
             double instance_height = m_model->objects[i]->get_instance_max_z(index++);
             //skip the object for not current plate
             if (!plate_bb.overlap(bouding_box))
                 continue;
-            convex_and_bounding_boxes.push_back({instance_height, bouding_box, convex_hull});
+            convex_and_bounding_boxes.push_back({instance_height, bouding_box, convex_hull, instance_id});
             polygons.emplace_back(std::move(convex_hull));
         }
+        sort_sequential_clearance_instances(convex_and_bounding_boxes.begin() + first_instance, convex_and_bounding_boxes.end(), print_order);
     }
 
-    //sort the print instance
-    std::sort(convex_and_bounding_boxes.begin(), convex_and_bounding_boxes.end(),
-        [](auto &l, auto &r) {
-            auto ly1 = l.bounding_box.min.y();
-            auto ly2 = l.bounding_box.max.y();
-            auto ry1 = r.bounding_box.min.y();
-            auto ry2 = r.bounding_box.max.y();
-            auto inter_min = std::max(ly1, ry1);
-            auto inter_max = std::min(ly2, ry2);
-            auto lx = l.bounding_box.min.x();
-            auto rx = r.bounding_box.min.x();
-            if (inter_max - inter_min > 0)
-                return (lx < rx) || ((lx == rx)&&(ly1 < ry1));
-            else
-                return (ly1 < ry1);
-        });
-
-    /*bool has_interlaced_objects = false;
-    for (int k = 0; k < bounding_box_count; k++)
-    {
-        Polygon& convex = convex_and_bounding_boxes[k].hull_polygon;
-        BoundingBox& bbox = convex_and_bounding_boxes[k].bounding_box;
-        auto iy1 = bbox.min.y();
-        auto iy2 = bbox.max.y();
-
-        for (int i = k+1; i < bounding_box_count; i++)
-        {
-            Polygon&     next_convex = convex_and_bounding_boxes[i].hull_polygon;
-            BoundingBox& next_bbox   = convex_and_bounding_boxes[i].bounding_box;
-            auto py1 = next_bbox.min.y();
-            auto py2 = next_bbox.max.y();
-            auto inter_min = std::max(iy1, py1); // min y of intersection
-            auto inter_max = std::min(iy2, py2); // max y of intersection. length=max_y-min_y>0 means intersection exists
-            if (inter_max - inter_min > 0) {
-                has_interlaced_objects = true;
-                break;
-            }
-        }
-        if (has_interlaced_objects)
-            break;
-    }*/
-
-    int bounding_box_count = convex_and_bounding_boxes.size();
-    double printable_height = fff_print()->config().printable_height;
-    double hc1 = fff_print()->config().extruder_clearance_height_to_lid;
-    double hc2 = fff_print()->config().extruder_clearance_height_to_rod;
-    for (int k = 0; k < bounding_box_count; k++)
-    {
-        Polygon& convex = convex_and_bounding_boxes[k].hull_polygon;
-        BoundingBox& bbox = convex_and_bounding_boxes[k].bounding_box;
-        auto iy1 = bbox.min.y();
-        auto iy2 = bbox.max.y();
-        double height = (k == (bounding_box_count - 1))?printable_height:hc1;
-
-        /*if (has_interlaced_objects) {
-            if ((k < (bounding_box_count - 1)) && (convex_and_bounding_boxes[k].instance_height > hc2)) {
-                height_polygons.emplace_back(std::make_pair(convex, hc2));
-            }
-        }
-        else {
-            if ((k < (bounding_box_count - 1)) && (convex_and_bounding_boxes[k].instance_height > hc1)) {
-                height_polygons.emplace_back(std::make_pair(convex, hc1));
-            }
-        }*/
-
-        for (int i = k+1; i < bounding_box_count; i++)
-        {
-            Polygon&     next_convex = convex_and_bounding_boxes[i].hull_polygon;
-            BoundingBox& next_bbox   = convex_and_bounding_boxes[i].bounding_box;
-            auto py1 = next_bbox.min.y();
-            auto py2 = next_bbox.max.y();
-            auto inter_min = std::max(iy1, py1); // min y of intersection
-            auto inter_max = std::min(iy2, py2); // max y of intersection. length=max_y-min_y>0 means intersection exists
-            if (inter_max - inter_min > 0) {
-                height = hc2;
-                break;
-            }
-        }
-        if (height < convex_and_bounding_boxes[k].instance_height)
-            height_polygons.emplace_back(std::make_pair(convex, height));
-    }
+    const auto& config = fff_print()->config();
+    const auto height_polygons = sequential_clearance_height_polygons(convex_and_bounding_boxes, config.printable_height,
+        config.extruder_clearance_height_to_lid, config.extruder_clearance_height_to_rod);
 
     // sends instances 2d hulls to be rendered
     set_sequential_print_clearance_visible(true);

src/slic3r/GUI/GLShadersManager.cpp

@@ -114,6 +114,8 @@ std::pair<bool, std::string> GLShadersManager::init()
         valid &= append_shader("mm_gouraud", { prefix + "mm_gouraud.vs", prefix + "mm_gouraud.fs" });
     // Fast shaded preview for the texture displacement gizmo (see libslic3r/TextureDisplacement.hpp).
     valid &= append_shader("texture_displacement_shaded", { prefix + "texture_displacement_shaded.vs", prefix + "texture_displacement_shaded.fs" });
+    // ChromaStack: the printed colours on the model, read per fragment from its own height.
+    valid &= append_shader("chromastack", { prefix + "chromastack.vs", prefix + "chromastack.fs" });
     // UV-check overlay for the same gizmo: a procedural checker or a distortion heatmap over the
     // painted patch, to sanity-check the unwrap.
     valid &= append_shader("texture_displacement_uvcheck", { prefix + "texture_displacement_uvcheck.vs", prefix + "texture_displacement_uvcheck.fs" });

src/slic3r/GUI/GUI_App.cpp

@@ -252,9 +252,11 @@
 #include "slic3r/GUI/ParamsDialog.hpp"
 #include "slic3r/GUI/ParamsPanel.hpp"
 #include "slic3r/GUI/PartPlate.hpp"
+#include "slic3r/GUI/TextureLibrary.hpp"
 #include "slic3r/GUI/Widgets/Button.hpp"
 #include "slic3r/GUI/wxExtensions.hpp"
 #include "slic3r/plugin/host/PluginPages.hpp"
+#include "slic3r/plugin/host/PluginResourceTextures.hpp"
 #include <wx/defs.h>
 #include "slic3r/GUI/Widgets/WebView.hpp"
 #include <cwchar>
@@ -333,6 +335,23 @@ static std::shared_ptr<OrcaCloudServiceAgent> get_orca_cloud_agent(NetworkAgent*
     return agent ? std::dynamic_pointer_cast<OrcaCloudServiceAgent>(agent->get_cloud_agent()) : nullptr;
 }
 
+// Full {web} page URL for a same-site OrcaCloud path, or empty when it cannot be built. An already
+// open cloud pane is navigated straight to it, without re-running the session handoff.
+static wxString cloud_page_url(NetworkAgent* agent, const std::string& path)
+{
+    if (path.empty())
+        return {};
+    auto orca = get_orca_cloud_agent(agent);
+    if (!orca)
+        return {};
+    std::string base = orca->get_cloud_base_url();
+    while (!base.empty() && base.back() == '/')
+        base.pop_back();
+    if (base.empty())
+        return {};
+    return from_u8(base + path);
+}
+
 void start_ping_test()
 {
     return;
@@ -3063,6 +3082,37 @@ void GUI_App::init_plugin_gui_wiring()
             refresh_plugins_dialog();
             switch_printer_agent_after_unload(capability.plugin_key);
         });
+    plugin_mgr.subscribe_on_load_callback([](const std::string& plugin_key) {
+        if (wxTheApp == nullptr || wxGetApp().is_closing())
+            return;
+        PluginResourceTextures::instance().on_plugin_register(plugin_key);
+        Slic3r::GUI::bump_texture_library_generation();
+    });
+    plugin_mgr.subscribe_on_unload_callback([](const std::string& plugin_key) {
+        if (wxTheApp == nullptr || wxGetApp().is_closing())
+            return;
+        PluginResourceTextures::instance().on_plugin_deregister(plugin_key);
+        Slic3r::GUI::bump_texture_library_generation();
+    });
+    plugin_mgr.subscribe_on_capability_load_callback([](const PluginCapabilityId& capability) {
+        if (capability.type != PluginCapabilityType::ResourceTexture)
+            return;
+        if (wxTheApp == nullptr || wxGetApp().is_closing())
+            return;
+        PluginResourceTextures::instance().on_cap_register(capability);
+        Slic3r::GUI::bump_texture_library_generation();
+    });
+    plugin_mgr.subscribe_on_capability_unload_callback([](const PluginCapabilityId& capability) {
+        if (capability.type != PluginCapabilityType::ResourceTexture)
+            return;
+        if (wxTheApp == nullptr || wxGetApp().is_closing())
+            return;
+        PluginResourceTextures::instance().on_cap_deregister(capability);
+        Slic3r::GUI::bump_texture_library_generation();
+    });
+    // Packs already loaded before this wiring, or in tests where no callback fires.
+    PluginResourceTextures::instance().initial_sweep();
+    Slic3r::GUI::bump_texture_library_generation();
 }
 
 bool GUI_App::on_init_inner()
@@ -5434,12 +5484,7 @@ std::string GUI_App::handle_web_request(std::string cmd)
                 this->request_open_project({});
             }
             else if (command_str.compare("homepage_open_cloud") == 0) {
-                // Deferred: set_cloud_mode() may create a second native webview, which must
-                // not run inside this webview's script-message callback.
-                CallAfter([this]() {
-                    if (WebViewPanel* home = WebViewPanel::if_built())
-                        home->set_cloud_mode(true, cloud_webview_handoff());
-                });
+                this->open_orca_cloud();
             }
             else if (command_str.compare("homepage_close_cloud") == 0) {
                 CallAfter([]() {
@@ -5588,6 +5633,22 @@ std::string GUI_App::handle_web_request(std::string cmd)
     return "";
 }
 
+void GUI_App::open_orca_cloud(const std::string& next)
+{
+    // Remembered so a login while the pane is open re-seeds the same page rather than the default.
+    m_cloud_open_path = next;
+    // Deferred: set_cloud_mode() may create the second native webview, so this must not run inside
+    // the caller's frame (a webview script-message callback or an ImGui frame).
+    CallAfter([this]() {
+        if (mainframe != nullptr)
+            mainframe->select_tab(TAB_ID_HOME);
+        // The Home page is built lazily, so the tab switch may not have created it yet: ensure() builds
+        // it if needed, matching what showing the tab does, rather than dropping the open request.
+        if (WebViewPanel* home = WebViewPanel::ensure())
+            home->set_cloud_mode(true, cloud_webview_handoff(m_cloud_open_path), cloud_page_url(m_agent, m_cloud_open_path));
+    });
+}
+
 void GUI_App::handle_script_message(std::string msg, const std::string& provider)
 {
     try {
@@ -5865,7 +5926,7 @@ void GUI_App::on_update_machine_list(wxCommandEvent &evt)
      }*/
 }
 
-wxString GUI_App::cloud_webview_handoff()
+wxString GUI_App::cloud_webview_handoff(const std::string& next)
 {
     // The cloud pane loads the Start URL, not a token: the page checks the webview's own session
     // and asks for a handoff itself (guide §4). Signed out, user_id is omitted and the page will
@@ -5873,7 +5934,7 @@ wxString GUI_App::cloud_webview_handoff()
     auto orca = get_orca_cloud_agent(m_agent);
     if (!orca)
         return {};
-    return from_u8(orca->build_web_session_start_url());
+    return from_u8(orca->build_web_session_start_url(next));
 }
 
 void GUI_App::request_cloud_web_session(const std::string& state)
@@ -6016,7 +6077,7 @@ void GUI_App::on_user_login_handle(wxCommandEvent &evt)
     m_cloud_handoff_state.clear();
     if (WebViewPanel* home = WebViewPanel::if_built())
         if (home->cloud_mode_active())
-            home->set_cloud_mode(true, cloud_webview_handoff());
+            home->set_cloud_mode(true, cloud_webview_handoff(m_cloud_open_path));
 }
 
 
@@ -6863,7 +6924,7 @@ void  GUI_App::push_notification(const MachineObject* obj, wxString msg, wxStrin
     }
 }
 
-void GUI_App::reload_settings()
+bool GUI_App::reload_settings()
 {
     if (preset_bundle && m_agent) {
         // Load user's personal presets
@@ -6871,7 +6932,7 @@ void GUI_App::reload_settings()
         int result = m_agent->get_user_presets(&user_presets);
         if (result != 0) {
             BOOST_LOG_TRIVIAL(error) << __FUNCTION__ << ": get_user_presets failed with code " << result << ", skipping sync";
-            return;
+            return false;
         }
         BOOST_LOG_TRIVIAL(info) << __FUNCTION__ << __LINE__ << " cloud user preset number is: " << user_presets.size();
         auto refresh_synced_ui = [this, user_presets = std::move(user_presets)]() mutable {
@@ -6946,13 +7007,30 @@ void GUI_App::reload_settings()
             refresh_synced_ui();
         else
             CallAfter(refresh_synced_ui);
+        return true;
     }
+    return false;
 }
 
 void GUI_App::check_legacy_user_filament_presets(bool by_user)
 {
+    // One check at a time: the startup check can come due while the manual one is open, and on macOS the menu bar
+    // stays usable while a modal dialog is up.
+    if (m_legacy_filament_check_running)
+        return;
     if (!by_user && app_config->get_bool("legacy_user_filament_presets_dont_remind"))
         return;
+    // The local copies may be older than the cloud ones another machine changed: migrating them would push changes
+    // made against stale copies and conflict with the cloud. The startup check waits for the presets to be pulled.
+    if (!by_user && m_user_preset_pull_pending) {
+        m_legacy_filament_check_due = true;
+        return;
+    }
+    // This check also answers a startup check that is still waiting.
+    m_legacy_filament_check_due = false;
+    m_legacy_filament_check_running = true;
+    ScopeGuard running_guard([this] { m_legacy_filament_check_running = false; });
+
     const std::vector<LegacyUserFilamentPreset> legacy_presets = preset_bundle->legacy_user_filament_presets();
     if (legacy_presets.empty()) {
         if (by_user)
@@ -6997,6 +7075,9 @@ void GUI_App::check_legacy_user_filament_presets(bool by_user)
         show_error(mainframe, format_wxstr(_L("Your filament presets could not be migrated. A backup of your presets was saved to:\n%1%"), backup));
         return;
     }
+    // Ask the sync loop to push on its next pass instead of its next 60 s round, which shortens the time the migrated
+    // presets stay unpushed.
+    m_sync_user_presets_now.store(true, std::memory_order_release);
     preset_bundle->update_compatible(PresetSelectCompatibleType::Never);
     mainframe->update_side_preset_ui();
     for (auto tab : tabs_list) {
@@ -7763,6 +7844,7 @@ void GUI_App::start_sync_user_preset(bool with_progress_dlg)
     BOOST_LOG_TRIVIAL(info) << "start_sync_service...";
     // BBS
     m_user_sync_token.reset(new int(0));
+    m_user_preset_pull_pending = true;
     if (with_progress_dlg) {
         // Mark a manual progress dialog as active so restart_sync_user_preset() ignores
         // repeat triggers while it is on screen (prevents stacking modal dialogs).
@@ -7836,8 +7918,16 @@ void GUI_App::start_sync_user_preset(bool with_progress_dlg)
 
             finishFn(ret == 0);
 
-            if (ret == 0 && m_agent && !t.expired())
-                reload_settings();
+            // Queued after the refresh reload_settings() queues, so it runs once the pulled presets are loaded.
+            if (ret == 0 && m_agent && !t.expired() && reload_settings())
+                CallAfter([this, t] {
+                    // A stopped or restarted sync: the next sync, if one starts, pulls the presets again.
+                    if (t.expired() || is_closing())
+                        return;
+                    m_user_preset_pull_pending = false;
+                    if (m_legacy_filament_check_due)
+                        check_legacy_user_filament_presets(false);
+                });
 
             // For orca specific syncing
             auto orca_agent = std::dynamic_pointer_cast<OrcaCloudServiceAgent>(m_agent->get_cloud_agent());
@@ -7852,8 +7942,11 @@ void GUI_App::start_sync_user_preset(bool with_progress_dlg)
             // Sync once immediately, then every 60 seconds.
             while (!t.expired()) {
                 ++tick_tock;
-                // Sync once immediately, then every 60s, or right away when a force-push asked for it.
-                if (tick_tock % 120 == 0 || m_sync_user_presets_now.exchange(false, std::memory_order_acq_rel)) {
+                // Sync once immediately, then every 60s, or right away when a force-push or a preset migration asked
+                // for it. The request is consumed on every tick, so one made before a scheduled pass is served by it
+                // instead of running a second pass straight after.
+                const bool push_now = m_sync_user_presets_now.exchange(false, std::memory_order_acq_rel);
+                if (tick_tock % 120 == 0 || push_now) {
                     tick_tock = 0;
                     if (m_agent) {
                         if (!m_agent->is_user_login()) {

src/slic3r/GUI/GUI_App.hpp

@@ -355,6 +355,9 @@ private:
     std::atomic<bool>    m_restart_sync_pending {false};
     std::atomic<bool>    m_sync_user_preset_dlg_active {false}; // a manual "Sync Presets" progress dialog is on screen (see restart_sync_user_preset)
     std::atomic<bool>    m_sync_user_presets_now {false}; // request the sync loop to push user presets on its next tick
+    bool                 m_user_preset_pull_pending {false}; // set when a sync thread starts, cleared once its pull is applied; a failed pull or a stopped sync leaves it set
+    bool                 m_legacy_filament_check_due {false}; // the startup check waits for that pull
+    bool                 m_legacy_filament_check_running {false}; // a check is in progress, its dialog may be open
     std::atomic<bool>    m_migration_retry_pending {false};
     bool             m_is_dark_mode{ false };
     bool             m_adding_script_handler { false };
@@ -365,6 +368,8 @@ private:
     // and a generation counter so a superseded mint/login result is dropped (guide §5, "only the
     // newest request counts").
     std::string      m_cloud_handoff_state;
+    // The OrcaCloud path the open requests target, remembered so a login re-seed keeps the same page.
+    std::string      m_cloud_open_path;
     uint64_t         m_cloud_handoff_gen { 0 };
     bool             m_show_error_msgdlg{false};
     wxString         m_info_dialog_content;
@@ -466,6 +471,9 @@ public:
 #if wxUSE_WEBVIEW_EDGE
     void            init_webview_runtime();
 #endif
+    // Switches to the Home tab and opens the OrcaCloud pane beside the homepage on `next`, a
+    // same-site OrcaCloud path (empty = the page's default). Work is deferred internally.
+    void            open_orca_cloud(const std::string& next = std::string());
     static unsigned get_colour_approx_luma(const wxColour& colour);
     static bool     dark_mode();
     const wxColour  get_label_default_clr_system();
@@ -572,7 +580,7 @@ public:
     // called from the pane when the page asks for a session; it opens the desktop login dialog if
     // signed out, otherwise mints a one-time token on a worker thread and loads the Complete URL
     // in the same webview.
-    wxString        cloud_webview_handoff();
+    wxString        cloud_webview_handoff(const std::string& next = std::string());
     void            request_cloud_web_session(const std::string& state);
     void            mint_cloud_web_session(const std::string& state, uint64_t gen);
     void            on_cloud_web_session_minted(const std::string& state, uint64_t gen, const std::string& token_hash);
@@ -589,7 +597,8 @@ public:
     void            check_new_version_sf(bool show_tips = false, int by_user = 0);
     // Offers to migrate user filament presets made against an older release that lost printers their parent
     // listed then. At startup it asks on every start that finds such presets, unless the user chose not to be
-    // reminded; by_user also reports when there is nothing to migrate.
+    // reminded; while the user presets sync with the cloud, it waits until they have been pulled. by_user also
+    // reports when there is nothing to migrate.
     void            check_legacy_user_filament_presets(bool by_user);
     bool            process_network_msg(std::string dev_id, std::string msg);
     void            request_new_version(int by_user);
@@ -600,7 +609,8 @@ public:
     std::string     format_IP(const std::string& ip);
     void            show_dialog(wxString msg);
     void            push_notification(const MachineObject* obj, wxString msg, wxString title = wxEmptyString, UserNotificationStyle style = UserNotificationStyle::UNS_NORMAL);
-    void            reload_settings();
+    // Returns false when the cloud presets could not be fetched, so nothing was reloaded.
+    bool            reload_settings();
     void            remove_user_presets();
 
     bool            maybe_migrate_user_presets_on_login();

src/slic3r/GUI/GUI_ObjectList.cpp

@@ -2875,6 +2875,37 @@ void ObjectList::del_layer_from_object(const int obj_idx, const t_layer_height_r
     changed_object(obj_idx);
 }
 
+void ObjectList::rebuild_layer_ranges(const int obj_idx)
+{
+    if (obj_idx < 0 || obj_idx >= int(m_objects->size()))
+        return;
+    const wxDataViewItem obj_item = m_objects_model->GetItemById(obj_idx);
+    if (!obj_item.IsOk())
+        return;
+
+    const t_layer_config_ranges& ranges = object(obj_idx)->layer_config_ranges;
+    // Adding and deleting tree items fires selection events on GTK; suppress them as every other
+    // bulk edit of the tree here does.
+    const bool prevent_events = m_prevent_list_events;
+    m_prevent_list_events = true;
+    wxDataViewItem root = m_objects_model->GetLayerRootItem(obj_item);
+    if (root.IsOk()) {
+        m_objects_model->DeleteChildren(root);
+        if (ranges.empty())
+            m_objects_model->Delete(root);
+        else {
+            for (const auto& range : ranges)
+                add_layer_item(range.first, root);
+            Expand(root);
+        }
+    } else {
+        add_layer_root_item(obj_item);
+    }
+    m_prevent_list_events = prevent_events;
+
+    changed_object(obj_idx);
+}
+
 void ObjectList::del_layers_from_object(const int obj_idx)
 {
     object(obj_idx)->layer_config_ranges.clear();

src/slic3r/GUI/GUI_ObjectList.hpp

@@ -321,6 +321,11 @@ public:
     void                del_settings_from_config(const wxDataViewItem& parent_item);
     void                del_instances_from_object(const int obj_idx);
     void                del_layer_from_object(const int obj_idx, const t_layer_height_range& layer_range);
+    // Mirrors an object's height ranges in the tree after they have been rewritten from outside the
+    // list - the texture displacement gizmo writes a whole set at once for a ChromaStack stack's
+    // filament swaps. Deliberately leaves the selection alone: changing it while a gizmo is open
+    // tears the gizmo down under the user.
+    void                rebuild_layer_ranges(const int obj_idx);
     void                del_layers_from_object(const int obj_idx);
     bool                del_from_cut_object(bool is_connector, bool is_model_part = false, bool is_negative_volume = false);
     bool                del_subobject_from_object(const int obj_idx, const int idx, const int type);

src/slic3r/GUI/Gizmos/GLGizmoTextureDisplacement.hpp

@@ -264,7 +264,11 @@ private:
     ModelVolume*       texture_volume();
     const ModelVolume* texture_volume() const;
 
-    void add_texture_layer();
+    // `chroma` picks the kind: a relief layer, a height map pushed into the surface, or a ChromaStack
+    // picture, a colour picture made of filament swaps. The kind is a property of the layer and is
+    // fixed when it is created - there is no panel-wide mode, so a plaque can carry a picture on its
+    // face and a knurl on its rim at the same time.
+    void add_texture_layer(bool chroma = false);
     void remove_texture_layer(int slot);
     // Moves the layer in `slot` to position `to_index` of the slot-ordered stack (an insertion index,
     // 0..layer count), which is the order blend modes apply in. Layers keep their data; what moves is
@@ -428,9 +432,116 @@ private:
     void render_texture_library_popup(const ImVec2 &panel_min, const ImVec2 &panel_max);
     int  m_picker_slot         = -1;
     bool m_picker_open_request = false;
+    // Set when the popup opens, so the next layout scrolls the layer's current texture into view.
+    bool  m_picker_reveal_selected = false;
+    float m_picker_scroll_target   = 0.f;  // see ease_wheel_scroll()
+    float m_picker_scroll_applied  = -1.f;
     void set_layer_texture(TextureDisplacementLayer &layer, const TextureLibraryEntry &entry);
     void import_custom_texture(TextureDisplacementLayer &layer);
 
+    // ---- ChromaStack (see libslic3r/ChromaStack.hpp) ----
+    //
+    // A ChromaStack layer is an ordinary texture displacement layer whose height map is computed rather
+    // than imported: the picture the user picked is kept in `chroma_source`, the stack in
+    // `chroma`, and `image_data` holds whatever the last solve produced. Everything downstream -
+    // projection, preview, bake - therefore needs no changes at all, and this is the whole of the
+    // feature's GUI side: pick a picture, edit the stack, solve, hand the filament swaps to the
+    // object's height ranges.
+
+    // The project's loaded filaments as the colour model sees them: colour from the project, TD from
+    // each filament's preset. Index is the filament index the swaps are written from.
+    static std::vector<ChromaFilament> chroma_filaments();
+    // The layer height the print profile is set to, mm; 0 when there is no print preset. The solve is
+    // only valid for the layer height it was run at, so the panel compares it against the stack's.
+    static float print_layer_height();
+
+    // Where the active layer's painted patch sits, in the object's own printed coordinates - which is
+    // what the stack has to be lined up with, because a filament swap happens at a print Z and the
+    // relief grows upward from this patch.
+    struct ChromaPatch
+    {
+        bool  valid     = false;
+        float top_mm    = 0.f; // the patch's surface above the bottom of the object
+        float spread_mm = 0.f; // how far from flat the patch is in Z; 0 is a horizontal face
+        // The patch's extent in the plane the texture is projected from, in the bake frame - i.e. the
+        // (x, y) the Z-facing plane of the tri-planar projection samples by. This is what the picture
+        // has to be placed against: tiling_scale and offset are in that space, not in the patch's.
+        Vec2f planar_min = Vec2f::Zero();
+        Vec2f planar_max = Vec2f::Zero();
+    };
+    ChromaPatch chroma_patch(const TextureDisplacementLayer &layer) const;
+    // The painted face's height in layers, which is what the stack's base has to be: -1 when nothing
+    // is painted yet or there is no layer height to divide by.
+    int         chroma_measured_base(const TextureDisplacementLayer &layer) const;
+    // Extracting the painted patch is far too expensive to repeat once a frame, which is how often the
+    // panel wants the number. The paint's own timestamp and the transform the patch was measured in are
+    // between them everything the answer depends on.
+    mutable std::array<ChromaPatch, TEXTURE_DISPLACEMENT_MAX_LAYERS> m_chroma_patch_cache{};
+    mutable std::array<uint64_t, TEXTURE_DISPLACEMENT_MAX_LAYERS>      m_chroma_patch_stamp{};
+    mutable std::array<Transform3d, TEXTURE_DISPLACEMENT_MAX_LAYERS>   m_chroma_patch_trafo{};
+    // The patch is measured in two frames at once - the instance's, for the print height, and the
+    // bake's, for the placement - so both are part of what the cached answer depends on.
+    mutable std::array<Transform3d, TEXTURE_DISPLACEMENT_MAX_LAYERS>   m_chroma_patch_bake{};
+
+    // Picks a colour picture for the layer, turns ChromaStack on, and seeds a stack from the loaded
+    // filaments and the patch if the layer has none yet. Solves straight away, so the relief appears
+    // without a second click.
+    void chroma_pick_picture(TextureDisplacementLayer &layer);
+    // Re-runs the solve and replaces the layer's height map with its result. False (with the reason
+    // shown to the user) when the layer has no picture, no stack or no relief layers.
+    bool chroma_solve_layer(TextureDisplacementLayer &layer);
+    // Writes the stack's filament swaps onto the object as height ranges, replacing whatever ranges it
+    // had. Takes its own undo snapshot.
+    void chroma_apply_swaps(const TextureDisplacementLayer &layer);
+    // The stack's swaps as the object's height ranges already express them: true when applying would
+    // change nothing, which is what the panel's hint keys off.
+    bool chroma_swaps_applied(const TextureDisplacementLayer &layer) const;
+    // The whole of a ChromaStack card between its header and More settings: the picture and the print
+    // it predicts, the base, the filaments and the colours they reach, and the two actions.
+    void render_chroma_card(TextureDisplacementLayer &layer, float card_pad, float label_w, float wrap_w);
+
+    // The surface drawn in the colours the stack will print, over whatever neutral surface is there:
+    // the live preview's displaced mesh before a bake, the volume's own mesh after one. Every triangle
+    // takes the palette colour of the printed layer its centroid falls in (chroma_level_at()), which
+    // is what the eye will see at that height - so the relief's terraces read as the picture, the walls
+    // between them show the bands they cut through, and looking straight down gives the blended colours
+    // the whole feature exists to produce.
+    //
+    // Nothing else in the application can show this: the colours are blends of translucent filaments
+    // that no single filament can make, so they cannot be expressed as the per-triangle filament
+    // assignment the 3D scene paints with. They exist only here and in the sliced preview.
+    void rebuild_chroma_colors(const TextureDisplacementLayer &layer);
+    bool render_chroma_colors(const TextureDisplacementLayer &layer);
+    GLModel m_chroma_color_model;
+    // The surface the overlay was built from. The mesh alone: the colours are not in it, so a change
+    // to the stack costs a palette upload rather than a mesh rebuild.
+    size_t  m_chroma_color_key = 0;
+    // The stack's colours, one texel per printed layer, and what they were built from. A texture
+    // because a uniform array cannot hold a base of several hundred layers.
+    std::unique_ptr<GLTexture> m_chroma_palette_tex;
+    size_t                     m_chroma_palette_key = 0;
+    bool                       m_show_chroma_colors = true;
+
+    // What the last solve of each slot reported, for the panel. Not saved: it is a property of the
+    // solve, and a project is reopened with whatever height map the solve produced.
+    struct ChromaReport
+    {
+        bool   valid = false;
+        double mean  = 0.;
+        double max   = 0.;
+    };
+    std::array<ChromaReport, TEXTURE_DISPLACEMENT_MAX_LAYERS>              m_chroma_report{};
+    // The predicted print, per slot: the palette colours the solve chose, as a panel thumbnail. The
+    // 3D view cannot show this - the colours do not exist until the print swaps filament - so it is
+    // the only place the user sees what the stack will actually produce.
+    std::array<std::unique_ptr<GLTexture>, TEXTURE_DISPLACEMENT_MAX_LAYERS>  m_chroma_preview;
+    // The source picture's thumbnail, beside the prediction, keyed by the image it was built from.
+    std::array<std::unique_ptr<GLTexture>, TEXTURE_DISPLACEMENT_MAX_LAYERS>  m_chroma_source_thumb;
+    std::array<const void *, TEXTURE_DISPLACEMENT_MAX_LAYERS>                m_chroma_source_key{};
+    // Per slot: the stack has been edited since the height map was solved, so the relief on screen is the
+    // previous stack's. Nothing else can show this - a stale height map looks like a perfectly good one.
+    std::array<bool, TEXTURE_DISPLACEMENT_MAX_LAYERS>                        m_chroma_stale{};
+
     // Paint / Erase, from the panel. See get_left_button_state_type().
     bool m_erase_mode = false;
     // Per slot: whether the layer's card shows every setting or only Depth, Tile size and Rotation.
@@ -508,120 +619,25 @@ private:
     // is where the live adjustment is applied. Called whenever the panel moves or reverts it.
     void invalidate_adjust_textures();
 
-    // "Adjust Texture" mode: instead of painting, the active layer's placement is a control lying
-    // on the model. The gizmo is one tile of the texture in the paint patch's own tangent plane:
-    // where it sits is the layer's offset, how it is turned is rotation_deg, how big it is is the
-    // tile extents - so each handle is one of the values it edits rather than a stand-in for it.
-    // Anchored to the centroid/average-normal of the active layer's paint patch (see
-    // libslic3r::compute_layer_paint_anchor()), so nothing is drawn until something is painted.
+    // "Adjust Texture" mode: instead of painting, dragging an on-canvas handle changes the active
+    // layer's offset. The handle is a flat panel lying in the paint patch's own tangent plane
+    // (a "pan" - drag anywhere on it for free 2D movement), plus two arrows along the patch's
+    // own U/V axes that constrain the drag to just that one axis for precise nudging. Anchored to
+    // the centroid/average-normal of the active layer's current paint patch (see
+    // libslic3r::compute_layer_paint_anchor()), so nothing is drawn if it has nothing painted yet.
     //
-    // Drawn through ImGui's foreground draw list rather than the "flat" shader, which is what gives
-    // it antialiased sub-pixel strokes, rounded joins, a contrast halo and the numeric readout. The
-    // cost is that render_adjust_texture_gizmo() has to be called from inside the gizmo's ImGui
-    // frame (see on_render_input_window()), not from on_render() with the rest of the GL drawing.
-    // Move is the whole interior, free in both axes. AxisMove is the four arrows, each holding a
-    // move to one of the tile's axes. EdgeScale is the four edge bars, one tile extent each, and
-    // CornerScale the four corner grips, which take both extents at once.
-    //
-    // The arrows and the bars share the same four rays out of the tile, so they are kept apart by
-    // distance rather than direction: the bar hugs its edge, then a clear gap, then the arrow well
-    // outside it. adjust_pick() leans on that - see the order it tests them in.
-    enum class AdjustHandle { None, Move, AxisMove, EdgeScale, CornerScale, Rotate };
-
-    // The gizmo's geometry for one frame, in screen pixels. Worked out once and shared by the
-    // drawing and the hit test, so the two cannot disagree about where a handle is. Everything
-    // except the frame itself is a fixed number of pixels, which is what keeps the control equally
-    // usable on a 5 mm part and a 500 mm one - the old handle scaled with the volume's bounding box
-    // and so was unusable at both ends.
-    struct AdjustFrame
-    {
-        bool  valid  = false;
-        Vec2d center = Vec2d::Zero();
-        // +u+v, +u-v, -u-v, -u+v. A scale grip sits on each.
-        std::array<Vec2d, 4> corner{};
-        // +u, -u, +v, -v: the middle of each edge, and the two ends of the bar that sits just
-        // outside it. The bar's ends are kept rather than recomputed so the drawing and the hit test
-        // cannot end up with it in two different places.
-        std::array<Vec2d, 4> edge_mid{};
-        std::array<Vec2d, 4> bar_a{};
-        std::array<Vec2d, 4> bar_b{};
-        // The axis arrow further out on the same ray: where its shaft starts and where its chevron
-        // ends. Kept, like the bar's ends, so the drawing and the hit test cannot disagree.
-        std::array<Vec2d, 4> arm_base{};
-        std::array<Vec2d, 4> arm_tip{};
-        std::array<Vec2d, 4> arm_dir{}; // unit, pointing outward from the centre
-        // A ring is placed off every corner, but only one is drawn and only that one can be picked:
-        // four marks for a single operation is clutter, and whichever corner the cursor is near is
-        // the one worth offering. `rotate_shown` says which, and see update_rotate_corner() for why
-        // it does not simply track the nearest.
-        std::array<Vec2d, 4> rotate_at{};
-        int                  rotate_shown = 0;
-        // The tile projects to less than ADJUST_MIN_TILE_PX across, so the frame was drawn at that
-        // floor instead. It is dashed to say so, and the readout still gives the true millimetres.
-        bool clamped = false;
-        // What the frame was built from, in mesh-local mm: the tile's own (rotated) edge directions
-        // and its extents. The drag arithmetic reuses these rather than recomputing them, so it
-        // cannot end up working against a different tile than the one on screen.
-        Vec3f e_u = Vec3f::UnitX(), e_v = Vec3f::UnitY();
-        Vec2f tile_mm = Vec2f::Ones();     // (u, v) as stored on the layer
-        Vec2f half_planar = Vec2f::Ones(); // half extents in planar mm, v already aspect-corrected
-        Vec2f center_planar = Vec2f::Zero();
-    };
-
+    // NOTE: the drag direction/sign below is this session's best-effort reasoning about which way
+    // the texture should appear to move as the handle is dragged - it could not be visually
+    // confirmed while writing it (no way to render/see pixels in this environment), so it may
+    // need a one-line sign flip once actually tested.
     bool update_adjust_anchor(); // recomputes m_adjust_anchor_pos/normal; false if nothing painted
-    // The selected instance's volume-to-world transform, which the bake frame is derived from.
-    bool adjust_volume_to_world(Transform3d &out) const;
-    // The anchor, carried from mesh-local into the bake frame. Converted on every use rather than
-    // stored converted: the object can be moved, rotated or scaled while this mode is open, and a
-    // stored copy would quietly describe the shape it used to be. The expensive part - finding the
-    // paint patch's centroid - stays cached in the local members.
-    bool adjust_anchor_in_frame(Vec3f &pos, Vec3f &normal) const;
-    // Drops every scrap of live gizmo state. Called from each of the several ways out of the mode -
-    // the panel's toggle, collapsing the card, switching to seam marking, closing the gizmo - so a
-    // stale hover highlight or a half-finished drag cannot survive into the next time it is opened.
-    void end_adjust_texture_mode();
     bool on_mouse_adjust_texture(const wxMouseEvent &mouse_event);
     void render_adjust_texture_gizmo();
-    // This frame's gizmo geometry, or an invalid frame when there is nothing to draw.
-    AdjustFrame adjust_frame() const;
-    // Which handle is under `mouse`, with `index` set to the arm or corner it belongs to. Tested
-    // outermost first; see the hit-area board of the design for the order and the radii.
-    AdjustHandle adjust_pick(const Vec2d &mouse, const AdjustFrame &f, int &index) const;
-    // Chooses which corner carries the rotate ring. The nearest to the cursor, but it only moves once
-    // the newcomer is clearly nearer, because without that margin the ring flits between two corners
-    // every time the cursor wanders along the line equidistant from both. Never moves during a drag.
-    void update_rotate_corner(const AdjustFrame &f, const Vec2d &mouse);
-
-    // Where the tile's centre sits, in planar mm relative to the anchor, for a layer's placement.
-    // Not the patch centroid: the gizmo *is* the texture's placement, so it has to travel as the
-    // offset changes - pinning it to the centroid is what made the old handle look broken, the
-    // texture sliding while the handle stayed put.
-    static Vec2f adjust_handle_planar(const TextureDisplacementLayer &layer, float aspect);
-    // The exact inverse: the offset that puts the tile's centre at `planar`. Rotating or resizing
-    // the tile changes where a given offset places it, so every drag recomputes the offset from the
-    // centre it wants to keep - which is what makes rotate and scale act *in place* rather than
-    // swinging the tile around the anchor.
-    static Vec2f adjust_offset_for_planar(const Vec2f &planar, float rot_deg, const Vec2f &tile_mm,
-                                          const Vec2f &mirror_sign, float aspect);
-    // The mesh-local-to-world transform of the selected instance, the (unrotated) tangent basis and
-    // the lifted anchor the whole gizmo is built on. False when there is nothing to build on - which
-    // is the single place that decides the gizmo is not drawable, so the drawing, the hit test and
-    // the drag cannot disagree about it.
-    bool adjust_basis(Transform3d &trafo, Vec3f &u_axis, Vec3f &v_axis, Vec3f &base) const;
-    // Screen-space quad of a tile with the given placement. Used for the pre-drag ghost, which is
-    // drawn from the captured start placement rather than from the layer, the layer having already
-    // been moved by then.
-    bool adjust_quad_for(const Vec2f &center_planar, float rot_deg, const Vec2f &tile_mm,
-                         std::array<Vec2d, 4> &out) const;
-    // Where the cursor meets the drag plane, in planar mm relative to the anchor. The plane is
-    // deliberately the *fixed* anchor plane rather than the tile's current position: measuring a
-    // drag against a plane the drag itself moves is a feedback loop.
-    bool adjust_planar_under(const Vec2d &mouse, Vec2f &planar) const;
     // Draws a small '+'/'-' next to the mouse over the 3D view while painting/selecting, so it is
     // obvious whether the next stroke adds paint (default) or erases it (Shift). Uses ImGui's
     // foreground draw list, so it must be called from inside the gizmo's ImGui frame.
     void render_paint_cursor_hint();
-    // Bake-frame tangent-plane basis at m_adjust_anchor_normal: box_projection_axes(), so dragging
+    // Mesh-local tangent-plane basis at m_adjust_anchor_normal: box_projection_axes(), so dragging
     // on-canvas maps consistently onto offset whichever face the anchor is on.
     void adjust_tangent_basis(Vec3f &u_axis, Vec3f &v_axis) const;
 
@@ -1157,6 +1173,8 @@ private:
 
     // Library textures the picker has shown at least once, keyed by file path (see LibraryTexture).
     std::map<std::string, LibraryTexture> m_library_textures;
+    // Generation of the library the above were cached against; a change clears them.
+    uint64_t                              m_library_generation = 0;
 
     // Everything the *unwrap* depends on. update_uv_editor() runs from rebuild_preview(), i.e. on
     // every stroke end and every slider release - but depth/tiling/rotation/offset/blend change
@@ -1198,12 +1216,15 @@ private:
     // The config key behind m_undocked.
     static constexpr const char *UNDOCKED_CONFIG_KEY = "texture_displacement_panel_undocked";
 
-    // Smooth scrolling for the panel body (everything between the header and the pinned Bake footer).
-    // ImGui jumps a fixed number of lines per wheel notch, which on tall layer cards reads as a hard
-    // jolt rather than a scroll. The wheel is intercepted (ImGuiWindowFlags_NoScrollWithMouse) and
-    // moves a *target* offset instead; the real scroll is eased toward it over the following frames.
+    // Smooth scrolling for a child window, called inside it: the panel body (everything between the header
+    // and the pinned Bake footer) and the texture library's grid. ImGui jumps a fixed number of lines per
+    // wheel notch, which on tall layer cards reads as a hard jolt rather than a scroll. The child sets
+    // ImGuiWindowFlags_NoScrollWithMouse, and the wheel moves `target` instead; the real scroll is eased
+    // toward it over the following frames. `applied` is what the easing wrote last frame, <0 until the
+    // first one.
+    void  ease_wheel_scroll(float &target, float &applied);
     float m_panel_scroll_target  = 0.f;
-    float m_panel_scroll_applied = -1.f; // what the easing wrote last frame; <0 until the first one
+    float m_panel_scroll_applied = -1.f;
     // Last frame's body content height and footer height. The body is a child window that has to be
     // given its height before its content is laid out, so it is sized from the previous frame: as tall
     // as its content, capped so the footer still fits above the bottom of the canvas.
@@ -1213,47 +1234,23 @@ private:
     // See the "Adjust Texture" block of private methods above.
     bool  m_adjust_texture_mode      = false;
     bool  m_adjust_anchor_valid      = false;
-    // Mesh-local, as compute_layer_paint_anchor() produces them. Everything else about the gizmo
-    // works in the bake frame - world orientation and scale, origin at the volume's own, which is
-    // where the texture is projected (texture_displacement_bake_frame()) - so these go through
-    // adjust_anchor_in_frame() at the point of use. A tile is so many *printed* millimetres whatever
-    // the instance's scale, which is why measuring the gizmo in mesh-local millimetres made its tile
-    // differ from the rendered one by exactly that scale.
-    Vec3f m_adjust_anchor_pos        = Vec3f::Zero();
-    Vec3f m_adjust_anchor_normal     = Vec3f::UnitZ();
+    Vec3f m_adjust_anchor_pos        = Vec3f::Zero();  // mesh-local
+    Vec3f m_adjust_anchor_normal     = Vec3f::UnitZ(); // mesh-local
 
-    // What the cursor is over, and what is being dragged. `index` picks the arm or the corner
-    // (0..3, in AdjustFrame's order); it is -1 for Move and Rotate, which have only one of each.
-    // Only ever one of each is set, so two handles can never both look live.
-    // Which corner currently carries the rotate ring. Persists between frames, which is what gives
-    // update_rotate_corner() something to apply its margin against.
-    int          m_adjust_rotate_corner = 0;
-    AdjustHandle m_adjust_hover_kind  = AdjustHandle::None;
-    int          m_adjust_hover_index = -1;
-    AdjustHandle m_adjust_drag_kind   = AdjustHandle::None;
-    int          m_adjust_drag_index  = -1;
+    // Pan: free drag anywhere on the flat panel, moves offset along both axes. AxisU/AxisV: drag
+    // the corresponding arrow, moves offset along only that one axis.
+    enum class AdjustHandle { None, Pan, AxisU, AxisV };
+    AdjustHandle m_adjust_drag_handle       = AdjustHandle::None;
+    Vec2f        m_adjust_drag_start_offset = Vec2f::Zero();
+    // Anchor-relative planar position (see project_planar()) of the point under the mouse at the
+    // moment the current drag started; every subsequent frame's delta is measured against this,
+    // rather than accumulated frame-to-frame, to avoid drift.
+    Vec2f m_adjust_drag_start_planar = Vec2f::Zero();
 
-    // Everything the live drag is measured against, captured once on mouse-down. Every frame's
-    // delta is taken from these rather than accumulated frame to frame, so a long drag cannot
-    // drift away from the cursor.
-    Vec2f m_adjust_drag_start_center = Vec2f::Zero(); // tile centre, planar mm
-    Vec2f m_adjust_drag_start_grab   = Vec2f::Zero(); // the grabbed point, planar mm
-    // How far the grab landed from the handle's own position, in planar mm. A scale drag reads the
-    // extents straight off the cursor, so without subtracting this the tile would jump by however
-    // far inside the generous hit radius the click happened to fall.
-    Vec2f m_adjust_drag_grab_bias    = Vec2f::Zero();
-    float m_adjust_drag_start_rot    = 0.f;
-    Vec2f m_adjust_drag_start_tile   = Vec2f::Ones(); // (u, v) extents as stored
-    float m_adjust_drag_start_angle  = 0.f;           // planar angle of the grab, radians
-    // The mirroring the layer had when the drag began. A width or height handle dragged back through
-    // the tile's middle toggles its axis against this, rather than against whatever the last frame
-    // left behind, so dragging back and forth across the middle cannot ratchet the flag.
-    bool  m_adjust_drag_start_mirror_u = false;
-    bool  m_adjust_drag_start_mirror_v = false;
-
-    // The live value to put in the readout pill, built by the drag and drawn by the renderer in the
-    // same frame. Empty when no drag is running, which is also when no pill is drawn.
-    std::string m_adjust_readout;
+    // Lazily-built unit quad (the pan panel) and unit line-with-arrowhead (reused, rotated, for
+    // both the U and V axis arrows), transformed into place at render time.
+    GLModel m_adjust_panel_glmodel;
+    GLModel m_adjust_arrow_glmodel;
 
     std::map<std::string, wxString> m_desc;
 

src/slic3r/GUI/Gizmos/GLGizmoUtils.cpp

@@ -83,11 +83,13 @@ namespace Slic3r::GUI::GLGizmoUtils {
         bool  clicked = toolbar_circular_button(icon, "##wiki_guide_btn", scale);
         if (url && *url && clicked)
             wxLaunchDefaultBrowser(wxString::FromUTF8(url));
+        // ImGuiWrapper::tooltip keeps the text readable on the dark tooltip background whatever text colour the
+        // host window pushed; a negative wrap position keeps the URL on one line.
         if (ImGui::IsItemHovered()){
             if(url && *url)
-                ImGui::SetTooltip("%s\n%s", _u8L("Wiki Guide").c_str(), url);
+                wxGetApp().imgui()->tooltip(_u8L("Wiki Guide") + "\n" + url, -1.f);
             else
-                ImGui::SetTooltip("%s", _u8L("Wiki Guide").c_str());
+                wxGetApp().imgui()->tooltip(_u8L("Wiki Guide"), -1.f);
         }
         return clicked; // for dynamically generated links
     }
@@ -101,9 +103,9 @@ namespace Slic3r::GUI::GLGizmoUtils {
             wxLaunchDefaultBrowser(wxString::FromUTF8(url));
         if (ImGui::IsItemHovered()){
             if(url && *url)
-                ImGui::SetTooltip("%s\n%s", _u8L("Video Guide").c_str(), url);
+                wxGetApp().imgui()->tooltip(_u8L("Video Guide") + "\n" + url, -1.f);
             else
-                ImGui::SetTooltip("%s", _u8L("Video Guide").c_str());
+                wxGetApp().imgui()->tooltip(_u8L("Video Guide"), -1.f);
         }
         return clicked; // for dynamically generated links
     }

src/slic3r/GUI/Gizmos/GizmoObjectManipulation.cpp

@@ -857,11 +857,12 @@ void GizmoObjectManipulation::do_render_move_window(ImGuiWrapper *imgui_wrapper,
     if (render_combo(imgui_wrapper, "", modes, selection_idx, 0, coord_combo_width)) {
         combox_changed = true;
     }
+    ImGuiWrapper::pop_combo_style();
+    // After pop_combo_style(): its light popup background would hide the tooltip's white text.
     if (ImGui::IsItemHovered()) {
         auto tooltip_str = _L("Coordinate system used for transform actions.");
         imgui_wrapper->tooltip(tooltip_str, imgui_wrapper->calc_text_size(tooltip_str).x + 3 * space_size);
     }
-    ImGuiWrapper::pop_combo_style();
 
     // ORCA use TextColored to match axes color
     float offset_to_center = (unit_size - ImGui::CalcTextSize("O").x) / 2;
@@ -964,11 +965,12 @@ void GizmoObjectManipulation::do_render_move_window(ImGuiWrapper *imgui_wrapper,
         if (render_combo(imgui_wrapper, "", align_modes, align_mode_idx, 0, coord_combo_width))
             m_align_to_parent = align_mode_idx + 1 == align_modes.size();
         ImGui::PopID();
+        ImGuiWrapper::pop_combo_style();
+        // After pop_combo_style(): its light popup background would hide the tooltip's white text.
         if (ImGui::IsItemHovered()) {
             wxString tooltip_str = _L("Reference for alignment");
             imgui_wrapper->tooltip(tooltip_str, imgui_wrapper->calc_text_size(tooltip_str).x + 3 * space_size);
         }
-        ImGuiWrapper::pop_combo_style();
 
         const bool dark_mode    = wxGetApp().dark_mode();
         int        current_axis = -1;
@@ -1377,11 +1379,12 @@ void GizmoObjectManipulation::do_render_scale_input_window(ImGuiWrapper* imgui_w
     if (render_combo(imgui_wrapper, "", modes, selection_idx, 0, coord_combo_width)) {
         combox_changed = true;
     }
+    ImGuiWrapper::pop_combo_style();
+    // After pop_combo_style(): its light popup background would hide the tooltip's white text.
     if (ImGui::IsItemHovered()) {
         auto tooltip_str = _L("Coordinate system used for transform actions.");
         imgui_wrapper->tooltip(tooltip_str, imgui_wrapper->calc_text_size(tooltip_str).x + 3 * space_size);
     }
-    ImGuiWrapper::pop_combo_style();
 
     // ORCA use TextColored to match axes color
     float offset_to_center = (unit_size - ImGui::CalcTextSize("O").x) / 2;

src/slic3r/GUI/ImGuiWrapper.cpp

@@ -505,10 +505,12 @@ bool ImGuiWrapper::update_mouse_data(wxMouseEvent& evt)
     io.MouseDown[0] = evt.LeftIsDown();
     io.MouseDown[1] = evt.RightIsDown();
     io.MouseDown[2] = evt.MiddleIsDown();
-    float wheel_delta = static_cast<float>(evt.GetWheelDelta());
-    if (wheel_delta != 0.0f) {
+    // Only a vertical wheel event with a direction is a wheel step. A macOS trackpad reports sideways motion
+    // as separate horizontal events, and its wheel events often carry a rotation of 0: wx truncates the
+    // sub-pixel deltas to whole ones and passes on events with no delta at all. Read as vertical steps,
+    // either kind scrolls against the gesture, and scrolling up bounces back and forth.
+    if (evt.GetWheelAxis() == wxMOUSE_WHEEL_VERTICAL && evt.GetWheelRotation() != 0)
         io.MouseWheel = evt.GetWheelRotation() > 0 ? 1.f : -1.f;
-    }
     unsigned buttons = (evt.LeftIsDown() ? 1 : 0) | (evt.RightIsDown() ? 2 : 0) | (evt.MiddleIsDown() ? 4 : 0);
     m_mouse_buttons = buttons;
 

src/slic3r/GUI/MainFrame.cpp

@@ -1739,6 +1739,9 @@ void MainFrame::fit_tab_labels()
     for (size_t i = count - 1; i >= 1; --i) {
         int right = ScreenToClient(m_slice_option_btn->ClientToScreen({})).x;
         int left  = sizer->GetSize().GetWidth();
+        // The overflow button sits between the tab buttons and the side tools, so it is not free space.
+        if (const wxWindow* overflow = ctrl->GetOverflowButton())
+            left += overflow->GetSize().GetWidth();
         if (right - left - FromDIP(15) > 0) return;
         ctrl->SetCompact(i, true);
         m_tabpanel->Refresh();

src/slic3r/GUI/Notebook.hpp

@@ -57,6 +57,7 @@ public:
     // ORCA: a companion widget shown right after the tab buttons (before any side_tools), e.g.
     // an overflow indicator. Pass nullptr to remove it; ownership stays with the caller.
     void SetOverflowButton(wxWindow* button);
+    wxWindow* GetOverflowButton() const { return m_overflow_button; }
 
 private:
     wxFlexGridSizer*                m_buttons_sizer;

src/slic3r/GUI/PublishSettingsDialog.cpp

@@ -2225,8 +2225,10 @@ void PublishSettingsDialog::refresh_tab_indicators()
             if (section.variant_switch != nullptr) {
                 const wxString& title = m_categories[section.categories[i]].title;
                 const wxString  text  = on ? title + wxString(" ") + wxString(wxUniChar(0x2022)) : title;
-                if (section.variant_switch->GetOptionText(static_cast<unsigned int>(i)) != text)
+                if (section.variant_switch->GetOptionText(static_cast<unsigned int>(i)) != text) {
                     section.variant_switch->SetOptionText(static_cast<unsigned int>(i), text);
+                    section.page->Layout();
+                }
             }
             any = any || on;
         }

src/slic3r/GUI/SequentialPrintClearance.hpp

@@ -0,0 +1,63 @@
+#pragma once
+
+#include "libslic3r/BoundingBox.hpp"
+#include "libslic3r/ObjectID.hpp"
+#include "libslic3r/Polygon.hpp"
+
+#include <algorithm>
+#include <cstddef>
+#include <map>
+#include <utility>
+#include <vector>
+
+namespace Slic3r::GUI {
+
+struct SequentialClearanceInstance
+{
+    double      instance_height;
+    BoundingBox bounding_box;
+    Polygon     hull_polygon;
+    ObjectID    instance_id;
+};
+
+// PrintApply may group rotated copies in a different order from ModelObject::instances.
+// Reorder one object's instances only when validation has assigned all of them an order.
+inline void sort_sequential_clearance_instances(
+    std::vector<SequentialClearanceInstance>::iterator first, std::vector<SequentialClearanceInstance>::iterator last,
+    const std::map<ObjectID, int>& print_order)
+{
+    if (!std::all_of(first, last, [&print_order](const SequentialClearanceInstance& instance) {
+            const auto order = print_order.find(instance.instance_id);
+            return order != print_order.end() && order->second > 0;
+        }))
+        return;
+
+    std::stable_sort(first, last, [&print_order](const SequentialClearanceInstance& lhs, const SequentialClearanceInstance& rhs) {
+        return print_order.at(lhs.instance_id) < print_order.at(rhs.instance_id);
+    });
+}
+
+// Input follows object-list order with validated print order within each object.
+inline std::vector<std::pair<Polygon, float>> sequential_clearance_height_polygons(
+    const std::vector<SequentialClearanceInstance>& instances, double printable_height, double height_to_lid, double height_to_rod)
+{
+    std::vector<std::pair<Polygon, float>> height_polygons;
+    height_polygons.reserve(instances.size());
+    for (size_t k = 0; k < instances.size(); ++k) {
+        const auto& instance = instances[k];
+        double height = k + 1 == instances.size() ? printable_height : height_to_lid;
+        for (size_t i = k + 1; i < instances.size(); ++i) {
+            const auto& next_bbox = instances[i].bounding_box;
+            if (std::min(instance.bounding_box.max.y(), next_bbox.max.y()) >
+                std::max(instance.bounding_box.min.y(), next_bbox.min.y())) {
+                height = height_to_rod;
+                break;
+            }
+        }
+        if (height < instance.instance_height)
+            height_polygons.emplace_back(instance.hull_polygon, float(height));
+    }
+    return height_polygons;
+}
+
+} // namespace Slic3r::GUI

src/slic3r/GUI/Tab.cpp

@@ -4538,6 +4538,7 @@ void TabFilament::build()
         optgroup->append_single_option_line("filament_adhesiveness_category", "material_basic_information#adhesiveness-category");
 
         optgroup->append_single_option_line("filament_density", "material_basic_information#density");
+        optgroup->append_single_option_line("filament_transmission_distance", "material_basic_information#transmission-distance");
         optgroup->append_single_option_line("filament_shrink", "material_basic_information#shrinkage-xy");
         optgroup->append_single_option_line("filament_shrinkage_compensation_z", "material_basic_information#shrinkage-z");
         optgroup->append_single_option_line("filament_cost", "material_basic_information#price");

src/slic3r/GUI/TextureLibrary.cpp

@@ -1,6 +1,7 @@
 #include "TextureLibrary.hpp"
 
 #include <algorithm>
+#include <atomic>
 #include <boost/filesystem/operations.hpp>
 #include <cstddef>
 #include <boost/filesystem/directory.hpp>
@@ -21,6 +22,7 @@
 #include <boost/system/error_code.hpp>
 
 #include <vector>
+#include <wx/app.h>
 #include <wx/gdicmn.h>
 #include <utility>
 #include <wx/image.h>
@@ -36,6 +38,9 @@
 #include "slic3r/GUI/GUI_App.hpp"
 #include "slic3r/GUI/I18N.hpp"
 
+#include "slic3r/plugin/PluginFsUtils.hpp"
+#include "slic3r/plugin/host/PluginResourceTextures.hpp"
+
 namespace Slic3r::GUI {
 
 namespace {
@@ -50,7 +55,7 @@ bool is_image_file(const boost::filesystem::path &path)
     return ext == ".png" || ext == ".jpg" || ext == ".jpeg" || ext == ".bmp";
 }
 
-void scan_dir(const boost::filesystem::path &dir, bool is_user, std::vector<TextureLibraryEntry> &out)
+void scan_dir(const boost::filesystem::path &dir, TextureSource source, std::vector<TextureLibraryEntry> &out)
 {
     boost::system::error_code ec;
     if (!boost::filesystem::is_directory(dir, ec))
@@ -60,12 +65,43 @@ void scan_dir(const boost::filesystem::path &dir, bool is_user, std::vector<Text
     for (boost::filesystem::directory_iterator it(dir, ec), end; it != end && !ec; it.increment(ec)) {
         if (!boost::filesystem::is_regular_file(it->path(), ec) || !is_image_file(it->path()))
             continue;
-        out.push_back({ it->path().stem().string(), it->path().string(), is_user });
+        out.push_back({ it->path().stem().string(), it->path().string(), source });
     }
     std::sort(out.begin() + first, out.end(),
               [](const TextureLibraryEntry &a, const TextureLibraryEntry &b) { return a.name < b.name; });
 }
 
+// Appends every loaded pack's textures, ordered by the pack's display name and then by file name. Each
+// entry carries the contributing pack's display name, which the picker uses to group a pack's textures
+// into one section; a pack name is unique among loaded packs of this capability type, so the name both
+// groups the section and keeps its entries contiguous. A pack whose directory fails validation is
+// skipped, so one bad pack never hides the others.
+void scan_plugin_textures(std::vector<TextureLibraryEntry> &out)
+{
+    std::vector<TextureLibraryEntry> found;
+    for (const ResourceTextureEntry &pack : PluginResourceTextures::instance().entries()) {
+        std::string                    error;
+        const std::vector<std::string> files =
+            collect_resource_texture_files(pack.texture_dir, pack.plugin_root, error);
+        if (!error.empty()) {
+            BOOST_LOG_TRIVIAL(warning) << "Texture pack '" << pack.display_name << "' was skipped: " << error;
+            continue;
+        }
+        for (const std::string &file : files) {
+            const boost::filesystem::path path(file);
+            found.push_back({ path.stem().string(), file, TextureSource::Plugin, pack.display_name });
+        }
+    }
+
+    std::stable_sort(found.begin(), found.end(), [](const TextureLibraryEntry &a, const TextureLibraryEntry &b) {
+        if (a.pack_name != b.pack_name)
+            return a.pack_name < b.pack_name;
+        return a.name < b.name;
+    });
+    for (TextureLibraryEntry &entry : found)
+        out.push_back(std::move(entry));
+}
+
 // True if the image carries any colour at all, i.e. some pixel's channels are not all equal. A
 // photo saved as RGB but actually grey should still be stored as a grayscale PNG - it is a third of
 // the bytes, and the colour feature keys off has_color(), so storing a grey image as RGB would offer
@@ -187,27 +223,20 @@ std::vector<unsigned char> read_file_bytes(const std::string &path)
     return std::vector<unsigned char>(std::istreambuf_iterator<char>(ifs), std::istreambuf_iterator<char>());
 }
 
-// True if these bytes are a PNG libslic3r can decode, i.e. can be stored on a layer as-is. Mirrors
-// exactly what decode_height_texture() accepts: 8-bit grayscale, or 8-bit colour (whose luminance is
-// the height and whose RGB is available to colour the model). Anything else is converted on load.
+// True if decode_height_texture() can read these bytes, so they can be stored on a layer as-is.
+// Anything else is converted on load.
 bool is_supported_height_map(const std::vector<unsigned char> &bytes)
 {
-    if (bytes.empty())
-        return false;
-    const png::ReadBuf rbuf{ bytes.data(), bytes.size() };
-    if (!png::is_png(rbuf))
-        return false;
-    png::ImageGreyscale gray;
-    if (png::decode_png(rbuf, gray) && gray.cols > 0 && gray.rows > 0)
-        return true;
-    png::ImageColorscale color;
-    return png::decode_colored_png(rbuf, color) && color.cols > 0 && color.rows > 0 &&
-           color.bytes_per_pixel >= 3 &&
-           color.buf.size() == color.cols * color.rows * size_t(color.bytes_per_pixel);
+    return !bytes.empty() && png::is_decodable_8bit_png(png::ReadBuf{ bytes.data(), bytes.size() });
 }
 
 std::vector<TextureLibraryEntry> g_library;
 bool                             g_library_scanned = false;
+// The generation g_library was last built from, so texture_library() can rescan after a pack change.
+uint64_t g_library_scan_generation = 0;
+
+// Bumped by plugin callbacks, read by the GUI frame. Atomic so a read never races a write.
+std::atomic<uint64_t> g_library_generation{ 0 };
 
 // Writes `bytes` into the user's own texture folder as `stem`.png, uniquified so an existing texture -
 // shipped-name collision, or a second edit of the same source - is never overwritten. Returns the new
@@ -255,6 +284,8 @@ const size_t      RECENT_MAX     = 200;
 std::vector<std::string> recent_texture_names()
 {
     std::vector<std::string> names;
+    if (wxTheApp == nullptr) // headless or shutting down: there is no app config to read
+        return names;
     if (wxGetApp().app_config == nullptr)
         return names;
     const std::string value = wxGetApp().app_config->get(RECENT_SECTION, RECENT_KEY);
@@ -269,6 +300,8 @@ std::vector<std::string> recent_texture_names()
 
 void store_recent_texture_names(std::vector<std::string> names)
 {
+    if (wxTheApp == nullptr) // headless or shutting down: there is no app config to write
+        return;
     if (wxGetApp().app_config == nullptr)
         return;
     if (names.size() > RECENT_MAX)
@@ -279,8 +312,9 @@ void store_recent_texture_names(std::vector<std::string> names)
     wxGetApp().app_config->set(RECENT_SECTION, RECENT_KEY, value);
 }
 
-// Reorders the user's entries (the tail of `library`) by recent use. Stable, so textures never used keep
-// the name order scan_dir() gave them, after every used one.
+// Reorders the user's entries by recent use. Stable, so textures never used keep the name order
+// scan_dir() gave them, after every used one. Only the user's own range is touched: shipped and pack
+// entries must keep the order their scans produced.
 void sort_user_textures(std::vector<TextureLibraryEntry> &library)
 {
     const std::vector<std::string>       names = recent_texture_names();
@@ -291,8 +325,11 @@ void sort_user_textures(std::vector<TextureLibraryEntry> &library)
         const auto it = rank.find(boost::filesystem::path(e.path).filename().string());
         return it == rank.end() ? std::numeric_limits<int>::max() : it->second;
     };
-    const auto first_user = std::find_if(library.begin(), library.end(), [](const TextureLibraryEntry &e) { return e.is_user; });
-    std::stable_sort(first_user, library.end(),
+    const auto first_user = std::find_if(library.begin(), library.end(),
+                                         [](const TextureLibraryEntry &e) { return e.source == TextureSource::User; });
+    const auto user_end   = std::find_if(first_user, library.end(),
+                                         [](const TextureLibraryEntry &e) { return e.source != TextureSource::User; });
+    std::stable_sort(first_user, user_end,
                      [&rank_of](const TextureLibraryEntry &a, const TextureLibraryEntry &b) { return rank_of(a) < rank_of(b); });
 }
 
@@ -311,6 +348,16 @@ bool in_user_texture_dir(const std::string &path)
 
 } // namespace
 
+uint64_t texture_library_generation()
+{
+    return g_library_generation.load();
+}
+
+void bump_texture_library_generation()
+{
+    g_library_generation.fetch_add(1);
+}
+
 std::string user_texture_dir()
 {
     const boost::filesystem::path dir = boost::filesystem::path(Slic3r::data_dir()) / "textures" / "displacement";
@@ -325,17 +372,20 @@ std::string user_texture_dir()
 
 const std::vector<TextureLibraryEntry> &texture_library(bool force_rescan)
 {
-    if (g_library_scanned && !force_rescan)
+    const uint64_t generation = texture_library_generation();
+    if (g_library_scanned && !force_rescan && generation == g_library_scan_generation)
         return g_library;
 
     g_library.clear();
-    scan_dir(boost::filesystem::path(Slic3r::resources_dir()) / "textures" / "displacement", false, g_library);
+    scan_dir(boost::filesystem::path(Slic3r::resources_dir()) / "textures" / "displacement", TextureSource::Shipped, g_library);
     const std::string user_dir = user_texture_dir();
     if (!user_dir.empty())
-        scan_dir(boost::filesystem::path(user_dir), true, g_library);
+        scan_dir(boost::filesystem::path(user_dir), TextureSource::User, g_library);
     sort_user_textures(g_library);
+    scan_plugin_textures(g_library);
 
-    g_library_scanned = true;
+    g_library_scan_generation = generation;
+    g_library_scanned          = true;
     return g_library;
 }
 

src/slic3r/GUI/TextureLibrary.hpp

@@ -1,6 +1,7 @@
 #ifndef slic3r_TextureLibrary_hpp_
 #define slic3r_TextureLibrary_hpp_
 
+#include <cstdint>
 #include <memory>
 #include <optional>
 #include <string>
@@ -10,26 +11,45 @@
 
 namespace Slic3r::GUI {
 
-// One selectable height-map texture in the texture-displacement gizmo's texture picker.
+// Where a library texture came from. The picker lists each source under its own heading.
+enum class TextureSource { Shipped, User, Plugin };
+
+// One selectable height-map texture in the texture-displacement gizmo's texture picker. A texture that
+// came from a plugin pack also carries the provenance the picker groups its sections by.
 struct TextureLibraryEntry
 {
-    std::string name;    // display name (the file's stem, e.g. "Wood Grain")
-    std::string path;    // absolute path on disk
-    bool        is_user; // imported by the user, as opposed to shipped with OrcaSlicer
+    std::string   name;   // display name (the file's stem, e.g. "Wood Grain")
+    std::string   path;   // absolute path on disk
+    TextureSource source; // which of the three sources it came from
+
+    std::string pack_name; // contributing pack's display name; empty for Shipped and User
 };
 
-// Every available height-map texture: the ones shipped in resources/textures/displacement first, sorted
-// by name, then the user's own from <data_dir>/textures/displacement - most recently used first (see
-// touch_user_texture()), then any never used, by name.
+// Every available height-map texture, in the order the picker lists them: the ones shipped in
+// resources/textures/displacement first, by name; then the user's own from <data_dir>/textures/displacement,
+// most recently used first (see touch_user_texture()) and the rest by name; then the loaded texture
+// packs, ordered by pack display name and then by file name.
 //
-// The two live in separate directories deliberately: an app update replaces the resources tree
-// wholesale, so anything the user imported has to sit somewhere that update can never overwrite or
-// delete. `is_user` is what the picker uses to show them under separate headings.
+// The shipped and user textures live in separate directories deliberately: an app update replaces the
+// resources tree wholesale, so anything the user imported has to sit somewhere that update can never
+// overwrite or delete. Pack textures are read-only like the shipped ones. Every texture carries its
+// `source`; a pack texture also carries `pack_name`, and the picker groups pack textures into one
+// collapsible section per pack name. A pack's name is unique among loaded packs of this capability type -
+// the loader refuses a collision with "Capability collision for type ... and name ...", see
+// PluginManager::check_registry_locked (PluginManager.cpp:211-225) - so the name on its own identifies a
+// section and keeps that pack's entries contiguous.
 //
 // Scanned once and cached. Pass force_rescan after an import, or to pick up a file the user dropped
-// into either folder by hand while the app was running.
+// into either folder by hand while the app was running. A rescan also happens on its own when the pack
+// registry changes, which is what texture_library_generation() tracks.
 const std::vector<TextureLibraryEntry> &texture_library(bool force_rescan = false);
 
+// Monotonic counter, bumped whenever the set of loaded texture packs changes. The GUI reads it to
+// decide whether texture_library() must rescan, and plugin callbacks bump it. Safe to read from any
+// thread at any time.
+uint64_t texture_library_generation();
+void     bump_texture_library_generation();
+
 // Moves one of the user's own textures to the front of the recently-used order texture_library() lists
 // them in, and stores that order in the app config so it survives a restart. No-op for anything that is
 // not in user_texture_dir().

src/slic3r/GUI/UVEditorCanvas.cpp

@@ -297,12 +297,9 @@ void UVEditorCanvas::set_island_fill_colors(std::vector<ColorRGBA> colors)
     Refresh();
 }
 
-void UVEditorCanvas::set_uv_transform(float tiling_scale, float tiling_scale_v, const Vec2f &mirror_sign,
-                                      float rotation_deg, bool tile_enabled, bool tile_mirrored)
+void UVEditorCanvas::set_uv_transform(float tiling_scale, float rotation_deg, bool tile_enabled, bool tile_mirrored)
 {
     m_tiling_scale          = (std::abs(tiling_scale) > 1e-6f) ? tiling_scale : 1.f;
-    m_tiling_scale_v        = (tiling_scale_v > 1e-6f) ? tiling_scale_v : m_tiling_scale;
-    m_mirror_sign           = mirror_sign;
     m_rotation_deg          = rotation_deg;
     m_tile_enabled          = tile_enabled;
     m_tile_mirrored         = tile_mirrored;
@@ -664,18 +661,11 @@ Vec2f UVEditorCanvas::island_centroid(int island) const
 
 Vec2f UVEditorCanvas::uv_delta_to_unwrap(const Vec2f &delta_uv) const
 {
-    // apply_uv_transform() maps unwrap -> uv as  uv = S^-1 R(unwrap) + offset, with S the tile
-    // extents, so the inverse of a *delta* (the translation drops out) is  unwrap = R^-1(S delta_uv).
-    // The extents therefore multiply *before* the inverse rotation, which is the mirror image of the
-    // forward order; with a square tile the two are interchangeable, with a rectangular one they are
-    // not.
+    // apply_uv_transform() maps unwrap -> uv as  uv = R(unwrap / tiling) + offset, so the inverse of
+    // a *delta* (the translation drops out) is  unwrap = tiling * R^-1(delta_uv).
     const float rad = m_rotation_deg * float(M_PI) / 180.f;
     const float cs = std::cos(rad), sn = std::sin(rad);
-    // The mirror is its own sign on the same axis, and a negation is its own inverse, so it comes
-    // back out here exactly as it went in.
-    const Vec2f scaled(delta_uv.x() * m_tiling_scale * m_mirror_sign.x(),
-                       delta_uv.y() * m_tiling_scale_v * m_mirror_sign.y());
-    return Vec2f(scaled.x() * cs + scaled.y() * sn, -scaled.x() * sn + scaled.y() * cs);
+    return Vec2f(delta_uv.x() * cs + delta_uv.y() * sn, -delta_uv.x() * sn + delta_uv.y() * cs) * m_tiling_scale;
 }
 
 Vec2f UVEditorCanvas::snap_correction(int island) const
@@ -2190,10 +2180,7 @@ UVEditorPanel::UVEditorPanel(wxWindow *parent) : wxPanel(parent, wxID_ANY)
 void UVEditorPanel::apply_state(const UVEditorCanvas::PaneState &s)
 {
     const wxString name = s.has_layer ? s.layer_name : _L("No layer mapped with Unwrap");
-    // A square tile reads as one number; only a genuinely rectangular one is worth two.
-    const wxString tile = !s.has_layer                         ? wxString() :
-                          std::abs(s.tile_mm_v - s.tile_mm) < 0.005f ? wxString::Format(_L("%.1f mm tile"), s.tile_mm) :
-                          wxString::Format(_L("%.1f x %.1f mm tile"), s.tile_mm, s.tile_mm_v);
+    const wxString tile = s.has_layer ? wxString::Format(_L("%.1f mm tile"), s.tile_mm) : wxString();
     const bool     relayout = m_layer_name->GetLabel() != name || m_tile->GetLabel() != tile;
     m_layer_name->SetLabel(name);
     m_tile->SetLabel(tile);

src/slic3r/GUI/UVEditorCanvas.hpp

@@ -91,10 +91,7 @@ public:
     // space, back into the unwrap's mm space - which is where a TextureIsland's offset actually
     // lives (see apply_uv_transform()). The tile settings come along because the background has to
     // repeat exactly the way the height sampler does, or the pane would stop showing what gets baked.
-    // `tiling_scale_v` is the tile's extent along v; pass <= 0 for a square tile, matching
-    // layer_tile_mm().
-    void set_uv_transform(float tiling_scale, float tiling_scale_v, const Vec2f &mirror_sign,
-                          float rotation_deg, bool tile_enabled, bool tile_mirrored);
+    void set_uv_transform(float tiling_scale, float rotation_deg, bool tile_enabled, bool tile_mirrored);
 
     // Same 8-bit grayscale pixels build_texture_displacement()'s height sampling uses, shown
     // beneath the wireframe (expanded to RGBA on upload) so the unwrap can be checked against the
@@ -132,8 +129,7 @@ public:
     {
         bool                       has_layer       = false; // an active layer mapped with Unwrap
         wxString                   layer_name;
-        float                      tile_mm         = 0.f; // tile extent along u
-        float                      tile_mm_v       = 0.f; // along v; equal to tile_mm for a square tile
+        float                      tile_mm         = 0.f;
         std::vector<unsigned char> thumbnail_rgb;           // thumbnail_px square, 3 bytes per pixel; empty for none
         int                        thumbnail_px    = 0;
         float                      seam_angle_deg  = 40.f;
@@ -296,10 +292,8 @@ private:
     GLModel m_grid_glmodel;
     float   m_grid_step = 0.f; // the UV step m_grid_glmodel was built for; 0 = not built
 
-    float m_tiling_scale   = 1.f;
-    float m_tiling_scale_v = 1.f; // already resolved: never the "unset" sentinel, see set_uv_transform()
-    Vec2f m_mirror_sign    = Vec2f::Ones(); // +1 or -1 per axis, see layer_mirror_sign()
-    float m_rotation_deg   = 0.f;
+    float m_tiling_scale  = 1.f;
+    float m_rotation_deg  = 0.f;
     bool  m_tile_enabled  = true;
     bool  m_tile_mirrored = false;
     bool  m_snap_enabled  = false;

src/slic3r/GUI/WebViewDialog.cpp

@@ -537,7 +537,7 @@ void WebViewPanel::cloud_signed_out(const wxString& signout_url)
     }
 }
 
-void WebViewPanel::set_cloud_mode(bool on, const wxString& url)
+void WebViewPanel::set_cloud_mode(bool on, const wxString& url, const wxString& page_url)
 {
     if (on) {
         // load_cloud_url records the origin from whatever it loads, including the plain cloud
@@ -546,6 +546,10 @@ void WebViewPanel::set_cloud_mode(bool on, const wxString& url)
             create_cloud_browser();
             m_cloud_seed_url = url;
             schedule_cloud_load(url.empty() ? wxString(kCloudUrl) : url);
+        } else if (!page_url.empty()) {
+            // The pane already holds a session, so an explicit page loads straight in; the handoff is
+            // only needed to establish one, not to reach an ordinary page.
+            load_cloud_url(page_url);
         } else if (!url.empty() && url != m_cloud_seed_url) {
             // Re-seed only when the Start URL actually changed — a login, an account switch or a
             // sign-out changes it; an ordinary re-open does not (guide §8). Reloading an unchanged
@@ -556,6 +560,10 @@ void WebViewPanel::set_cloud_mode(bool on, const wxString& url)
             load_cloud_after_signout(url);
         }
         m_cloud_browser->Show();
+        // The homepage shell owns the nav highlight and the cloud-mode layout class. A native open
+        // (not the page's own GotoCloud) must ask it to switch; the page's SetCloudMode no-ops if it
+        // already did. OnDocumentLoaded re-applies this when the homepage was still loading.
+        RunScript("SetCloudMode(true)");
     } else if (m_cloud_browser) {
         // Closing keeps the pane alive with its session, so re-opening does not have to sign in
         // again (guide §10). Nothing is cleared here.

src/slic3r/GUI/WebViewDialog.hpp

@@ -118,7 +118,10 @@ public:
     // Shows/hides the native OrcaCloud pane beside this homepage nav column. `url` is the
     // cloud Start URL ({web}/auth/handoff?next=…&user_id=…), which the page uses to decide
     // whether it already has a session; empty falls back to the plain cloud page.
-    void set_cloud_mode(bool on, const wxString& url = wxString());
+    // `page_url` is a full {web} page URL: an already-open pane is navigated straight there,
+    // without re-running the handoff, which is what an explicit "open this page" wants. Empty
+    // keeps whatever page the pane holds.
+    void set_cloud_mode(bool on, const wxString& url = wxString(), const wxString& page_url = wxString());
 
     // True while the OrcaCloud pane exists and is visible.
     bool cloud_mode_active() const { return m_cloud_browser && m_cloud_browser->IsShown(); }

src/slic3r/GUI/Widgets/AMSItem.cpp

@@ -38,6 +38,8 @@
 #include <cstddef>
 #include <wx/simplebook.h>
 #include <wx/dcgraph.h>
+#include <wx/graphics.h>
+#include <memory>
 
 #include <boost/log/trivial.hpp>
 #include <wx/timer.h>
@@ -1409,22 +1411,17 @@ void AMSLib::render_lite_lib(wxDC& dc)
 
     dc.SetPen(wxPen(*wxTRANSPARENT_PEN));
     if (m_info.material_cols.size() > 1) {
-        int left = FromDIP(10);
-        int gwidth = std::round(libsize.x / (m_info.material_cols.size() - 1));
-        //gradient
+        wxRect color_rect(FromDIP(10), FromDIP(10), libsize.x - FromDIP(18), libsize.y - FromDIP(18));
         if (m_info.ctype == 0) {
-            for (int i = 0; i < m_info.material_cols.size() - 1; i++) {
-                auto rect = wxRect(left, FromDIP(10), libsize.x - FromDIP(18), libsize.y - FromDIP(18));
-                fill_gradient_rect_east(dc, rect, m_info.material_cols[i], m_info.material_cols[i + 1]);
-                left += gwidth;
-            }
+            fill_gradient_rect_east(dc, color_rect, m_info.material_cols.front(), m_info.material_cols.back());
         }
         else {
             int cols_size = m_info.material_cols.size();
             for (int i = 0; i < cols_size; i++) {
                 dc.SetBrush(wxBrush(m_info.material_cols[i]));
-                float x = FromDIP(10) + ((float)libsize.x - FromDIP(18)) * i / cols_size;
-                dc.DrawRoundedRectangle(x, FromDIP(10), ((float)libsize.x - FromDIP(17)) / cols_size, libsize.y - FromDIP(20), 0);
+                int x = color_rect.x + color_rect.width * i / cols_size;
+                int next_x = color_rect.x + color_rect.width * (i + 1) / cols_size;
+                dc.DrawRectangle(x, color_rect.y, next_x - x, color_rect.height);
             }
             dc.SetBrush(wxBrush(tmp_lib_colour));
         }
@@ -1516,7 +1513,27 @@ void AMSLib::render_generic_lib(wxDC &dc)
     int top = height - curr_height;
 
     if (m_ams_model == EXT_AMS){
-        dc.DrawRoundedRectangle(FromDIP(1), FromDIP(1), size.x - FromDIP(2), size.y - FromDIP(1), m_radius - 1);
+        wxRect color_rect(FromDIP(1), FromDIP(1), size.x - FromDIP(2), size.y - FromDIP(1));
+        if ((m_info.ctype == 0 || m_info.ctype == 1) && m_info.material_cols.size() > 1 && alpha != 0) {
+            if (m_info.ctype == 0) {
+                fill_gradient_rect_east(dc, color_rect, m_info.material_cols.front(), m_info.material_cols.back());
+            }
+            else {
+                int cols_size = static_cast<int>(m_info.material_cols.size());
+                for (int i = 0; i < cols_size; i++) {
+                    dc.SetPen(wxPen(*wxTRANSPARENT_PEN));
+                    dc.SetBrush(wxBrush(m_info.material_cols[i]));
+                    int x      = color_rect.x + color_rect.width * i / cols_size;
+                    int next_x = color_rect.x + color_rect.width * (i + 1) / cols_size;
+                    dc.DrawRectangle(x, color_rect.y, next_x - x, color_rect.height);
+                }
+            }
+            dc.SetPen(wxPen(*wxTRANSPARENT_PEN));
+            dc.SetBrush(wxBrush(tmp_lib_colour));
+        }
+        else {
+            dc.DrawRoundedRectangle(color_rect.x, color_rect.y, color_rect.width, color_rect.height, m_radius - 1);
+        }
         if (alpha == 0) {
             dc.DrawBitmap(m_bitmap_transparent_def.bmp(), FromDIP(2), FromDIP(2));
         }
@@ -2844,30 +2861,38 @@ void AMSPreview::doRender(wxDC &dc)
         //dc.DrawRoundedRectangle((size.x - rec_size.x) / 2, (size.y - rec_size.y) / 2, rec_size.x, rec_size.y, FromDIP(2));
         if (iter.material_cols.size() > 1)
         {
-            int fleft = (size.x - AMS_ITEM_CUBE_SIZE.x) / 2;
-
-            float total_width = AMS_ITEM_CUBE_SIZE.x;
-            int gwidth = (total_width / (iter.material_cols.size()));
-            if (iter.ctype == 0) {
-                for (int i = 0; i < iter.material_cols.size() - 1; i++) {
-
-                    if ((fleft + gwidth) > (AMS_ITEM_CUBE_SIZE.x)) {
-                        gwidth = (fleft + AMS_ITEM_CUBE_SIZE.x) - fleft;
-                    }
-
-                    auto rect = wxRect(fleft, (size.y - AMS_ITEM_CUBE_SIZE.y) / 2, gwidth, AMS_ITEM_CUBE_SIZE.y);
-                    fill_gradient_rect_east(dc, rect, iter.material_cols[i], iter.material_cols[i + 1]);
-                    fleft += gwidth;
-                }
+            const wxSize swatch_size = m_ams_item_type == AMSModel::N3S_AMS ? AMS_ITEM_CUBE_SIZE : AMS_ITEM_CUBE_SIZE2;
+            const wxRect rect((size.x - swatch_size.x) / 2, (size.y - swatch_size.y) / 2, swatch_size.x, swatch_size.y);
+            const int radius = m_ams_item_type == AMSModel::N3S_AMS ? 0 : FromDIP(3);
+            std::unique_ptr<wxGraphicsContext> owned_gc;
+            wxGraphicsContext* gc = dc.GetGraphicsContext();
+            if (!gc) {
+                owned_gc.reset(wxGraphicsContext::CreateFromUnknownDC(dc));
+                gc = owned_gc.get();
             }
-            else {
-                int cols_size = iter.material_cols.size();
-                for (int i = 0; i < cols_size; i++) {
-                    dc.SetPen(wxPen(*wxTRANSPARENT_PEN));
-                    dc.SetBrush(wxBrush(iter.material_cols[i]));
-                    float x = (size.x - AMS_ITEM_CUBE_SIZE.x) / 2 + total_width * i / cols_size;
-                    dc.DrawRectangle(x, (size.y - AMS_ITEM_CUBE_SIZE.y) / 2, total_width / cols_size, AMS_ITEM_CUBE_SIZE.y);
+            if (gc) {
+                wxGraphicsGradientStops stops(iter.material_cols.front(), iter.material_cols.back());
+                const size_t count = iter.material_cols.size();
+                if (iter.ctype == 0) {
+                    for (size_t i = 1; i + 1 < count; ++i)
+                        stops.Add(iter.material_cols[i], static_cast<float>(i) / (count - 1));
+                } else {
+                    // Two stops at each boundary keep multicolour bands distinct.
+                    for (size_t i = 1; i < count; ++i) {
+                        const float position = static_cast<float>(i) / count;
+                        stops.Add(iter.material_cols[i - 1], position);
+                        stops.Add(iter.material_cols[i], position);
+                    }
                 }
+                gc->PushState();
+                gc->SetPen(*wxTRANSPARENT_PEN);
+                gc->SetBrush(gc->CreateLinearGradientBrush(rect.x, rect.y, rect.x + rect.width, rect.y, stops));
+                gc->DrawRoundedRectangle(rect.x, rect.y, rect.width, rect.height, radius);
+                gc->PopState();
+            } else {
+                dc.SetPen(*wxTRANSPARENT_PEN);
+                dc.SetBrush(iter.material_colour);
+                dc.DrawRoundedRectangle(rect, radius);
             }
         }
         else {

src/slic3r/GUI/Widgets/TabCtrl.cpp

@@ -32,7 +32,7 @@ END_EVENT_TABLE()
  */
 
 #define TAB_BUTTON_SPACE 2
-#define TAB_BUTTON_PADDING_X 2
+#define TAB_BUTTON_PADDING_X 3
 #define TAB_BUTTON_PADDING_Y 2
 #define TAB_BUTTON_PADDING TAB_BUTTON_PADDING_X, TAB_BUTTON_PADDING_Y
 

src/slic3r/GUI/Widgets/WebView.cpp

@@ -74,6 +74,10 @@ webkit_web_view_run_javascript_finish                (WebKitWebView
 						      GError                    **error);
 WEBKIT_API void
 webkit_javascript_result_unref              (WebKitJavascriptResult *js_result);
+WEBKIT_API guint
+webkit_get_major_version                     (void);
+WEBKIT_API guint
+webkit_get_minor_version                     (void);
 }
 #endif
 
@@ -427,6 +431,29 @@ bool WebView::NeedsRecreateOnShow()
     return false;
 #endif
 }
+
+bool WebView::NeedsCompositingWorkaround()
+{
+#ifdef __linux__
+    const guint major = webkit_get_major_version();
+    const guint minor = webkit_get_minor_version();
+    return major < 2 || (major == 2 && minor < 42);
+#else
+    return false;
+#endif
+}
+
+bool WebView::SupportsForceSharedMemory()
+{
+#ifdef __linux__
+    const guint major = webkit_get_major_version();
+    const guint minor = webkit_get_minor_version();
+    return major > 2 || (major == 2 && minor >= 44);
+#else
+    return false;
+#endif
+}
+
 #if wxUSE_WEBVIEW_EDGE
 bool WebView::CheckWebViewRuntime()
 {

src/slic3r/GUI/Widgets/WebView.hpp

@@ -34,6 +34,16 @@ public:
     // ignoring every navigation. A panel that gets true here recreates its view on first Show().
     static bool NeedsRecreateOnShow();
 
+    // True when the linked WebKitGTK predates 2.42, where accelerated compositing
+    // could fail silently on XWayland sessions, so it is forced off as a fallback.
+    // False on every other platform. Safe to call before GTK initialises.
+    static bool NeedsCompositingWorkaround();
+
+    // True when the linked WebKitGTK is new enough (>= 2.44) to honour
+    // WEBKIT_DMABUF_RENDERER_FORCE_SHM, the shared-memory transport switch.
+    // False on every other platform. Safe to call before GTK initialises.
+    static bool SupportsForceSharedMemory();
+
     static void RecreateAll();
 };
 

src/slic3r/plugin/PluginConfig.cpp

@@ -529,6 +529,7 @@ std::string PluginConfig::capability_type_label(PluginCapabilityType type)
     case PluginCapabilityType::Visualization: return _u8L("Visualization");
     case PluginCapabilityType::Script: return _u8L("Script");
     case PluginCapabilityType::SlicingPipeline: return _u8L("Slicing Pipeline");
+    case PluginCapabilityType::ResourceTexture: return _u8L("Resource Texture");
     default: return _u8L("Unknown");
     }
 }

src/slic3r/plugin/PluginFsUtils.cpp

@@ -9,6 +9,7 @@
 #include <boost/filesystem/directory.hpp>
 #include <boost/filesystem/file_status.hpp>
 #include <boost/log/trivial.hpp>
+#include <boost/algorithm/string/case_conv.hpp>
 #include <boost/nowide/fstream.hpp>
 
 #include "PluginAuditManager.hpp"
@@ -33,10 +34,6 @@
 #include <unordered_set>
 #include <vector>
 
-#ifdef WIN32
-#include <boost/locale/encoding_utf.hpp>
-#endif
-
 namespace Slic3r {
 
 const char* const INSTALL_STATE_FILE = ".install_state.json";
@@ -664,7 +661,7 @@ bool extract_zip_to_directory(const boost::filesystem::path& zip_path, const boo
     namespace fs = boost::filesystem;
 
     boost::system::error_code ec;
-    fs::create_directories(destination, ec);
+    fs::create_directories(extended_path(destination), ec);
     if (ec) {
         error = "Failed to create plugin staging directory: " + ec.message();
         return false;
@@ -708,7 +705,7 @@ bool extract_zip_to_directory(const boost::filesystem::path& zip_path, const boo
 
         const fs::path output_path = destination / relative_path;
         if (stat.m_is_directory || mz_zip_reader_is_file_a_directory(&reader.archive, stat.m_file_index)) {
-            fs::create_directories(output_path, ec);
+            fs::create_directories(extended_path(output_path), ec);
             if (ec) {
                 error = "Failed to create plugin zip directory " + output_path.string() + ": " + ec.message();
                 return false;
@@ -716,26 +713,19 @@ bool extract_zip_to_directory(const boost::filesystem::path& zip_path, const boo
             continue;
         }
 
-        fs::create_directories(output_path.parent_path(), ec);
+        fs::create_directories(extended_path(output_path.parent_path()), ec);
         if (ec) {
             error = "Failed to create plugin zip parent directory " + output_path.parent_path().string() + ": " + ec.message();
             return false;
         }
 
-        if (fs::exists(output_path, ec) && fs::is_directory(output_path, ec)) {
+        if (fs::exists(extended_path(output_path), ec) && fs::is_directory(extended_path(output_path), ec)) {
             error = "Plugin zip file conflicts with an existing directory: " + output_path.string();
             return false;
         }
 
-        const std::string encoded_output_path = encode_path(output_path.string().c_str());
-        mz_bool extracted = mz_zip_reader_extract_to_file(&reader.archive, stat.m_file_index, encoded_output_path.c_str(), 0);
-#ifdef WIN32
-        if (!extracted) {
-            const std::wstring wide_output_path = boost::locale::conv::utf_to_utf<wchar_t>(output_path.generic_string());
-            extracted = mz_zip_reader_extract_to_file_w(&reader.archive, stat.m_file_index, wide_output_path.c_str(), 0);
-        }
-#endif
-        if (!extracted) {
+        // Shared with the rest of the archive code: long-path capable and binary-safe.
+        if (!extract_zip_entry_to_file(&reader.archive, stat.m_file_index, output_path)) {
             error = "Failed to extract plugin zip entry " + relative_key + ": " +
                     MZ_Archive::get_errorstr(mz_zip_get_last_error(&reader.archive));
             return false;
@@ -1255,4 +1245,76 @@ boost::filesystem::path find_installed_plugin_entry(const boost::filesystem::pat
     return {};
 }
 
+std::vector<std::string> collect_resource_texture_files(const std::string& dir,
+                                                        const std::string& plugin_root,
+                                                        std::string& error)
+{
+    namespace fs = boost::filesystem;
+
+    // Picker sanity bound, not a security control: the popup draws every entry in its grid.
+    constexpr size_t kMaxResourceTextureFiles = 64;
+
+    std::vector<std::string> files;
+    error.clear();
+
+    if (dir.empty()) {
+        error = _u8L("Resource texture directory is not set");
+        return files;
+    }
+    if (plugin_root.empty()) {
+        error = _u8L("Plugin folder could not be determined.");
+        return files;
+    }
+
+    boost::system::error_code ec;
+    const fs::path resolved_dir = fs::weakly_canonical(fs::path(dir), ec);
+    if (ec) {
+        error = _u8L("Resource texture directory could not be resolved");
+        return files;
+    }
+
+    if (!fs::is_directory(resolved_dir, ec) || ec) {
+        error = _u8L("Resource texture path is not a directory");
+        return files;
+    }
+
+    const fs::path resolved_root = fs::weakly_canonical(fs::path(plugin_root), ec);
+    if (ec) {
+        error = _u8L("Plugin folder could not be resolved");
+        return files;
+    }
+
+    if (!is_inside_allowed_root(resolved_dir, resolved_root)) {
+        error = _u8L("Resource texture directory is outside the plugin folder");
+        return files;
+    }
+
+    for (fs::directory_iterator it(resolved_dir, ec), end; it != end && !ec; it.increment(ec)) {
+        const fs::path& path = it->path();
+
+        if (!fs::is_regular_file(path, ec) || ec)
+            continue;
+
+        std::string ext = path.extension().string();
+        boost::algorithm::to_lower(ext);
+        if (ext != ".png" && ext != ".jpg" && ext != ".jpeg" && ext != ".bmp")
+            continue;
+
+        // Re-check per file: a symlink inside the pack must not resolve outside it.
+        if (!is_inside_allowed_root(path, resolved_root))
+            continue;
+
+        files.push_back(path.string());
+        if (files.size() >= kMaxResourceTextureFiles)
+            break;
+    }
+
+    if (ec) {
+        error = _u8L("Resource texture directory could not be read");
+        return {};
+    }
+
+    return files;
+}
+
 } // namespace Slic3r

src/slic3r/plugin/PluginFsUtils.hpp

@@ -176,6 +176,12 @@ bool read_wheel_plugin_metadata(const boost::filesystem::path& whl_path, PluginD
 // Returns the path to the entry file, or an empty path with error set if zero or multiple candidates.
 boost::filesystem::path find_installed_plugin_entry(const boost::filesystem::path& plugin_dir, std::string& error);
 
+// Validated non-recursive .png/.jpg/.jpeg/.bmp files directly inside `dir`, every one contained in
+// `plugin_root`. Empty with `error` set on failure.
+std::vector<std::string> collect_resource_texture_files(const std::string& dir,
+                                                        const std::string& plugin_root,
+                                                        std::string& error);
+
 // Canonical writer: emits the .install_state.json schema for the given state.
 bool write_install_state(const boost::filesystem::path& plugin_dir, const PluginInstallState& state);
 // Builds a PluginInstallState from the descriptor and delegates to the canonical writer.

src/slic3r/plugin/PluginLoader.cpp

@@ -9,6 +9,7 @@
 #include "PluginFsUtils.hpp"
 #include "PythonInterpreter.hpp"
 #include "PythonPluginBridge.hpp"
+#include "pluginTypes/resourceTexture/ResourceTexturePluginCapability.hpp"
 #include "libslic3r/Utils.hpp"
 #include "slic3r/GUI/I18N.hpp"
 #include "slic3r/GUI/format.hpp"
@@ -387,6 +388,37 @@ bool load(const PluginDescriptor&                          descriptor,
                         instance->set_config_ui_available(false);
                     }
 
+                    // ResourceTexture capabilities resolve their texture directory here, under the
+                    // GIL this block already holds, and cache the validated result. A resolution
+                    // fault only costs this capability its textures, so it is caught locally like
+                    // has_config_ui() above rather than failing the load.
+                    if (auto* resource_texture = dynamic_cast<ResourceTexturePluginCapability*>(instance.get())) {
+                        try {
+                            const std::string texture_dir = resource_texture->get_texture_dir();
+                            if (texture_dir.empty()) {
+                                resource_texture->set_texture_dir({});
+                            } else {
+                                std::string texture_error;
+                                collect_resource_texture_files(texture_dir, descriptor.plugin_root, texture_error);
+                                if (texture_error.empty()) {
+                                    resource_texture->set_texture_dir(texture_dir);
+                                } else {
+                                    BOOST_LOG_TRIVIAL(warning)
+                                        << "Plugin capability '" << found.name << "' of plugin '" << descriptor.plugin_key
+                                        << "': texture directory rejected (" << texture_error
+                                        << "); the capability will provide no textures";
+                                    resource_texture->set_texture_dir({});
+                                }
+                            }
+                        } catch (const std::exception& ex) {
+                            BOOST_LOG_TRIVIAL(warning)
+                                << "Plugin capability '" << found.name << "' of plugin '" << descriptor.plugin_key
+                                << "': get_texture_dir() failed (" << ex.what()
+                                << "); the capability will provide no textures";
+                            resource_texture->set_texture_dir({});
+                        }
+                    }
+
                     capabilities.push_back(std::move(instance));
                 }
             } catch (const std::exception& ex) {

src/slic3r/plugin/PythonPluginBridge.cpp

@@ -25,6 +25,7 @@
 #include "PyPluginTrampoline.hpp"
 #include "pluginTypes/printerAgent/PrinterAgentPluginCapability.hpp"
 #include "pluginTypes/pages/PagesPluginCapability.hpp"
+#include "pluginTypes/resourceTexture/ResourceTexturePluginCapability.hpp"
 #include "pluginTypes/script/ScriptPluginCapability.hpp"
 #include "pluginTypes/slicingPipeline/SlicingPipelinePluginCapability.hpp"
 #include <pybind11/pytypes.h>
@@ -374,6 +375,7 @@ void bind_python_api(pybind11::module_& m)
         .value("Visualization", PluginCapabilityType::Visualization)
         .value("Script", PluginCapabilityType::Script)
         .value("SlicingPipeline", PluginCapabilityType::SlicingPipeline)
+        .value("ResourceTexture", PluginCapabilityType::ResourceTexture)
         .value("Unknown", PluginCapabilityType::Unknown)
         .export_values();
 
@@ -527,6 +529,7 @@ void bind_python_api(pybind11::module_& m)
     PagesPluginCapability::RegisterBindings(m);
     ScriptPluginCapability::RegisterBindings(m);
     SlicingPipelinePluginCapability::RegisterBindings(m);
+    ResourceTexturePluginCapability::RegisterBindings(m);
     PluginHost::RegisterBindings(m);
     BOOST_LOG_TRIVIAL(debug) << "Registered ScriptPluginCapability Python bindings";
 

src/slic3r/plugin/PythonPluginInterface.hpp

@@ -14,7 +14,7 @@
 
 namespace Slic3r {
 
-enum class PluginCapabilityType { PrinterConnection = 0, Pages, Analysis, Importer, Exporter, Visualization, Script, SlicingPipeline, Unknown };
+enum class PluginCapabilityType { PrinterConnection = 0, Pages, Analysis, Importer, Exporter, Visualization, Script, SlicingPipeline, ResourceTexture, Unknown };
 
 struct PluginCapabilityId
 {
@@ -48,6 +48,7 @@ inline std::string plugin_capability_type_to_string(PluginCapabilityType type)
     case PluginCapabilityType::Visualization: return "visualization";
     case PluginCapabilityType::Script: return "script";
     case PluginCapabilityType::SlicingPipeline: return "slicing-pipeline";
+    case PluginCapabilityType::ResourceTexture: return "resource-texture";
     default: return "unknown";
     }
 }
@@ -63,6 +64,7 @@ inline std::string plugin_capability_type_display_name(PluginCapabilityType type
     case PluginCapabilityType::Visualization: return "Visualization";
     case PluginCapabilityType::Script: return "Script";
     case PluginCapabilityType::SlicingPipeline: return "Slicing Pipeline";
+    case PluginCapabilityType::ResourceTexture: return "Resource Texture";
     default: return "Unknown";
     }
 }
@@ -92,6 +94,8 @@ inline PluginCapabilityType plugin_capability_type_from_string(std::string_view
         return PluginCapabilityType::Script;
     if (lowered == "slicing-pipeline")
         return PluginCapabilityType::SlicingPipeline;
+    if (lowered == "resource-texture")
+        return PluginCapabilityType::ResourceTexture;
     return PluginCapabilityType::Unknown;
 }
 

src/slic3r/plugin/host/PluginPages.cpp

@@ -4,6 +4,7 @@
 #include "slic3r/GUI/GUI.hpp"
 #include "slic3r/GUI/Notebook.hpp"
 #include "slic3r/GUI/GUI_App.hpp"
+#include "slic3r/GUI/MainFrame.hpp"
 #include "slic3r/GUI/Widgets/Button.hpp"
 #include "slic3r/GUI/wxExtensions.hpp"
 #include "slic3r/plugin/PluginManager.hpp"
@@ -382,6 +383,7 @@ void PluginPages::relayout()
             m_parent->SelectPageByName(id_to_reselect);
     }
 
+    const bool overflow_changed = need_overflow != (m_overflow_button != nullptr);
     if (need_overflow) {
         if (m_overflow_button == nullptr) {
             auto* btn = new Button(m_parent->GetBtnsListCtrl(), wxString(L"\u25BE"), wxString(), wxNO_BORDER);
@@ -398,6 +400,11 @@ void PluginPages::relayout()
         m_overflow_button->Destroy();
         m_overflow_button = nullptr;
     }
+
+    // The tab strip changed width; compact the labels again so the side tools stay visible.
+    if ((!up_to_date || overflow_changed) && !GUI::wxGetApp().is_closing())
+        if (GUI::MainFrame* mainframe = GUI::wxGetApp().mainframe)
+            mainframe->fit_tab_labels();
 }
 
 void PluginPages::select_page(const PluginCapabilityId& id)

src/slic3r/plugin/host/PluginResourceTextures.cpp

@@ -0,0 +1,115 @@
+#include "PluginResourceTextures.hpp"
+
+#include "slic3r/plugin/PluginDescriptor.hpp"
+#include "slic3r/plugin/PluginManager.hpp"
+#include "slic3r/plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapability.hpp"
+
+#include <map>
+#include <memory>
+#include <mutex>
+#include <optional>
+#include <slic3r/plugin/PythonPluginInterface.hpp>
+#include <string>
+#include <utility>
+#include <vector>
+
+namespace Slic3r {
+
+PluginResourceTextures &PluginResourceTextures::instance()
+{
+    static PluginResourceTextures registry;
+    return registry;
+}
+
+std::optional<ResourceTextureEntry> PluginResourceTextures::build_entry(const PluginCapabilityId &id) const
+{
+    // Query by id and re-check state; the manager copies the shared_ptr and releases its lock.
+    auto capability = PluginManager::instance().get_plugin_capability(id, /*only_enabled=*/false);
+    if (!capability || !capability->is_enabled() || capability->type() != PluginCapabilityType::ResourceTexture)
+        return std::nullopt;
+
+    auto texture_capability = std::dynamic_pointer_cast<ResourceTexturePluginCapability>(capability);
+    if (!texture_capability)
+        return std::nullopt;
+
+    // An empty directory means resolution failed at load, so the pack contributes nothing.
+    const std::string &texture_dir = texture_capability->texture_dir();
+    if (texture_dir.empty())
+        return std::nullopt;
+
+    PluginDescriptor descriptor;
+    if (!PluginManager::instance().try_get_plugin_descriptor_for_capability(capability->name(),
+                                                                            PluginCapabilityType::ResourceTexture,
+                                                                            descriptor))
+        return std::nullopt;
+
+    ResourceTextureEntry entry;
+    entry.id           = capability->identity();
+    entry.plugin_key   = entry.id.plugin_key;
+    entry.display_name = capability->name();
+    entry.texture_dir  = texture_dir;
+    entry.plugin_root  = descriptor.plugin_root;
+    return entry;
+}
+
+void PluginResourceTextures::on_plugin_register(const std::string &plugin_key)
+{
+    for (const auto &capability : PluginManager::instance().get_plugin_capabilities(plugin_key, PluginCapabilityType::ResourceTexture)) {
+        if (capability)
+            on_cap_register(capability->identity());
+    }
+}
+
+void PluginResourceTextures::on_plugin_deregister(const std::string &plugin_key)
+{
+    std::lock_guard<std::mutex> lock(m_mutex);
+    for (auto it = m_entries.begin(); it != m_entries.end();) {
+        if (it->first.plugin_key == plugin_key)
+            it = m_entries.erase(it);
+        else
+            ++it;
+    }
+}
+
+void PluginResourceTextures::on_cap_register(const PluginCapabilityId &id)
+{
+    std::optional<ResourceTextureEntry> entry = build_entry(id);
+
+    std::lock_guard<std::mutex> lock(m_mutex);
+    if (entry)
+        m_entries[entry->id] = std::move(*entry);
+}
+
+void PluginResourceTextures::on_cap_deregister(const PluginCapabilityId &id)
+{
+    std::lock_guard<std::mutex> lock(m_mutex);
+    m_entries.erase(id);
+}
+
+void PluginResourceTextures::initial_sweep()
+{
+    std::map<PluginCapabilityId, ResourceTextureEntry> entries;
+    for (const auto &capability : PluginManager::instance().get_plugin_capabilities("", PluginCapabilityType::ResourceTexture)) {
+        if (!capability)
+            continue;
+        std::optional<ResourceTextureEntry> entry = build_entry(capability->identity());
+        if (entry)
+            entries[entry->id] = std::move(*entry);
+    }
+
+    std::lock_guard<std::mutex> lock(m_mutex);
+    m_entries = std::move(entries);
+}
+
+std::vector<ResourceTextureEntry> PluginResourceTextures::entries() const
+{
+    std::lock_guard<std::mutex> lock(m_mutex);
+
+    std::vector<ResourceTextureEntry> result;
+    result.reserve(m_entries.size());
+    for (const auto &entry : m_entries)
+        result.push_back(entry.second);
+    return result;
+}
+
+} // namespace Slic3r

src/slic3r/plugin/host/PluginResourceTextures.hpp

@@ -0,0 +1,50 @@
+#pragma once
+
+#include <slic3r/plugin/PythonPluginInterface.hpp>
+
+#include <map>
+#include <mutex>
+#include <optional>
+#include <string>
+#include <vector>
+
+namespace Slic3r {
+
+// One ResourceTexture capability contributing a texture directory to the texture picker.
+struct ResourceTextureEntry
+{
+    PluginCapabilityId id;
+    std::string        plugin_key;
+    std::string        display_name;
+    std::string        texture_dir;
+    std::string        plugin_root;   // so a scan can re-check containment
+};
+
+// Tracks every loaded and enabled ResourceTexture capability by id. Holds strings only: a
+// shared_ptr here would keep a Python object alive past Py_Finalize.
+class PluginResourceTextures
+{
+public:
+    static PluginResourceTextures &instance();
+
+    PluginResourceTextures(const PluginResourceTextures &)            = delete;
+    PluginResourceTextures &operator=(const PluginResourceTextures &) = delete;
+
+    void on_plugin_register(const std::string &plugin_key);
+    void on_plugin_deregister(const std::string &plugin_key);
+    void on_cap_register(const PluginCapabilityId &id);
+    void on_cap_deregister(const PluginCapabilityId &id);
+    void initial_sweep();
+
+    std::vector<ResourceTextureEntry> entries() const;
+
+private:
+    PluginResourceTextures() = default;
+
+    std::optional<ResourceTextureEntry> build_entry(const PluginCapabilityId &id) const;
+
+    mutable std::mutex                                 m_mutex;
+    std::map<PluginCapabilityId, ResourceTextureEntry> m_entries;
+};
+
+} // namespace Slic3r

src/slic3r/plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapability.cpp

@@ -0,0 +1,32 @@
+#include "ResourceTexturePluginCapability.hpp"
+#include "ResourceTexturePluginCapabilityTrampoline.hpp"
+
+#include <pybind11/pybind11.h>
+#include <pybind11/pytypes.h>
+
+#include <boost/log/trivial.hpp>
+
+#include <memory>
+#include <slic3r/plugin/PythonPluginInterface.hpp>
+#include <string>
+
+namespace py = pybind11;
+
+namespace Slic3r {
+
+void ResourceTexturePluginCapability::RegisterBindings(pybind11::module_& module)
+{
+    BOOST_LOG_TRIVIAL(debug) << "Registering orca.resource_texture bindings";
+
+    auto resource_texture = module.def_submodule("resource_texture", "Plugin resource texture API");
+
+    py::class_<ResourceTexturePluginCapability, PluginCapabilityInterface,
+               PyResourceTexturePluginCapabilityTrampoline,
+               std::shared_ptr<ResourceTexturePluginCapability>>(resource_texture,
+                                                                 "ResourceTexturePluginCapabilityBase")
+        .def(py::init<>())
+        .def("get_type", &ResourceTexturePluginCapability::get_type)
+        .def("get_texture_dir", &ResourceTexturePluginCapability::get_texture_dir);
+}
+
+} // namespace Slic3r

src/slic3r/plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapability.hpp

@@ -0,0 +1,28 @@
+#pragma once
+
+#include "../../PythonPluginInterface.hpp"
+#include "pybind11/pybind11.h"
+
+#include <string>
+#include <utility>
+
+namespace Slic3r {
+class ResourceTexturePluginCapability : public PluginCapabilityInterface
+{
+public:
+    static void RegisterBindings(pybind11::module_& module);
+
+    PluginCapabilityType get_type() const override { return PluginCapabilityType::ResourceTexture; }
+
+    // Python contract: the directory the host scans for textures. Resolved once at materialization
+    // under the GIL; never called from the GUI. Overridden in Python through the trampoline.
+    virtual std::string get_texture_dir() const = 0;
+
+    // C++-only read of the validated directory cached by the loader; empty if resolution failed.
+    const std::string& texture_dir() const { return m_texture_dir; }
+    void               set_texture_dir(std::string dir) { m_texture_dir = std::move(dir); }
+
+private:
+    std::string m_texture_dir;
+};
+} // namespace Slic3r

src/slic3r/plugin/pluginTypes/resourceTexture/ResourceTexturePluginCapabilityTrampoline.hpp

@@ -0,0 +1,30 @@
+#pragma once
+
+#include "ResourceTexturePluginCapability.hpp"
+#include "../../PyPluginTrampoline.hpp"
+#include <pybind11/pybind11.h>
+#include <pybind11/pytypes.h>
+
+#include <string>
+#include "slic3r/plugin/PythonPluginInterface.hpp"
+
+namespace Slic3r {
+
+class PyResourceTexturePluginCapabilityTrampoline :
+    public PyPluginCommonTrampoline<ResourceTexturePluginCapability>
+{
+public:
+    using PyPluginCommonTrampoline<ResourceTexturePluginCapability>::PyPluginCommonTrampoline;
+
+    std::string get_texture_dir() const override
+    {
+        ORCA_PY_OVERRIDE_AUDITED(
+            [] {},
+            PYBIND11_OVERRIDE_PURE,
+            std::string,
+            ResourceTexturePluginCapability,
+            get_texture_dir);
+    }
+};
+
+} // namespace Slic3r

tests/data/wipe_tower_temperature_trace_main.txt

@@ -163,7 +163,7 @@ M109 S240 T0 ; set nozzle temperature and wait for it to be reached
 ; CP TOOLCHANGE END
 M104 S240 T1 ; preheat T1 time: 30s	lead 30.0s
 ; CP TOOLCHANGE START
-M104 S200 T0 ; set nozzle temperature ;cooldown
+M104 S0 T0 ; set nozzle temperature ;cooldown
 T1 ; change extruder
 M109 S240 T1 ; set nozzle temperature and wait for it to be reached
 ; CP TOOLCHANGE END

tests/fff_print/test_multifilament.cpp

@@ -1742,3 +1742,112 @@ TEST_CASE("Each filament cools with the fan speeds of its extruder variant", "[M
         CHECK(gcode.find("; start range high " + std::to_string(range_high) + "\n") != std::string::npos);
     }
 }
+
+TEST_CASE("Shared heater on IMEX printer stays hot until all filaments on it are done", "[MultiFilament]")
+{
+    DynamicPrintConfig cfg = multifilament_config(7);
+    cfg.set_deserialize_strict({
+        {"ooze_prevention", 1},
+        {"preheat_time", 0},
+        {"single_extruder_multi_material", 0},
+    });
+    imex_7x4_printer(cfg);
+
+    const std::string gcode = slice_with_object_overrides({make_cube(20., 20., 10.), make_cube(20., 20., 4.), make_cube(20., 20., 10.)},
+                                                          cfg,
+                                                          {{{"extruder", 1}, {"layer_height", 0.30}},
+                                                           {{"extruder", 2}, {"layer_height", 0.15}},
+                                                           {{"extruder", 5}}});
+
+    bool premature_s0_t0 = false;
+    double current_z     = 0.0;
+
+    GCodeReader reader;
+    reader.parse_buffer(gcode, [&](GCodeReader& /*self*/, const GCodeReader::GCodeLine& line) {
+        if (line.has_z())
+            current_z = line.z();
+
+        const std::string raw = line.raw();
+        if (raw.find("M104 S0 T0") != std::string::npos && raw.find(";cooldown") != std::string::npos) {
+            if (current_z < 9.0)
+                premature_s0_t0 = true;
+        }
+    });
+
+    // Filaments 1 and 2 share head 0, and filament 1 prints up to 10mm.
+    // T0 must not shut down at ~4mm when filament 2 finishes.
+    CHECK_FALSE(premature_s0_t0);
+}
+
+TEST_CASE("Unused extruder turns off after its final layer on multi-object print", "[MultiFilament]")
+{
+    const std::string gcode = slice_with_object_overrides(
+        { make_cube(20., 20., 10.), make_cube(20., 20., 4.) },
+        multifilament_config(2, {
+            { "single_extruder_multi_material", 0 },
+            { "ooze_prevention",                1 },
+            { "standby_temperature_delta",      -40 },
+            { "nozzle_temperature",             "240,240" },
+        }),
+        {
+            { { "extruder", 1 }, { "layer_height", 0.20 } },
+            { { "extruder", 2 }, { "layer_height", 0.15 } }
+        });
+
+    int t1_s0_cooldown_count = 0;
+    int t0_s0_cooldown_count = 0;
+
+    std::istringstream stream(gcode);
+    for (std::string line; std::getline(stream, line);) {
+        if (line.find(";cooldown") != std::string::npos) {
+            if (line.find("M104 S0 T1") != std::string::npos)
+                ++t1_s0_cooldown_count;
+            if (line.find("M104 S0 T0") != std::string::npos)
+                ++t0_s0_cooldown_count;
+        }
+    }
+
+    // T1 (extruder 2, 4mm cube) gets exactly one S0 after its last layer.
+    // T0 (extruder 1, 10mm cube) never gets S0 mid-print.
+    CHECK(t1_s0_cooldown_count == 1);
+    CHECK(t0_s0_cooldown_count == 0);
+}
+
+TEST_CASE("Unused extruder turns off on non-IMEX three-tool print", "[MultiFilament]")
+{
+    const std::string gcode = slice_with_object_overrides(
+        { make_cube(20., 20., 4.), make_cube(20., 20., 10.), make_cube(20., 20., 10.) },
+        multifilament_config(3, {
+            { "single_extruder_multi_material", 0 },
+            { "ooze_prevention",                1 },
+            { "standby_temperature_delta",      -40 },
+            { "nozzle_temperature",             "240,240,240" },
+        }),
+        {
+            { { "extruder", 1 } },
+            { { "extruder", 2 } },
+            { { "extruder", 3 } }
+        });
+
+    bool premature_s0_t1 = false;
+    bool premature_s0_t2 = false;
+    double current_z     = 0.0;
+
+    GCodeReader reader;
+    reader.parse_buffer(gcode, [&](GCodeReader& /*self*/, const GCodeReader::GCodeLine& line) {
+        if (line.has_z())
+            current_z = line.z();
+
+        const std::string raw = line.raw();
+        if (raw.find(";cooldown") != std::string::npos && current_z < 9.0) {
+            if (raw.find("M104 S0 T1") != std::string::npos)
+                premature_s0_t1 = true;
+            if (raw.find("M104 S0 T2") != std::string::npos)
+                premature_s0_t2 = true;
+        }
+    });
+
+    // T1 and T2 print up to 10mm — neither should get S0 before that.
+    CHECK_FALSE(premature_s0_t1);
+    CHECK_FALSE(premature_s0_t2);
+}

tests/libslic3r/CMakeLists.txt

@@ -37,6 +37,7 @@ add_executable(${_TEST_NAME}_tests
     test_fill_plane_path.cpp
     test_fill_tpms_adaptive.cpp
     test_geometry.cpp
+    test_chromastack.cpp
     test_imex_arrange.cpp
     test_imex_helpers.cpp
     test_imex_zones.cpp

tests/libslic3r/test_chromastack.cpp

@@ -0,0 +1,379 @@
+#include <catch2/catch_approx.hpp>
+#include <catch2/catch_test_macros.hpp>
+#include <catch2/matchers/catch_matchers_floating_point.hpp>
+
+#include <algorithm>
+#include <cmath>
+#include <cstdint>
+#include <vector>
+
+#include "libslic3r/Color.hpp"
+#include "libslic3r/ChromaStack.hpp"
+#include "libslic3r/PNGReadWrite.hpp"
+
+using namespace Slic3r;
+using Catch::Matchers::WithinAbs;
+using Catch::Matchers::WithinRel;
+
+namespace {
+
+ChromaFilament filament(int index, const ColorRGB &color, float td_mm)
+{
+    ChromaFilament f;
+    f.filament = index;
+    f.color    = color;
+    f.td_mm    = td_mm;
+    return f;
+}
+
+ChromaBand band(const ChromaFilament &f, int layers)
+{
+    ChromaBand b;
+    b.filament = f;
+    b.layers   = layers;
+    return b;
+}
+
+// An opaque filament (TD 1 hides what is under it within 0.1 mm) and a translucent one (TD 8 needs
+// most of a millimetre), which is the whole range the model has to behave sensibly over.
+const ChromaFilament WHITE_OPAQUE      = filament(0, ColorRGB(1.f, 1.f, 1.f), 1.f);
+const ChromaFilament BLACK_OPAQUE      = filament(1, ColorRGB(0.f, 0.f, 0.f), 1.f);
+const ChromaFilament BLACK_TRANSLUCENT = filament(2, ColorRGB(0.f, 0.f, 0.f), 8.f);
+
+float luminance(const ColorRGB &c) { return 0.2126f * c.r() + 0.7152f * c.g() + 0.0722f * c.b(); }
+
+} // namespace
+
+TEST_CASE("A stack's palette has one entry per printed layer", "[ChromaStack]")
+{
+    ChromaStack stack;
+    stack.layer_height_mm = 0.08f;
+    stack.base_layers     = 4;
+    stack.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 10) };
+
+    // The base is backing under the picture, not part of a band: the bands are the 12 layers the
+    // picture is built from, and the stack is those on top of the 4 the model already has.
+    REQUIRE(stack.relief_layers() == 12);
+    REQUIRE(stack.total_layers() == 16);
+    REQUIRE_THAT(stack.depth_mm(), WithinAbs(12 * 0.08f, 1e-6f));
+
+    const std::vector<ColorRGB> palette = chroma_palette(stack);
+    REQUIRE(palette.size() == 17);
+    // Index 0 is the base's own colour, so a column one layer tall is exactly its filament - whatever
+    // that filament's transmission distance.
+    REQUIRE(palette[0] == palette[1]);
+    REQUIRE_THAT(luminance(palette[1]), WithinAbs(1.f, 1e-3f));
+}
+
+TEST_CASE("A run of one filament gives the same palette however it is split into bands", "[ChromaStack]")
+{
+    // The property that justifies compositing layer by layer at all: transmittance multiplies, so
+    // cutting a run in two must not change the colours. A model without it would make the palette
+    // depend on how the stack was typed in.
+    ChromaStack one;
+    one.layer_height_mm = 0.1f;
+    one.base_layers     = 0;
+    one.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 8) };
+
+    ChromaStack split = one;
+    split.bands         = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 3), band(BLACK_TRANSLUCENT, 5) };
+
+    const std::vector<ColorRGB> a = chroma_palette(one);
+    const std::vector<ColorRGB> b = chroma_palette(split);
+    REQUIRE(a.size() == b.size());
+    for (size_t i = 0; i < a.size(); ++i)
+        REQUIRE_THAT(luminance(a[i]), WithinAbs(luminance(b[i]), 1e-5f));
+}
+
+TEST_CASE("A translucent filament darkens gradually and an opaque one at once", "[ChromaStack]")
+{
+    ChromaStack translucent;
+    translucent.layer_height_mm = 0.08f;
+    translucent.base_layers     = 0;
+    translucent.bands           = { band(WHITE_OPAQUE, 1), band(BLACK_TRANSLUCENT, 12) };
+    const std::vector<ColorRGB> slow = chroma_palette(translucent);
+
+    // Monotonically darker, and still short of the filament's own black after a single layer: that
+    // gradient is what a ChromaStack stack prints its midtones with.
+    for (size_t i = 2; i < slow.size(); ++i)
+        REQUIRE(luminance(slow[i]) < luminance(slow[i - 1]));
+    REQUIRE(luminance(slow[2]) > 0.5f);
+    // Twelve layers of a TD 8 filament pass under 3% of the light, which is the dark end of the ramp
+    // rather than true black - and that is the honest prediction, not a rounding artifact.
+    REQUIRE(luminance(slow.back()) < 0.25f);
+
+    ChromaStack opaque;
+    opaque.layer_height_mm = 0.08f;
+    opaque.base_layers     = 0;
+    opaque.bands           = { band(WHITE_OPAQUE, 1), band(BLACK_OPAQUE, 12) };
+    const std::vector<ColorRGB> fast = chroma_palette(opaque);
+    // TD 1 is opaque within about 0.1 mm, so two layers of it leave little of the white showing.
+    REQUIRE(luminance(fast[3]) < 0.2f);
+}
+
+TEST_CASE("Solving a picture of the palette's own colors reproduces it exactly", "[ChromaStack]")
+{
+    ChromaStack stack;
+    stack.layer_height_mm = 0.08f;
+    stack.base_layers     = 2;
+    stack.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 8) };
+
+    const std::vector<ColorRGB> palette = chroma_palette(stack);
+    const int                   relief  = stack.relief_layers();
+
+    // One pixel per reachable height, in order.
+    std::vector<uint8_t> rgb;
+    for (int i = 0; i <= relief; ++i) {
+        const ColorRGB &c = palette[size_t(stack.base_layers + i)];
+        rgb.push_back(uint8_t(std::lround(c.r() * 255.f)));
+        rgb.push_back(uint8_t(std::lround(c.g() * 255.f)));
+        rgb.push_back(uint8_t(std::lround(c.b() * 255.f)));
+    }
+
+    const ChromaSolve solve = chroma_solve(rgb.data(), relief + 1, 1, stack);
+    REQUIRE_FALSE(solve.empty());
+    REQUIRE(solve.width == relief + 1);
+    REQUIRE(solve.levels.size() == size_t(relief + 1));
+    // Every pixel comes back as the colour it was generated from, and the error is rounding only.
+    //
+    // The colour rather than the height: the bottom band prints in the same filament as the base, so
+    // its first layers show exactly the colour the base already showed, and any of those heights is a
+    // correct answer for that pixel. Asserting the height would be asserting which of several equally
+    // right answers the search happens to reach first.
+    for (int i = 0; i <= relief; ++i) {
+        const int chosen = int(solve.levels[size_t(i)]);
+        REQUIRE(chosen >= stack.base_layers);
+        REQUIRE(chosen <= stack.total_layers());
+        const ColorRGB &got  = palette[size_t(chosen)];
+        const ColorRGB &want = palette[size_t(stack.base_layers + i)];
+        REQUIRE_THAT(got.r(), WithinAbs(want.r(), 2.f / 255.f));
+        REQUIRE_THAT(got.g(), WithinAbs(want.g(), 2.f / 255.f));
+        REQUIRE_THAT(got.b(), WithinAbs(want.b(), 2.f / 255.f));
+    }
+    // Only 8-bit rounding of the palette separates the two, which near black is still under one dE.
+    REQUIRE(solve.max_delta_e < 2.);
+
+    // The height map's ends are the base and the top of the stack, which is what ties a grey level to
+    // a printed layer.
+    REQUIRE(solve.grey.front() == 0);
+    REQUIRE(solve.grey.back() == 255);
+}
+
+TEST_CASE("A picture the filaments cannot reach is reported as a poor match", "[ChromaStack]")
+{
+    ChromaStack stack;
+    stack.layer_height_mm = 0.08f;
+    stack.base_layers     = 2;
+    stack.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 8) };
+
+    // Saturated red against a stack of white and black: no height can produce it, and the panel's
+    // warning depends on the solve saying so rather than silently picking a grey.
+    const std::vector<uint8_t> red{ 255, 0, 0 };
+    const ChromaSolve        solve = chroma_solve(red.data(), 1, 1, stack);
+    REQUIRE_FALSE(solve.empty());
+    REQUIRE(solve.mean_delta_e > 15.);
+}
+
+TEST_CASE("Dithering keeps every pixel in the stack's range", "[ChromaStack]")
+{
+    ChromaStack stack;
+    stack.layer_height_mm = 0.08f;
+    stack.base_layers     = 2;
+    stack.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 8) };
+    stack.dither          = true;
+
+    // A flat mid-grey: the error diffusion has to spread it over neighbouring heights without ever
+    // leaving the stack, which is what a clamped palette index would quietly get wrong.
+    std::vector<uint8_t> grey(size_t(16 * 16) * 3, 128);
+    const ChromaSolve  solve = chroma_solve(grey.data(), 16, 16, stack);
+    REQUIRE_FALSE(solve.empty());
+    REQUIRE(solve.levels.size() == size_t(16 * 16));
+    const auto [lo, hi] = std::minmax_element(solve.levels.begin(), solve.levels.end());
+    REQUIRE(int(*lo) >= stack.base_layers);
+    REQUIRE(int(*hi) <= stack.total_layers());
+}
+
+TEST_CASE("The solved height map is an 8-bit grayscale PNG of the picture's size", "[ChromaStack]")
+{
+    // The displacement pipeline reads one thing from a layer: an 8-bit grayscale PNG. A solve that
+    // produced anything else would leave the relief silently flat.
+    ChromaStack stack;
+    stack.layer_height_mm = 0.08f;
+    stack.base_layers     = 2;
+    stack.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_TRANSLUCENT, 6) };
+
+    std::vector<uint8_t> picture(size_t(8 * 4) * 3);
+    for (size_t i = 0; i < picture.size(); ++i)
+        picture[i] = uint8_t(i % 256);
+
+    const ChromaSolve              solve = chroma_solve(picture.data(), 8, 4, stack);
+    const std::vector<unsigned char> png   = chroma_encode_height_png(solve);
+    REQUIRE_FALSE(png.empty());
+
+    const png::ReadBuf  buf{ png.data(), png.size() };
+    png::ImageGreyscale decoded;
+    REQUIRE(png::is_png(buf));
+    REQUIRE(png::decode_png(buf, decoded));
+    REQUIRE(decoded.cols == 8);
+    REQUIRE(decoded.rows == 4);
+    REQUIRE(decoded.buf == solve.grey);
+}
+
+TEST_CASE("Each band becomes a height range, and a repeated filament is not a swap", "[ChromaStack]")
+{
+    ChromaStack stack;
+    stack.layer_height_mm = 0.1f;
+    stack.base_layers     = 0;
+    stack.bands           = { band(WHITE_OPAQUE, 5), band(BLACK_OPAQUE, 3), band(BLACK_OPAQUE, 2) };
+
+    const std::vector<ChromaRange> ranges = chroma_ranges(stack, 0.);
+    // Three bands, two filaments: the second and third merge, because nothing is swapped between them.
+    REQUIRE(ranges.size() == 2);
+    REQUIRE_THAT(ranges[0].z_min, WithinAbs(0., 1e-9));
+    REQUIRE_THAT(ranges[0].z_max, WithinAbs(0.5, 1e-9));
+    REQUIRE(ranges[0].filament == WHITE_OPAQUE.filament);
+    REQUIRE_THAT(ranges[1].z_min, WithinAbs(0.5, 1e-9));
+    REQUIRE_THAT(ranges[1].z_max, WithinAbs(1.0, 1e-9));
+    REQUIRE(ranges[1].filament == BLACK_OPAQUE.filament);
+}
+
+TEST_CASE("A default stack puts the lightest filament at the bottom", "[ChromaStack]")
+{
+    // Light to dark is what gives the long tonal ramp; the order is the one thing a seeded stack has
+    // to get right, since everything else is a layer count the user can drag.
+    const std::vector<ChromaFilament> loaded{ BLACK_OPAQUE, WHITE_OPAQUE,
+                                                filament(3, ColorRGB(0.5f, 0.5f, 0.5f), 4.f) };
+    const ChromaStack stack = chroma_default_stack(loaded, 0.08f, 4, 24);
+
+    REQUIRE(stack.bands.size() == 3);
+    REQUIRE(stack.bands.front().filament.filament == WHITE_OPAQUE.filament);
+    REQUIRE(stack.bands.back().filament.filament == BLACK_OPAQUE.filament);
+    REQUIRE(stack.base_layers == 4);
+    // The base is printed in the bottom band, so the relief is what was asked for and the base sits
+    // underneath it.
+    REQUIRE(stack.total_layers() == 4 + 24);
+    REQUIRE(stack.relief_layers() == 24);
+    REQUIRE_THAT(stack.layer_height_mm, WithinRel(0.08f));
+}
+
+TEST_CASE("An empty stack or picture solves to nothing", "[ChromaStack]")
+{
+    ChromaStack empty;
+    REQUIRE(chroma_palette(empty).empty());
+    REQUIRE(chroma_ranges(empty, 0.).empty());
+
+    // A base with nothing on it: there is no picture to solve, and no filament to print the base in.
+    ChromaStack base_only;
+    base_only.layer_height_mm = 0.08f;
+    base_only.base_layers     = 4;
+    REQUIRE(base_only.relief_layers() == 0);
+    REQUIRE(chroma_palette(base_only).empty());
+    const std::vector<uint8_t> pixel{ 128, 128, 128 };
+    REQUIRE(chroma_solve(pixel.data(), 1, 1, base_only).empty());
+
+    ChromaStack usable;
+    usable.layer_height_mm = 0.08f;
+    usable.base_layers     = 2;
+    usable.bands           = { band(WHITE_OPAQUE, 2), band(BLACK_OPAQUE, 4) };
+    REQUIRE(chroma_solve(nullptr, 4, 4, usable).empty());
+}
+
+TEST_CASE("The base extends the bottom band's range down to the first layer", "[ChromaStack]")
+{
+    // The base is not a swap of its own: it prints in whatever is loaded at the start, which is the
+    // bottom band's filament, so the first range has to reach the build plate.
+    ChromaStack stack;
+    stack.layer_height_mm = 0.1f;
+    stack.base_layers     = 20;
+    stack.bands           = { band(WHITE_OPAQUE, 5), band(BLACK_OPAQUE, 5) };
+
+    const std::vector<ChromaRange> ranges = chroma_ranges(stack, 0.);
+    REQUIRE(ranges.size() == 2);
+    REQUIRE_THAT(ranges[0].z_min, WithinAbs(0., 1e-9));
+    REQUIRE_THAT(ranges[0].z_max, WithinAbs(2.5, 1e-9)); // 20 base + 5 band layers at 0.1 mm
+    REQUIRE_THAT(ranges[1].z_min, WithinAbs(2.5, 1e-9));
+    REQUIRE_THAT(ranges[1].z_max, WithinAbs(3.0, 1e-9));
+    REQUIRE(ranges.back().z_max == Catch::Approx(double(stack.total_layers()) * double(stack.layer_height_mm)));
+}
+
+TEST_CASE("A surface's color follows the printed layer it sits at", "[ChromaStack]")
+{
+    // What the preview colours by: a swap happens at one print Z, so the layer a surface sits in is
+    // what decides its colour, whatever the height map says.
+    ChromaStack stack;
+    stack.layer_height_mm = 0.1f;
+    stack.base_layers     = 2;
+    stack.bands           = { band(WHITE_OPAQUE, 3), band(BLACK_OPAQUE, 5) };
+
+    REQUIRE(chroma_level_at(stack, 0.05) == 1);  // inside the first layer
+    REQUIRE(chroma_level_at(stack, 0.10) == 1);  // exactly its top, which that layer printed
+    REQUIRE(chroma_level_at(stack, 0.15) == 2);
+    REQUIRE(chroma_level_at(stack, 0.50) == 5);  // the top of the base plus the white band
+    // Clamped at both ends: below the plate and above the stack there is nothing else to show.
+    REQUIRE(chroma_level_at(stack, -1.) == 1);
+    REQUIRE(chroma_level_at(stack, 99.) == stack.total_layers());
+}
+
+TEST_CASE("How fast a filament covers is its TD against the layer height", "[ChromaStack]")
+{
+    // The number that decides whether a band is a gradient or a step, and the one the panel reports:
+    // a filament hides what is under it at about a tenth of its TD, so how many layers that is
+    // depends entirely on how thick they are.
+    REQUIRE_THAT(chroma_layers_to_opaque(4.f, 0.2f), WithinRel(2.f, 1e-4f));
+    REQUIRE_THAT(chroma_layers_to_opaque(4.f, 0.08f), WithinRel(5.f, 1e-4f));
+    REQUIRE_THAT(chroma_layers_to_opaque(8.f, 0.08f), WithinRel(10.f, 1e-4f));
+    // An opaque filament at a coarse layer height covers within one layer, which is where a stack
+    // stops having any colours between its filaments at all.
+    REQUIRE(chroma_layers_to_opaque(1.f, 0.2f) < 1.f);
+    REQUIRE_THAT(chroma_layers_to_opaque(4.f, 0.f), WithinAbs(0.f, 1e-6f));
+}
+
+TEST_CASE("A finer layer height buys colors between the filaments", "[ChromaStack]")
+{
+    // The whole of the user-visible problem in one assertion: the same stack, the same filaments, two
+    // layer heights. The coarse one steps from white to black in two layers; the fine one fades.
+    const auto distinct = [](float layer_height) {
+        ChromaStack stack;
+        stack.layer_height_mm = layer_height;
+        stack.base_layers     = 0;
+        stack.bands           = { band(WHITE_OPAQUE, 1), band(filament(5, ColorRGB(0.f, 0.f, 0.f), 8.f), 10) };
+        const std::vector<ColorRGB> palette = chroma_palette(stack);
+        int                         n       = 0;
+        for (size_t i = 1; i < palette.size(); ++i)
+            if (std::abs(luminance(palette[i]) - luminance(palette[i - 1])) > 0.02f)
+                ++n;
+        return n;
+    };
+    REQUIRE(distinct(0.08f) > distinct(0.3f));
+}
+
+TEST_CASE("Layers past a filament's covering depth add no color", "[ChromaStack]")
+{
+    // The band is twice as long as the filament needs to hide what is under it, so half of it shows a
+    // colour the layer below already showed. That is invisible in the stack and obvious here.
+    ChromaStack stack;
+    stack.layer_height_mm = 0.08f;
+    stack.base_layers     = 0;
+    stack.bands           = { band(WHITE_OPAQUE, 1), band(filament(5, ColorRGB(0.f, 0.f, 0.f), 4.f), 10) };
+
+    // Trimming keeps every layer that still changes the colour and drops the rest, so the stack gets
+    // shorter and loses none of its colours.
+    const int distinct = chroma_distinct_levels(stack);
+    const int before   = stack.relief_layers();
+    chroma_trim_bands(stack);
+    REQUIRE(stack.relief_layers() <= before);
+    REQUIRE(chroma_distinct_levels(stack) == distinct);
+    // The white band is a single layer and has nothing to give up.
+    REQUIRE(stack.bands.front().layers == 1);
+
+    // A light band over a dark one is the case where layers really are wasted: it saturates within a
+    // few layers, and the rest of a long band shows what the layer below already showed.
+    ChromaStack light_over_dark;
+    light_over_dark.layer_height_mm = 0.08f;
+    light_over_dark.base_layers     = 0;
+    light_over_dark.bands = { band(filament(5, ColorRGB(0.f, 0.f, 0.f), 4.f), 1), band(WHITE_OPAQUE, 14) };
+    const int kept = chroma_distinct_levels(light_over_dark);
+    chroma_trim_bands(light_over_dark);
+    REQUIRE(light_over_dark.bands.back().layers < 14);
+    REQUIRE(chroma_distinct_levels(light_over_dark) == kept);
+}

tests/libslic3r/test_miniz_extension.cpp

@@ -18,6 +18,7 @@
 #include <fstream>
 #include <ios>
 #include <iterator>
+#include <libslic3r/Utils.hpp>
 #include <miniz.h>
 #include <string>
 #include <utility>
@@ -176,6 +177,55 @@ TEST_CASE("Confined extraction writes nothing outside the target for Windows-spe
     CHECK(list_dir(tmp.path()) == std::vector<std::string>{"bundle.zip", "cache"});
 }
 
+TEST_CASE("Confined extraction writes files under a target path longer than MAX_PATH", "[MinizExtension]")
+{
+    ScopedTemporaryDir tmp;
+    const fs::path     zip_file = tmp.path() / "bundle.zip";
+    // Directory entries are explicit: extract_archive_confined recreates directories only for them.
+    write_zip(zip_file, {{"vendor/", ""}, {"vendor/machine/", ""}, {"vendor.json", "{\"a\":1}"},
+                         {"vendor/machine/printer.json", "{\"b\":2}"}});
+
+    // One 240-char component keeps every name below the NTFS 255 limit while pushing the full
+    // target path past the 260-char MAX_PATH the Win32 APIs enforce when LongPathsEnabled is off.
+    const fs::path target = tmp.path() / std::string(240, 'd');
+    REQUIRE(target.string().size() > 260);
+    fs::create_directories(extended_path(target));
+
+    REQUIRE(extract_archive_confined(zip_file.string(), target.string()));
+    // Verify through extended-length paths, otherwise the checks themselves are bounded by MAX_PATH.
+    CHECK(fs::is_directory(extended_path(target / "vendor")));
+    CHECK(read_file(extended_path(target / "vendor.json")) == "{\"a\":1}");
+    CHECK(read_file(extended_path(target / "vendor" / "machine" / "printer.json")) == "{\"b\":2}");
+
+    // The temporary dir guard cannot remove a deep tree on Windows, so drop it here.
+    boost::system::error_code ec;
+    fs::remove_all(extended_path(target), ec);
+    CHECK_FALSE(ec);
+}
+
+TEST_CASE("A zip archive under a path longer than MAX_PATH can be written and read", "[MinizExtension]")
+{
+    ScopedTemporaryDir tmp;
+
+    // Create the long directory with the extended-length path, then exercise the archive
+    // reader/writer against an archive whose own path is past MAX_PATH.
+    const fs::path long_dir = tmp.path() / std::string(240, 'z');
+    fs::create_directories(extended_path(long_dir));
+    const fs::path zip_file = long_dir / "bundle.zip";
+    REQUIRE(zip_file.string().size() > 260);
+
+    write_zip(zip_file, {{"vendor.json", "{\"a\":1}"}});
+
+    const fs::path target = tmp.path() / "out";
+    fs::create_directories(target);
+    REQUIRE(extract_archive_confined(zip_file.string(), target.string()));
+    CHECK(read_file(target / "vendor.json") == "{\"a\":1}");
+
+    boost::system::error_code ec;
+    fs::remove_all(extended_path(long_dir), ec);
+    CHECK_FALSE(ec);
+}
+
 #ifndef _WIN32
 TEST_CASE("Confined extraction replaces a symlink at the destination instead of writing through it", "[MinizExtension]")
 {

tests/libslic3r/test_utils.cpp

@@ -521,3 +521,49 @@ TEST_CASE("ascii_iequals compares ASCII letters regardless of case", "[Utils]")
     CHECK_FALSE(ascii_iequals("[", "{"));
     CHECK_FALSE(ascii_iequals("@", "`"));
 }
+
+TEST_CASE("extended_path turns a Windows path into its long-path form", "[Utils]") {
+#ifdef _WIN32
+    namespace fs = boost::filesystem;
+
+    SECTION("a normal absolute drive path is prefixed") {
+        CHECK(extended_path(fs::path(R"(C:\models\model.3mf)")).wstring() == LR"(\\?\C:\models\model.3mf)");
+    }
+    SECTION("separators are made native") {
+        CHECK(extended_path(fs::path("C:/models/model.3mf")).wstring() == LR"(\\?\C:\models\model.3mf)");
+    }
+    SECTION("dot segments are collapsed before prefixing") {
+        // The "\\?\" prefix disables Win32 normalisation, so "." and ".." must be gone beforehand.
+        CHECK(extended_path(fs::path(R"(C:\models\..\model.3mf)")).wstring() == LR"(\\?\C:\model.3mf)");
+    }
+    SECTION("a trailing separator is dropped, not turned into a dot element") {
+        CHECK(extended_path(fs::path(R"(C:\models\)")).wstring() == LR"(\\?\C:\models)");
+    }
+    SECTION("a UNC path becomes an extended UNC path") {
+        CHECK(extended_path(fs::path(R"(\\server\share\model.3mf)")).wstring() == LR"(\\?\UNC\server\share\model.3mf)");
+    }
+    SECTION("an already extended path is returned unchanged") {
+        const fs::path extended(R"(\\?\C:\models\model.3mf)");
+        CHECK(extended_path(extended) == extended);
+    }
+    SECTION("a device path is returned unchanged") {
+        const fs::path device(R"(\\.\C:\models\model.3mf)");
+        CHECK(extended_path(device) == device);
+    }
+    SECTION("a bare device path without a root directory is returned unchanged") {
+        // Not "absolute" by Boost's rules, so it must be recognized before absolutizing.
+        const fs::path device(R"(\\.\PhysicalDrive0)");
+        CHECK(extended_path(device) == device);
+    }
+    SECTION("a relative path is completed against the working directory") {
+        const fs::path result = extended_path(fs::path("model.3mf"));
+        CHECK(result.wstring().rfind(L"\\\\?\\", 0) == 0);
+        CHECK(result.filename() == fs::path("model.3mf"));
+    }
+#else
+    SECTION("non-Windows returns the path unchanged") {
+        const boost::filesystem::path p = boost::filesystem::temp_directory_path() / "model.3mf";
+        CHECK(extended_path(p) == p);
+    }
+#endif
+}

tests/slic3rutils/CMakeLists.txt

@@ -32,6 +32,7 @@ add_executable(${_TEST_NAME}_tests
     test_printer_agent.cpp
     test_plugin_install.cpp
     test_plugin_lifecycle.cpp
+    test_plugin_resource_texture.cpp
     test_plugin_printer_agent.cpp
     test_printhost.cpp
     test_slicing_pipeline_bindings.cpp
@@ -44,7 +45,9 @@ add_executable(${_TEST_NAME}_tests
     test_texture_color_palette.cpp
     test_texture_island_view.cpp
     test_file_url.cpp
+    test_imgui_wrapper.cpp
     test_user_manager.cpp
+    test_sequential_clearance.cpp
     ../fff_print/test_helpers.cpp
     )
 

tests/slic3rutils/test_device_manager_integration.cpp

@@ -1,5 +1,6 @@
 #include <catch2/catch_test_macros.hpp>
 #include <catch2/catch_all.hpp>
+#include <catch2/generators/catch_generators.hpp>
 
 #include <slic3r/GUI/DeviceCore/DevManager.h>
 #include <slic3r/GUI/DeviceManager.hpp>
@@ -16,6 +17,7 @@
 #include <cstdint>
 #include <string>
 #include <utility>
+#include <vector>
 
 using namespace Slic3r;
 using json = nlohmann::json;
@@ -49,6 +51,21 @@ public:
 
     AgentInfo get_agent_info() override { return m_info; }
 
+    int send_message(std::string dev_id, std::string json_str, int, int) override
+    {
+        published_device = std::move(dev_id);
+        published_message = json::parse(json_str);
+        return 0;
+    }
+
+    int send_message_to_printer(std::string dev_id, std::string json_str, int qos, int flag) override
+    {
+        return send_message(std::move(dev_id), std::move(json_str), qos, flag);
+    }
+
+    std::string published_device;
+    json published_message;
+
 private:
     AgentInfo m_info;
 };
@@ -160,3 +177,62 @@ TEST_CASE("Device manager filters and rehomes devices by printer-agent ownership
     CHECK(manager.get_my_machine_list("integration-agent-a").empty());
     CHECK(manager.get_my_machine_list("integration-agent-b").count(machine.dev_id) == 1);
 }
+
+TEST_CASE("AMS filament settings keep legacy single-color commands compatible", "[DeviceManager][AMSMaterialsSetting][integration]")
+{
+    ScopedAppConfig app_config;
+    auto agent = std::make_shared<TestPrinterAgent>("integration-agent");
+    NetworkAgent network(nullptr, agent);
+    DeviceManager manager(&network, false, &app_config.config);
+    MachineObject object(&manager, &network, "Test printer", "integration-device", "192.0.2.10");
+    object.dev_connection_type = GENERATE("lan", "cloud");
+
+    REQUIRE(object.command_ams_filament_settings(1, 2, "filament-id", "setting-id", "FF000080", "PLA", 190, 220) == 0);
+
+    const json& payload = agent->published_message.at("print");
+    CHECK(agent->published_device == object.get_dev_id());
+    CHECK(payload.at("command") == "ams_filament_setting");
+    CHECK(payload.at("ams_id") == 1);
+    CHECK(payload.at("slot_id") == 2);
+    CHECK(payload.at("tray_color") == "FF000080");
+    CHECK_FALSE(payload.contains("cols"));
+    CHECK_FALSE(payload.contains("ctype"));
+}
+
+TEST_CASE("AMS filament settings preserve ordered gradient and multicolor RGBA components", "[DeviceManager][AMSMaterialsSetting][integration]")
+{
+    ScopedAppConfig app_config;
+    auto agent = std::make_shared<TestPrinterAgent>("integration-agent");
+    NetworkAgent network(nullptr, agent);
+    DeviceManager manager(&network, false, &app_config.config);
+    MachineObject object(&manager, &network, "Test printer", "integration-device", "192.0.2.10");
+    object.dev_connection_type = GENERATE("lan", "cloud");
+    const int color_type = GENERATE(0, 1);
+    const std::vector<std::string> colors{"FF000080", "0000FF40", "00FF00FF"};
+
+    REQUIRE(object.command_ams_filament_settings(1, 2, "filament-id", "setting-id", colors.front(), "PLA", 190, 220,
+                                                 colors, color_type) == 0);
+
+    const json& payload = agent->published_message.at("print");
+    CHECK(payload.at("tray_color") == colors.front());
+    CHECK(payload.at("cols").get<std::vector<std::string>>() == colors);
+    CHECK(payload.at("ctype") == color_type);
+}
+
+TEST_CASE("AMS filament settings default a supplied single-color list to type two", "[DeviceManager][AMSMaterialsSetting][integration]")
+{
+    ScopedAppConfig app_config;
+    auto agent = std::make_shared<TestPrinterAgent>("integration-agent");
+    NetworkAgent network(nullptr, agent);
+    DeviceManager manager(&network, false, &app_config.config);
+    MachineObject object(&manager, &network, "Test printer", "integration-device", "192.0.2.10");
+    object.dev_connection_type = GENERATE("lan", "cloud");
+    const std::vector<std::string> colors{"FF000080"};
+
+    REQUIRE(object.command_ams_filament_settings(1, 2, "filament-id", "setting-id", colors.front(), "PLA", 190, 220,
+                                                 colors) == 0);
+
+    const json& payload = agent->published_message.at("print");
+    CHECK(payload.at("cols").get<std::vector<std::string>>() == colors);
+    CHECK(payload.at("ctype") == 2);
+}

tests/slic3rutils/test_imgui_wrapper.cpp

@@ -0,0 +1,61 @@
+// How ImGuiWrapper turns wx mouse events into ImGui input. ImGuiWrapper lives in libslic3r_gui; this is
+// the suite that links it. Same Windows include prologue as test_texture_color_palette.cpp (wx pulls in
+// <windows.h>; keep WIN32_LEAN_AND_MEAN / NOMINMAX ahead of it).
+#ifdef WIN32
+    #ifndef WIN32_LEAN_AND_MEAN
+        #define WIN32_LEAN_AND_MEAN
+    #endif
+    #ifndef NOMINMAX
+        #define NOMINMAX
+    #endif
+    #include <Windows.h>
+#endif
+
+#include <catch2/catch_test_macros.hpp>
+#include <catch2/generators/catch_generators.hpp>
+#include <catch2/matchers/catch_matchers.hpp>
+#include <catch2/matchers/catch_matchers_floating_point.hpp>
+
+#include <imgui/imgui.h>
+#include <wx/event.h>
+
+#include "slic3r/GUI/ImGuiWrapper.hpp"
+
+using Catch::Matchers::WithinAbs;
+using Slic3r::GUI::ImGuiWrapper;
+
+namespace {
+
+// ImGui's wheel after one wheel event, as wx on macOS builds it: a delta of 10 and the rotation in whole
+// points of trackpad travel.
+float wheel_after(int rotation, wxMouseWheelAxis axis)
+{
+    ImGuiWrapper imgui;
+    imgui.set_display_size(800.f, 600.f); // update_mouse_data() ignores input until the display has a size
+
+    wxMouseEvent evt(wxEVT_MOUSEWHEEL);
+    evt.m_wheelRotation = rotation;
+    evt.m_wheelDelta    = 10;
+    evt.m_wheelAxis     = axis;
+    imgui.update_mouse_data(evt);
+    return ImGui::GetIO().MouseWheel;
+}
+
+} // namespace
+
+TEST_CASE("a vertical wheel event steps ImGui's wheel in its own direction", "[ImGuiWrapper]")
+{
+    const int rotation = GENERATE(1, 7, 120, -1, -7, -120);
+    CHECK_THAT(wheel_after(rotation, wxMOUSE_WHEEL_VERTICAL), WithinAbs(rotation > 0 ? 1.0 : -1.0, 0.0));
+}
+
+// Both kinds arrive from a macOS trackpad while scrolling; see ImGuiWrapper::update_mouse_data().
+TEST_CASE("a wheel event with no rotation or on the horizontal axis does not step ImGui's wheel", "[ImGuiWrapper][Regression]")
+{
+    const auto [rotation, axis] = GENERATE(table<int, wxMouseWheelAxis>({
+        { 0, wxMOUSE_WHEEL_VERTICAL },
+        { 5, wxMOUSE_WHEEL_HORIZONTAL },
+        { -5, wxMOUSE_WHEEL_HORIZONTAL },
+    }));
+    CHECK_THAT(wheel_after(rotation, axis), WithinAbs(0.0, 0.0));
+}

tests/slic3rutils/test_plugin_install.cpp

@@ -3,6 +3,7 @@
 #include <catch2/catch_all.hpp>
 
 #include <ios>
+#include <libslic3r/Utils.hpp>
 #include <slic3r/plugin/PluginLoader.hpp>
 #include <slic3r/plugin/PluginDescriptor.hpp>
 #include <slic3r/plugin/PluginFsUtils.hpp>
@@ -10,13 +11,17 @@
 #include <catch2/catch_test_macros.hpp>
 #include <catch2/matchers/catch_matchers.hpp>
 #include <catch2/matchers/catch_matchers_string.hpp>
+#include "catch2/catch_message.hpp"
 #include "plugin_test_utils.hpp"
+#include "test_utils.hpp"
 
 #include <boost/filesystem.hpp>
 
 #include <fstream>
 #include <string>
 
+#include <libslic3r/Zipper.hpp>
+
 using namespace Slic3r;
 namespace fs = boost::filesystem;
 
@@ -31,6 +36,14 @@ fs::path write_py_file(const fs::path& dir, const std::string& filename, const s
     return p;
 }
 
+// The shared production helper: makes the path absolute and lexically normalized and, on
+// Windows, prefixes it with the extended-length "\\?\" form, so the test can create, verify and
+// clean up a tree past MAX_PATH even with LongPathsEnabled off. Identity on other platforms.
+fs::path to_extended_path(const fs::path& p)
+{
+    return Slic3r::extended_path(p);
+}
+
 } // namespace
 
 TEST_CASE("install_plugin rejects a cloud UUID containing path traversal", "[PluginInstall]")
@@ -184,3 +197,59 @@ TEST_CASE("an unsubscribed cloud plugin follows its header version", "[PluginIns
     REQUIRE(read_install_state(plugin_dir, scanned));
     CHECK(scanned.installed_version == "1.0.0");
 }
+
+TEST_CASE("extract_zip_to_directory writes entries under a destination longer than MAX_PATH",
+          "[PluginInstall][PluginFsUtils]")
+{
+    ScopedTemporaryDir dir("deep-extract");
+
+    const fs::path wheel = dir.path() / "deep_pkg-1.0-py3-none-any.whl";
+    const std::string payload = "print('hello from a deep path')\n";
+    {
+        Zipper zipper(wheel.string(), Zipper::NO_COMPRESSION);
+        zipper.add_entry("pkg/", "", 0); // explicit directory entry
+        zipper.add_entry("pkg/module.py", payload.data(), payload.size());
+        zipper.finalize();
+    }
+
+    // A single 240-char component keeps every filename below the NTFS 255 limit while pushing the
+    // full destination (temp dir + component) past the 260-char MAX_PATH the Win32 APIs enforce
+    // when LongPathsEnabled is off.
+    const fs::path destination = dir.path() / std::string(240, 'd');
+    REQUIRE(destination.string().size() > 260);
+
+    std::string error;
+    REQUIRE(extract_zip_to_directory(wheel, destination, error));
+    INFO("extract error: " << error);
+    CHECK(error.empty());
+
+    // Verify through a long-path-capable path, otherwise the check itself is bounded by MAX_PATH.
+    const fs::path module_py = destination / "pkg" / "module.py";
+    REQUIRE(fs::exists(to_extended_path(module_py)));
+    REQUIRE(fs::is_regular_file(to_extended_path(module_py)));
+
+    // Remove the deep subtree here: ScopedTemporaryDir's plain remove_all cannot reach past
+    // MAX_PATH on Windows.
+    boost::system::error_code ec;
+    fs::remove_all(to_extended_path(destination), ec);
+    CHECK_FALSE(ec);
+}
+
+TEST_CASE("extract_zip_to_directory rejects an entry that escapes the destination",
+          "[PluginInstall][PluginFsUtils]")
+{
+    ScopedTemporaryDir dir("escape-extract");
+
+    const fs::path wheel = dir.path() / "evil-1.0-py3-none-any.whl";
+    {
+        Zipper zipper(wheel.string(), Zipper::NO_COMPRESSION);
+        zipper.add_entry("pkg/module.py", "print('ok')\n", 12);
+        zipper.add_entry("../escape.txt", "escaped\n", 8);
+        zipper.finalize();
+    }
+
+    std::string error;
+    CHECK_FALSE(extract_zip_to_directory(wheel, dir.path() / "out", error));
+    CHECK_THAT(error, Catch::Matchers::ContainsSubstring("escapes the plugin package"));
+    CHECK_FALSE(fs::exists(dir.path() / "escape.txt"));
+}