juli/OrcaSlicer
1
1
Fork 0
mirror of https://github.com/OrcaSlicer/OrcaSlicer.git synced 2026-05-16 18:12:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

Optimize and simplify MarchingSquares.hpp. (#10747)

Browse Source 
* Optimize and simplify MarchingSquares.hpp, and fix it's test.

This changes the implementation to get the possible next directions for a cell
when building the tags and clearing them as the cells are visited during the
march, instead of adding the visited previous direction to the tags during the
march. The Dir enum has been turned into bit flags that for the possible next
directions with boolean operators for testing/setting/clearing them. This
simplifies and optimizes many operations during the march and building the
polygons.

The complicated/broken and unused partial support for cell overlap has been
removed, simplifying the overly confusing grid iteration logic.

The broken test has been fixed by removing the now gone `RasterBase` namespace
from `sla::RasterBase::Pixeldim` and `sla:RasterBase:Resolution`, and the
CMakeLists.txt entry uncommented.

make Dir into flags

* Further optimize MarchingSquares.hpp and improve comments.

* Switch from a single byte-vector containing tags and dirs for each cell to a
  m_tags vector of bit-packed tags for each grid corner and an m_dirs vector
  of packed 4bit dirs for each cell. Since each grid corner tag is shared by
  the 4 adjacent cells this significantly reduces storage space and avoids
  redundantly calculating each tag 4x. It also significantly improves memory
  locality with each phase of calculating tags, calculating dirs, calculating
  rings operating only on the tags or dirs data required without them being
  interleaved with the data they don't need.

* Change NEXT_CCW to be initialized with a static constexpr lambda instead of
  a manually entered table. This avoids typo errors manually building the
  table.

* Optimize search_start_cell() so it can efficiently skip over cleared blocks
  of 8 dirs in the packed m_dirs vector.

* Change the tags logical labeling to better suit the packed tags vector data.
  This makes it a tiny bit more efficient to extract from the m_tags bitmap.

* Remove the now unused SquareTag enum class.

* Add comments explaining the algorithm, including corner-cases in cell
  iteration.

* Remove unused Dir operators and get_dirs() argument, and clang-format.

* Fix some bugs and add stream output operators for debugging.

* Fix a bug building tags where `step(gcrd, Dir::right)` was not assigned to
  update the gcrd grid point. Perhaps this should be a mutating method, or
  even a += operator? Also when wrapping at the end of a row it was updating
  the gcrd grid point by mutating the p raster point instead of itself.
  Perhaps Grid and Raster points should be different types? Maybe even
  templated?

* Fix a bug in get_tags() when the second row tags are packed into any of the
  2 LSB's of the uint32_t blocks. In hind-sight obviously `>>(o - 2)` will not
  shift left when `o < 2`.

* Move interpolation of the edge-crossings into a `interpolate()` method, and
  make it shift bottom and right side points "out" by one to account for
  raster pixel width. This makes the results track the raster shapes much more
  accurately for very small windows.

* Make `interpolate_rings()` check for and remove duplicated points. It turns
  out it's pretty common that two edge-crossing-points at a corner interpolate
  to the same point. This can also happen for the first and last points.

* For Coord add `==` and `!=` operators, and use them wherever Coord's are
  compared.

* Add `<<` stream output operators for Coord, Ring, and Dir classes. Add
  `streamtags(<stream>)` and `streamdirs(<stream>)` methods for dumping the
  tags and dirs data in an easy to understand text format. These make
  print-debugging much easier.

* Add `assert(idx < m_gridlen)` in a bunch of places where grid-indexes are
  used.

* For test_clipper_utils.cpp fix three "ambiguous overloading" compiler errors.

This just adds three `Polygons` qualifications to fix compiler errors about
ambiguous overloaded methods.

Note this file was formated with a mixture of tabs and spaces and had lots of
trailing whitespace. My editor cleaned these up resulting in a large looking
diff, but if you use `git diff -w` to ignore the whitespace changes you will
see it is actually tiny.

errros

* Update SLA/RasterToPolygons.* for MarchingSquares.hpp improvements.

Change the minimum and default window size from 2x2 to 1x1. Also remove the
strange pixel size re-scaling by (resolution/resolution-1).

The old MarchingSquares implementation had complications around a default
minimum 1 pixel "overlap" between cells which messed with the scaling a tiny
bit and meant when you requested a 2x2 window size it actually used a 1x1
window. Both of these meant you had to specify a window 1 pixel larger than
you really wanted, and you needed to undo the strange scaling artifact for
accurate dimensions of your results.

This has been fixed/removed in the new implementation, so the window is the
window, there is no overlap, and no strange miss-scaling.

* Fix test_marchingsquares.cpp and add StreamUtils.hpp.

This fixes the MarchingSquares unittests to both pass and be more strict than
they were before.

It also adds libslic3r/StreamUtils.hpp which includes some handy streaming
operators for standard libslic3r classes used to show extracted polys in the
unittests.

* Change Format/SL1.cpp to support the min 1x1 window for MarchingSquares.

* Fix the ring-walk termination condition.

Terminate the ring-walk when we return to the starting cell instead of when we
reach a cell with no remaining directions. This ensures we don't merge two
polygons if we started on an ambiguous cell.

* Revert the removal of duplicate points in interpolate_rings().

It turns out that duplicate points are only relatively common when using a 1x1
window. These happen when the line passes through the corner pixel on a
top-left corner in the raster, and the probability of this rapidly declines as
the window increases, so in many cases this filtering is just overhead. It can
also be potentially useful to see the points for every edge crossing even if
they are duplicates. This kind of filtering is already done and done better in
the polygon post-processing.

* rename `interpolate()` to `interpolate_edge()`, make it update the point
  in-place, and add asserts to ensure the input point is a valid edge
  interpolation point.
* Remove the duplicate point filtering from `interpolate_rings()` and simplify
  it.

* Optimize directions building.

This optimizes `get_dirs_block8()` to rapidly skip over blocks where the tags
produce no directions (all tags are 1's or 0's), and also to build the
directions faster when it has to by fetching the whole blocks worth of tags at
once instead of cell-by-cell.

* Rename `get_tags()` to `get_tags9()` and make it fetch a row of nine tags
  instead of the tags for a single cell.

* Optimize `get_dirs_block8()` to use `get_tags9()` to get the next nine tags
  for the current and next rows and then shift through them to generate the
  tags and directions for each cell in the block. Also abort early and just
  return an empty block if the tags are all 0's or all 1's.

* Tiny optimization for `get_tags_block32()`.

This avoids using the `step()` method for a simple step-right that can be done
with a simple increment of the column. It also avoids re-calculating the
raster-coodinates for every corner, instead incrementing the column by
`m_window.c` until the end of a row.

* Fix svg output in test_marchingsquares.cpp for recreate_object_from_rasters.

These SVG's were not properly centered...

* Fix 2 static_casts for compiling on Windows.

Thanks to RF47 for pointing this out on the #10747 pull request.

* Make edge iteration use O(ln(N)) binary search instead of linear.

This should be much faster when the window size is large.

* Make `CellIt` into a `std::random_access_iterator_tag` so that
`std::lower_bound()` can use a binary search to find the point on the edge
instead of a linear search.

* Change `step()` to support an optional distance argument and make it modify
the `Coord` in-place instead of return a new one.

* Update tests for the `step()` change.

* Add Catch2 BENCHMARK tests for MarchingSquares.

This required enabling the benchmarks in the tests/CMakeLists.txt config.

* Add a _Loop<> specialization for parallel execution using ExecutionTBB.

This is something that could be added wherever you are going to use this, but
I intend on using this in multiple places so we might as add this once in one
place where it can be reused.

* Fix whitespace in messed up by tab-replacements.

My editor renders, and replaces, tabs as 8 spaces. This messed up the
indenting in tests/libslic3r/CMakeLists.txt and
tests/libslic3r/test_clipper_utils.cpp when I made tiny changes in them.
This fixes the indenting using 4 chars. Note it will still show as a diff
because it is replacing tabs with 4 spaces, and removing trailing whitespace.

But at least it's now indented correctly...

---------

Co-authored-by: Donovan Baarda <dbaarda@google.com>
Co-authored-by: SoftFever <softfeverever@gmail.com>
This commit is contained in:
Donovan Baarda
2025-10-19 22:27:55 +11:00
committed by GitHub
parent f12602c0bc
commit 2a3e761ab9
9 changed files with 1172 additions and 715 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/libslic3r/Format/SL1.cpp
View File

@@ -305,8 +305,8 @@ ConfigSubstitutions import_sla_archive(
std::function<bool(int)> progr)
{
// Ensure minimum window size for marching squares
windowsize.x() = std::max(2, windowsize.x());
windowsize.y() = std::max(2, windowsize.y());
windowsize.x() = std::max(1, windowsize.x());
windowsize.y() = std::max(1, windowsize.y());
std::string exclude_entries{"thumbnail"};
ArchiveData arch = extract_sla_archive(zipfname, exclude_entries);

880
src/libslic3r/MarchingSquares.hpp
View File

@@ -1,49 +1,160 @@
#ifndef MARCHINGSQUARES_HPP
#define MARCHINGSQUARES_HPP
#include "Execution/ExecutionTBB.hpp"
#include <type_traits>
#include <cstdint>
#include <vector>
#include <algorithm>
#include <cassert>
// Marching squares
//
// This algorithm generates 2D contour rings for a 2D scalar field (height
// map). See https://en.wikipedia.org/wiki/Marching_squares for details.
//
// This algorithm uses square cells (tiles) with corner tags that are set
// depending on which vertices are inside an ROI (higher than the selected
// isovalue threshold). We label the vertices counter-clockwise, and for ease
// of extraction from the m_tags bitmap we put them into a four bit bitmap in
// little-endian row-column order <cbda>;
//
// ^ u (up) ----> column "c" direction
// |
// a ----- d
// | | |
// l <--| |--> r | increasing row "r" direction
// (left) | | (right) \/
// b ----- c
// |
// \/ d (down)
//
// The m_tags bitmap has the top-left "a" corner of each cell packed
// into uint32_t values stored in little-endian order, with the "a" tag for
// the first cell (r=0,c=0) in the LSB bit0 of the first 32bit word of the
// first row. The grid has a one cell border round the raster area that is
// always clear so lines never extend out of the grid.
//
// From this grid of points we can get the <cbda> tags bitmap for the four
// corners of each cell with "a" in the LSB bit0 and "c" in bit3. Note the
// bits b,c,d are also the "a" bits of the cell's next-colum/row neighbour
// cells. Cells in the right-most column get their cleared right-side border
// <c_b_> tags by wrapping around the grid and getting the cleared values
// from the left-side border. Note the grid bitmap always includes at least
// one extra cleared boarder bit for the wrap-around of the last tag in the
// last cell. Cells in the bottom row get their cleared bottom-side border
// <cb__> tags by leaving them clear because their grid index exceeds the
// grid size.
//
// These tag bits are then translated into up/down/left/right cell-exit
// directions indicating the cell edge(s) that any lines will cross and exit
// the cell. As you walk counter-clock-wise around the cell corners in the
// order a->b->c->d, any edge from a unset-corner-tag to an set-corner-tag is
// where a line will cross, exiting the cell into the next cell.
//
// Note any set-to-unset-transition edge is where a line will enter a cell,
// but for marching the lines through the cells we only need to keep the next
// cell exit directions. The exit directions are stored as 4bit <urdl>
// left/down/right/up flags packed into uint32_t values in m_dirs, in
// counter-clockwise little-endian order.
//
// To build the line-rings we scan through the m_dirs directions for the
// cells until we find one with a set-edge. We then start a line and "march"
// it through the cells following the directions in each cell, recording each
// edge the line crosses that will become a point in the ring.
//
// Note that the outside-edge of the grid extends one cell past the raster
// grid, and this outer edge is always cleared. This means no line will cross
// out of the grid, and they will all eventually connect back to their
// starting cell to become a ring around an island of set tag bits, possibly
// containing rings around lakes of unset tag bits within them.
//
// As we march through the cells and build the lines, we clear the direction
// bits to indicate the edge-transition has been processed. After completing
// a ring we return to scanning for the next non-zeroed starting cell for the
// next ring. Eventually all rings will be found and completed, and the
// m_dirs map will be completely cleared.
//
// Note there are two ambigous cases when the diagonally opposite corner tags
// are set. In these cases two lines will enter and exit the cell. To avoid
// messy lines, we consistently choose the exit direction based on what edge
// the line entered the cell from. This also means when we start scanning for
// the next ring we should start checking at the same cell the last ring
// started from, because it could have a second ring passing through it.
//
namespace marchsq {
// Marks a square in the grid
struct Coord {
// Marks a square or point in grid or raster coordintes.
struct Coord
{
long r = 0, c = 0;
Coord() = default;
explicit Coord(long s) : r(s), c(s) {}
Coord(long _r, long _c): r(_r), c(_c) {}
size_t seq(const Coord &res) const { return r * res.c + c; }
Coord& operator+=(const Coord& b) { r += b.r; c += b.c; return *this; }
Coord operator+(const Coord& b) const { Coord a = *this; a += b; return a; }
Coord(long _r, long _c) : r(_r), c(_c) {}
size_t seq(const Coord& res) const { return r * res.c + c; }
Coord& operator+=(const Coord& b)
{
r += b.r;
c += b.c;
return *this;
}
Coord operator+(const Coord& b) const
{
Coord a = *this;
a += b;
return a;
}
bool operator==(const Coord& o) const { return (r == o.r) && (c == o.c); }
bool operator!=(const Coord& o) const { return !(*this == o); }
};
inline std::ostream& operator<<(std::ostream& os, const Coord& o) { return os << "(r=" << o.r << ",c=" << o.c << ")"; }
// Closed ring of cell coordinates
using Ring = std::vector<Coord>;
// Specialize this struct to register a raster type for the Marching squares alg
template<class T, class Enable = void> struct _RasterTraits {
inline std::ostream& operator<<(std::ostream& os, const Ring& r)
{
os << "[" << r.size() << "]:";
for (Coord c : r)
os << " " << c;
return os << "\n";
}
// Specialize this struct to register a raster type for MarchingSquares.
template<class T, class Enable = void> struct _RasterTraits
{
// The type of pixel cell in the raster
using ValueType = typename T::ValueType;
// Value at a given position
static ValueType get(const T &raster, size_t row, size_t col);
static ValueType get(const T& raster, size_t row, size_t col);
// Number of rows and cols of the raster
static size_t rows(const T &raster);
static size_t cols(const T &raster);
static size_t rows(const T& raster);
static size_t cols(const T& raster);
};
// Specialize this to use parellel loops within the algorithm
template<class ExecutionPolicy, class Enable = void> struct _Loop {
template<class It, class Fn> static void for_each(It from, It to, Fn &&fn)
template<class ExecutionPolicy, class Enable = void> struct _Loop
{
template<class It, class Fn> static void for_each_idx(It from, It to, Fn&& fn)
{
for (auto it = from; it < to; ++it) fn(*it, size_t(it - from));
for (auto it = from; it < to; ++it)
fn(*it, size_t(it - from));
}
};
// Add Specialization for using ExecutionTBB for parallel loops.
using namespace Slic3r;
template<> struct _Loop<ExecutionTBB>
{
template<class It, class Fn> static void for_each_idx(It from, It to, Fn&& fn)
{
execution::for_each(
ex_tbb, size_t(0), size_t(to - from), [&from, &fn](size_t i) { fn(from[i], i); }, execution::max_concurrency(ex_tbb));
}
};
@@ -52,388 +163,481 @@ namespace __impl {
template<class T> using RasterTraits = _RasterTraits<std::decay_t<T>>;
template<class T> using TRasterValue = typename RasterTraits<T>::ValueType;
template<class T> size_t rows(const T &raster)
template<class T> size_t rows(const T& raster) { return RasterTraits<T>::rows(raster); }
template<class T> size_t cols(const T& raster) { return RasterTraits<T>::cols(raster); }
template<class T> TRasterValue<T> isoval(const T& rst, const Coord& crd) { return RasterTraits<T>::get(rst, crd.r, crd.c); }
template<class ExecutionPolicy, class It, class Fn> void for_each_idx(ExecutionPolicy&& policy, It from, It to, Fn&& fn)
{
return RasterTraits<T>::rows(raster);
_Loop<ExecutionPolicy>::for_each_idx(from, to, fn);
}
template<class T> size_t cols(const T &raster)
{
return RasterTraits<T>::cols(raster);
}
template<class E> constexpr std::underlying_type_t<E> _t(E e) noexcept { return static_cast<std::underlying_type_t<E>>(e); }
template<class T> TRasterValue<T> isoval(const T &rst, const Coord &crd)
{
return RasterTraits<T>::get(rst, crd.r, crd.c);
}
template<class ExecutionPolicy, class It, class Fn>
void for_each(ExecutionPolicy&& policy, It from, It to, Fn &&fn)
{
_Loop<ExecutionPolicy>::for_each(from, to, fn);
}
// Type of squares (tiles) depending on which vertices are inside an ROI
// The vertices would be marked a, b, c, d in counter clockwise order from the
// bottom left vertex of a square.
// d --- c
// | |
// | |
// a --- b
enum class SquareTag : uint8_t {
// 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
none, a, b, ab, c, ac, bc, abc, d, ad, bd, abd, cd, acd, bcd, full
// A set of next direction flags for a cell to indicate what sides(s)
// any yet-to-be-marched line(s) will cross to leave the cell.
enum class Dir : uint8_t {
none = 0b0000,
left = 0b0001, /* exit through a-b edge */
down = 0b0010, /* exit through b-c edge */
right = 0b0100, /* exit through c-d edge */
leftright = left | right,
up = 0b1000, /* exit through d-a edge */
updown = up | down,
all = 0b1111
};
template<class E> constexpr std::underlying_type_t<E> _t(E e) noexcept
inline std::ostream& operator<<(std::ostream& os, const Dir& d) { return os << ".<v#>x##^#X#####"[_t(d)]; }
// This maps square tag column/row order <cbda> bitmaps to the next
// direction(s) a line will exit the cell. The directions ensure the
// line-rings circle counter-clock-wise around the clusters of set tags.
static const constexpr std::array<uint32_t, 16> NEXT_CCW = [] {
auto map = decltype(NEXT_CCW){};
for (uint32_t cbda = 0; cbda < 16; cbda++) {
const uint32_t dcba = (cbda & 0b0001) | ((cbda >> 1) & 0b0110) | ((cbda << 2) & 0b1000);
const uint32_t adcb = (dcba >> 1) | ((dcba << 3) & 0b1000);
map[cbda] = ~dcba & adcb;
}
return map;
}();
// Step a point in a direction, optionally by n steps.
inline void step(Coord& crd, const Dir d, const long n = 1)
{
return static_cast<std::underlying_type_t<E>>(e);
switch (d) {
case Dir::left: crd.c -= n; break;
case Dir::down: crd.r += n; break;
case Dir::right: crd.c += n; break;
case Dir::up: crd.r -= n; break;
}
}
enum class Dir: uint8_t { left, down, right, up, none};
static const constexpr Dir NEXT_CCW[] = {
/* 00 */ Dir::none, // SquareTag::none (empty square, nowhere to go)
/* 01 */ Dir::left, // SquareTag::a
/* 02 */ Dir::down, // SquareTag::b
/* 03 */ Dir::left, // SquareTag::ab
/* 04 */ Dir::right, // SquareTag::c
/* 05 */ Dir::none, // SquareTag::ac (ambiguous case)
/* 06 */ Dir::down, // SquareTag::bc
/* 07 */ Dir::left, // SquareTag::abc
/* 08 */ Dir::up, // SquareTag::d
/* 09 */ Dir::up, // SquareTag::ad
/* 10 */ Dir::none, // SquareTag::bd (ambiguous case)
/* 11 */ Dir::up, // SquareTag::abd
/* 12 */ Dir::right, // SquareTag::cd
/* 13 */ Dir::right, // SquareTag::acd
/* 14 */ Dir::down, // SquareTag::bcd
/* 15 */ Dir::none // SquareTag::full (full covered, nowhere to go)
};
static const constexpr uint8_t PREV_CCW[] = {
/* 00 */ 1 << _t(Dir::none),
/* 01 */ 1 << _t(Dir::up),
/* 02 */ 1 << _t(Dir::left),
/* 03 */ 1 << _t(Dir::left),
/* 04 */ 1 << _t(Dir::down),
/* 05 */ 1 << _t(Dir::up) | 1 << _t(Dir::down),
/* 06 */ 1 << _t(Dir::down),
/* 07 */ 1 << _t(Dir::down),
/* 08 */ 1 << _t(Dir::right),
/* 09 */ 1 << _t(Dir::up),
/* 10 */ 1 << _t(Dir::left) | 1 << _t(Dir::right),
/* 11 */ 1 << _t(Dir::left),
/* 12 */ 1 << _t(Dir::right),
/* 13 */ 1 << _t(Dir::up),
/* 14 */ 1 << _t(Dir::right),
/* 15 */ 1 << _t(Dir::none)
};
const constexpr uint8_t DIRMASKS[] = {
/*left: */ 0x01, /*down*/ 0x12, /*right */0x21, /*up*/ 0x10, /*none*/ 0x00
};
inline Coord step(const Coord &crd, Dir d)
template<class Rst> class Grid
{
uint8_t dd = DIRMASKS[uint8_t(d)];
return {crd.r - 1 + (dd & 0x0f), crd.c - 1 + (dd >> 4)};
}
const Rst* m_rst = nullptr;
Coord m_window, m_rastsize, m_gridsize;
size_t m_gridlen; // The number of cells in the grid.
std::vector<uint32_t> m_tags; // bit-packed squaretags for each corner.
std::vector<uint32_t> m_dirs; // bit-packed next directions for each cell.
template<class Rst> class Grid {
const Rst * m_rst = nullptr;
Coord m_cellsize, m_res_1, m_window, m_gridsize, m_grid_1;
std::vector<uint8_t> m_tags; // Assign tags to each square
Coord rastercoord(const Coord &crd) const
// Convert a grid coordinate point into raster coordinates.
inline Coord rastercoord(const Coord& crd) const
{
// Note the -1 offset for the grid border around the raster area.
return {(crd.r - 1) * m_window.r, (crd.c - 1) * m_window.c};
}
Coord bl(const Coord &crd) const { return tl(crd) + Coord{m_res_1.r, 0}; }
Coord br(const Coord &crd) const { return tl(crd) + Coord{m_res_1.r, m_res_1.c}; }
Coord tr(const Coord &crd) const { return tl(crd) + Coord{0, m_res_1.c}; }
Coord tl(const Coord &crd) const { return rastercoord(crd); }
bool is_within(const Coord &crd)
{
long R = rows(*m_rst), C = cols(*m_rst);
return crd.r >= 0 && crd.r < R && crd.c >= 0 && crd.c < C;
};
// Get the 4 corners of a grid coordinate cell in raster coordinates.
Coord bl(const Coord& crd) const { return tl(crd) + Coord{m_window.r, 0}; }
Coord br(const Coord& crd) const { return tl(crd) + Coord{m_window.r, m_window.c}; }
Coord tr(const Coord& crd) const { return tl(crd) + Coord{0, m_window.c}; }
Coord tl(const Coord& crd) const { return rastercoord(crd); }
// Calculate the tag for a cell (or square). The cell coordinates mark the
// top left vertex of a square in the raster. v is the isovalue
uint8_t get_tag_for_cell(const Coord &cell, TRasterValue<Rst> v)
{
Coord sqr[] = {bl(cell), br(cell), tr(cell), tl(cell)};
uint8_t t = ((is_within(sqr[0]) && isoval(*m_rst, sqr[0]) >= v)) +
((is_within(sqr[1]) && isoval(*m_rst, sqr[1]) >= v) << 1) +
((is_within(sqr[2]) && isoval(*m_rst, sqr[2]) >= v) << 2) +
((is_within(sqr[3]) && isoval(*m_rst, sqr[3]) >= v) << 3);
assert(t < 16);
return t;
}
// Get a cell coordinate from a sequential index
Coord coord(size_t i) const
{
return {long(i) / m_gridsize.c, long(i) % m_gridsize.c};
}
// Test if a raster coordinate point is within the raster area.
inline bool is_within(const Coord& crd) const { return crd.r >= 0 && crd.r < m_rastsize.r && crd.c >= 0 && crd.c < m_rastsize.c; }
size_t seq(const Coord &crd) const { return crd.seq(m_gridsize); }
bool is_visited(size_t idx, Dir d = Dir::none) const
// Get a block of 32 tags for a block index from the raster isovals.
uint32_t get_tags_block32(const size_t bidx, const TRasterValue<Rst> v) const
{
SquareTag t = get_tag(idx);
uint8_t ref = d == Dir::none ? PREV_CCW[_t(t)] : uint8_t(1 << _t(d));
return t == SquareTag::full || t == SquareTag::none ||
((m_tags[idx] & 0xf0) >> 4) == ref;
}
void set_visited(size_t idx, Dir d = Dir::none)
{
m_tags[idx] |= (1 << (_t(d)) << 4);
}
bool is_ambiguous(size_t idx) const
{
SquareTag t = get_tag(idx);
return t == SquareTag::ac || t == SquareTag::bd;
}
// Search for a new starting square
size_t search_start_cell(size_t i = 0) const
{
// Skip ambiguous tags as starting tags due to unknown previous
// direction.
while ((i < m_tags.size()) && (is_visited(i) || is_ambiguous(i))) ++i;
return i;
}
SquareTag get_tag(size_t idx) const { return SquareTag(m_tags[idx] & 0x0f); }
Dir next_dir(Dir prev, SquareTag tag) const
{
// Treat ambiguous cases as two separate regions in one square.
switch (tag) {
case SquareTag::ac:
switch (prev) {
case Dir::down: return Dir::right;
case Dir::up: return Dir::left;
default: assert(false); return Dir::none;
Coord gcrd = coord(bidx * 32); // position in grid coordinates of the block.
Coord rcrd = rastercoord(gcrd); // position in raster coordinates.
uint32_t tags = 0;
// Set a bit for each corner that has osoval > v.
for (uint32_t b = 1; b; b <<= 1) {
if (is_within(rcrd) && isoval(*m_rst, rcrd) > v)
tags |= b;
gcrd.c += 1;
rcrd.c += m_window.c;
// If we hit the end of the row, start on the next row.
if (gcrd.c >= m_gridsize.c) {
gcrd = Coord(gcrd.r + 1, 0);
rcrd = rastercoord(gcrd);
}
case SquareTag::bd:
switch (prev) {
case Dir::right: return Dir::up;
case Dir::left: return Dir::down;
default: assert(false); return Dir::none;
}
default:
return NEXT_CCW[uint8_t(tag)];
}
return Dir::none;
return tags;
}
struct CellIt {
Coord crd; Dir dir= Dir::none; const Rst *grid = nullptr;
TRasterValue<Rst> operator*() const { return isoval(*grid, crd); }
CellIt& operator++() { crd = step(crd, dir); return *this; }
CellIt operator++(int) { CellIt it = *this; ++(*this); return it; }
bool operator!=(const CellIt &it) { return crd.r != it.crd.r || crd.c != it.crd.c; }
// Get a block of 8 directions for a block index from the tags.
uint32_t get_dirs_block8(const size_t bidx) const
{
size_t gidx = bidx * 8;
uint32_t dirs = 0;
// Get the next 9 top-row tags at this grid index into the bottom 16
// bits, and the 9 bottom-row tags into the top 16 bits.
uint32_t tags9 = get_tags9(gidx) | (get_tags9(gidx + m_gridsize.c) << 16);
// Skip generating dirs if the tags are all 1's or all 0's.
if ((tags9 != 0) && (tags9 != 0x01ff01ff)) {
for (auto s = 0; s < 32; s += 4) {
uint8_t tags = (tags9 & 0b11) | ((tags9 >> 14) & 0b1100);
dirs |= NEXT_CCW[tags] << s;
tags9 >>= 1;
}
}
return dirs;
}
// Get the next 9 corner tags on a row for building a dirs block at a grid index.
uint32_t get_tags9(size_t gidx) const
{
uint32_t tags = 0; // the tags value.
size_t i = gidx / 32; // the tags block index
int o = gidx % 32; // the tags block offset
if (gidx < m_gridlen) {
// get the next 9 tags in the row.
tags = (m_tags[i] >> o) & 0x1ff;
// Some of the tags are in the next tags block.
if ((o > (32 - 9)) && ((i + 1) < m_tags.size())) {
tags |= (m_tags[i + 1] << (32 - o)) & 0x1ff;
}
}
return tags;
}
// Clear directions in a cell's <urdl> dirs for a grid index.
void clr_dirs(const size_t gidx, const Dir d)
{
assert(gidx < m_gridlen);
size_t i = gidx / 8; // the dirs block index
int o = (gidx % 8); // the dirs block offset
m_dirs[i] &= ~(static_cast<uint32_t>(d) << (o * 4));
}
// Get directions in a cell's <uldr> dirs from the m_dirs store.
Dir get_dirs(const size_t gidx) const
{
assert(gidx < m_gridlen);
size_t i = gidx / 8; // the dirs block index
int o = (gidx % 8); // the dirs block offset
return Dir((m_dirs[i] >> (o * 4)) & 0b1111);
}
// Get a cell coordinate from a sequential grid index.
Coord coord(size_t i) const { return {long(i) / m_gridsize.c, long(i) % m_gridsize.c}; }
// Get a sequential index from a cell coordinate.
size_t seq(const Coord& crd) const { return crd.seq(m_gridsize); }
// Step a sequential grid index in a direction.
size_t stepidx(const size_t idx, const Dir d) const
{
assert(idx < m_gridlen);
switch (d) {
case Dir::left: return idx - 1;
case Dir::down: return idx + m_gridsize.c;
case Dir::right: return idx + 1;
case Dir::up: return idx - m_gridsize.c;
default: return idx;
}
}
// Search for a new starting square.
size_t search_start_cell(size_t gidx = 0) const
{
size_t i = gidx / 8; // The dirs block index.
int o = gidx % 8; // The dirs block offset.
// Skip cells without any unvisited edges.
while (i < m_dirs.size()) {
if (!m_dirs[i]) {
// Whole block is clear, advance to the next block;
i++;
o = 0;
} else {
// Block not clear, find the next non-zero tags in the block. Note
// all dirs before gidx are cleared, so any set bits must be in
// block offsets >= o, not before.
for (uint32_t m = 0b1111 << (o * 4); !(m_dirs[i] & m); m <<= 4)
o++;
break;
}
}
return i * 8 + o;
}
// Get the next direction for a cell index after the prev direction.
Dir next_dir(size_t idx, Dir prev = Dir::all) const
{
Dir next = get_dirs(idx);
// Treat ambiguous cases as two separate regions in one square. If
// there are two possible next directions, pick based on the prev
// direction. If prev=all we are starting a new line so pick the one
// that leads "forwards" (right or down) in the search.
switch (next) {
case Dir::leftright:
// We must be coming from up, down, or starting a new line.
assert(prev == Dir::up || prev == Dir::down || prev == Dir::all);
return (prev == Dir::up) ? Dir::left : Dir::right;
case Dir::updown:
// We must be coming from left, right, or starting a new line.
assert(prev == Dir::left || prev == Dir::right || prev == Dir::all);
return (prev == Dir::right) ? Dir::up : Dir::down;
default:
// Next must be a single direction or none to stop.
assert(next == Dir::none || next == Dir::left || next == Dir::down || next == Dir::right || next == Dir::up);
return next;
}
}
struct CellIt
{
using iterator_category = std::random_access_iterator_tag;
using value_type = TRasterValue<Rst>;
using pointer = TRasterValue<Rst> *;
using reference = TRasterValue<Rst> &;
using pointer = TRasterValue<Rst>*;
using reference = TRasterValue<Rst>&;
using difference_type = long;
using iterator_category = std::forward_iterator_tag;
Coord crd;
Dir dir = Dir::none;
const Rst* grid = nullptr;
inline TRasterValue<Rst> operator*() const { return isoval(*grid, crd); }
inline TRasterValue<Rst>& operator[](long n) const { return *((*this) + n); }
inline CellIt& operator++()
{
step(crd, dir);
return *this;
}
inline CellIt operator++(int)
{
CellIt o = *this;
++(*this);
return o;
}
inline CellIt& operator--()
{
step(crd, dir, -1);
return *this;
}
inline CellIt operator--(int)
{
CellIt o = *this;
--(*this);
return o;
}
inline CellIt& operator+=(long n)
{
step(crd, dir, n);
return *this;
}
inline CellIt& operator-=(long n)
{
step(crd, dir, -n);
return *this;
}
inline CellIt operator+(long n) const
{
CellIt o = *this;
o += n;
return o;
}
inline CellIt operator-(long n) const
{
CellIt o = *this;
o -= n;
return o;
}
inline long operator-(const CellIt& o) const
{
switch (dir) {
case Dir::left: return o.crd.c - crd.c;
case Dir::down: return crd.r - o.crd.r;
case Dir::right: return crd.c - o.crd.c;
case Dir::up: return o.crd.r - crd.r;
default: return 0;
}
}
inline bool operator==(const CellIt& o) const { return crd == o.crd; }
inline bool operator!=(const CellIt& o) const { return crd != o.crd; }
inline bool operator<(const CellIt& o) const { return (*this - o) < 0; }
inline bool operator>(const CellIt& o) const { return (*this - o) > 0; }
inline bool operator<=(const CellIt& o) const { return (*this - o) <= 0; }
inline bool operator>=(const CellIt& o) const { return (*this - o) >= 0; }
};
// Two cell iterators representing an edge of a square. This is then
// used for binary search for the first active pixel on the edge.
struct Edge { CellIt from, to; };
Edge _edge(const Coord &ringvertex) const
struct Edge
{
size_t idx = ringvertex.r;
Coord cell = coord(idx);
uint8_t tg = m_tags[ringvertex.r];
SquareTag t = SquareTag(tg & 0x0f);
switch (t) {
case SquareTag::a:
case SquareTag::ab:
case SquareTag::abc:
return {{tl(cell), Dir::down, m_rst}, {bl(cell)}};
case SquareTag::b:
case SquareTag::bc:
case SquareTag::bcd:
return {{bl(cell), Dir::right, m_rst}, {br(cell)}};
case SquareTag::c:
return {{br(cell), Dir::up, m_rst}, {tr(cell)}};
case SquareTag::ac:
switch (Dir(ringvertex.c)) {
case Dir::left: return {{tl(cell), Dir::down, m_rst}, {bl(cell)}};
case Dir::right: return {{br(cell), Dir::up, m_rst}, {tr(cell)}};
default: assert(false);
}
case SquareTag::d:
case SquareTag::ad:
case SquareTag::abd:
return {{tr(cell), Dir::left, m_rst}, {tl(cell)}};
case SquareTag::bd:
switch (Dir(ringvertex.c)) {
case Dir::down: return {{bl(cell), Dir::right, m_rst}, {br(cell)}};
case Dir::up: return {{tr(cell), Dir::left, m_rst}, {tl(cell)}};
default: assert(false);
}
case SquareTag::cd:
case SquareTag::acd:
return {{br(cell), Dir::up, m_rst}, {tr(cell)}};
case SquareTag::full:
case SquareTag::none: {
Coord crd{tl(cell) + Coord{m_cellsize.r / 2, m_cellsize.c / 2}};
return {{crd, Dir::none, m_rst}, {crd}};
CellIt from, to;
};
Edge _edge(const Coord& ringvertex) const
{
size_t idx = ringvertex.r;
Coord cell = coord(idx);
Dir d = Dir(ringvertex.c);
switch (d) {
case Dir::left: return {{tl(cell), Dir::down, m_rst}, {bl(cell)}};
case Dir::down: return {{bl(cell), Dir::right, m_rst}, {br(cell)}};
case Dir::right: return {{br(cell), Dir::up, m_rst}, {tr(cell)}};
case Dir::up: return {{tr(cell), Dir::left, m_rst}, {tl(cell)}};
default: assert(false);
}
}
return {};
return {};
}
Edge edge(const Coord &ringvertex) const
Edge edge(const Coord& ringvertex) const
{
const long R = rows(*m_rst), C = cols(*m_rst);
const long R_1 = R - 1, C_1 = C - 1;
Edge e = _edge(ringvertex);
Edge e = _edge(ringvertex);
e.to.dir = e.from.dir;
++e.to;
e.from.crd.r = std::min(e.from.crd.r, R_1);
e.from.crd.r = std::max(e.from.crd.r, 0l);
e.from.crd.c = std::min(e.from.crd.c, C_1);
e.from.crd.c = std::max(e.from.crd.c, 0l);
e.to.crd.r = std::min(e.to.crd.r, R);
e.to.crd.r = std::max(e.to.crd.r, 0l);
e.to.crd.c = std::min(e.to.crd.c, C);
e.to.crd.c = std::max(e.to.crd.c, 0l);
e.from.crd.r = std::clamp(e.from.crd.r, 0l, m_rastsize.r - 1);
e.from.crd.c = std::clamp(e.from.crd.c, 0l, m_rastsize.c - 1);
e.to.crd.r = std::clamp(e.to.crd.r, 0l, m_rastsize.r);
e.to.crd.c = std::clamp(e.to.crd.c, 0l, m_rastsize.c);
return e;
}
public:
explicit Grid(const Rst &rst, const Coord &cellsz, const Coord &overlap)
: m_rst{&rst}
, m_cellsize{cellsz}
, m_res_1{m_cellsize.r - 1, m_cellsize.c - 1}
, m_window{overlap.r < cellsz.r ? cellsz.r - overlap.r : cellsz.r,
overlap.c < cellsz.c ? cellsz.c - overlap.c : cellsz.c}
, m_gridsize{2 + (long(rows(rst)) - overlap.r) / m_window.r,
2 + (long(cols(rst)) - overlap.c) / m_window.c}
, m_tags(m_gridsize.r * m_gridsize.c, 0)
{}
// Go through the cells and mark them with the appropriate tag.
template<class ExecutionPolicy>
void tag_grid(ExecutionPolicy &&policy, TRasterValue<Rst> isoval)
{
// parallel for r
for_each (std::forward<ExecutionPolicy>(policy),
m_tags.begin(), m_tags.end(),
[this, isoval](uint8_t& tag, size_t idx) {
tag = get_tag_for_cell(coord(idx), isoval);
});
void interpolate_edge(Coord& ecrd, TRasterValue<Rst> isoval) const
{
// The ecrd must have a grid index in r and a direction in c.
assert((0 <= ecrd.r) && (ecrd.r < m_gridlen));
assert((ecrd.c == long(Dir::left)) || (ecrd.c == long(Dir::down)) || (ecrd.c == long(Dir::right)) || (ecrd.c == long(Dir::up)));
Edge e = edge(ecrd);
ecrd = std::lower_bound(e.from, e.to, isoval).crd;
// Shift bottom and right side points "out" by one to account for
// raster pixel width. Note "dir" is the direction of interpolation
// along the cell edge, not the next move direction.
if (e.from.dir == Dir::up)
ecrd.r += 1;
else if (e.from.dir == Dir::left)
ecrd.c += 1;
}
// Scan for the rings on the tagged grid. Each ring vertex stores the
// sequential index of the cell and the next direction (Dir).
// This info can be used later to calculate the exact raster coordinate.
public:
explicit Grid(const Rst& rst, const Coord& window)
: m_rst{&rst}
, m_window{window}
, m_rastsize{static_cast<long>(rows(rst)), static_cast<long>(cols(rst))}
, m_gridsize{2 + m_rastsize.r / m_window.r, 2 + m_rastsize.c / m_window.c}
, m_gridlen{static_cast<size_t>(m_gridsize.r * m_gridsize.c)}
, m_tags(m_gridlen / 32 + 1, 0) // 1 bit per tag means 32 per uint32_t.
, m_dirs(m_gridlen / 8 + 1, 0) // 4 bits per cell means 32/4=8 per uint32_t.
{}
// Go through the cells getting their tags and dirs.
template<class ExecutionPolicy> void tag_grid(ExecutionPolicy&& policy, TRasterValue<Rst> isoval)
{
// Get all the tags. parallel?
for_each_idx(std::forward<ExecutionPolicy>(policy), m_tags.begin(), m_tags.end(),
[this, isoval](uint32_t& tag_block, size_t bidx) { tag_block = get_tags_block32(bidx, isoval); });
// streamtags(std::cerr);
// Get all the dirs. parallel?
for_each_idx(std::forward<ExecutionPolicy>(policy), m_dirs.begin(), m_dirs.end(),
[this](uint32_t& dirs_block, size_t bidx) { dirs_block = get_dirs_block8(bidx); });
// streamdirs(std::cerr);
}
// Scan for the rings on the tagged grid. Each ring vertex uses the Coord
// to store the sequential cell index (idx in r) and next direction (Dir
// in c) for the next point of the ring. This info can be used later to
// calculate the exact raster coordinate of the point.
std::vector<Ring> scan_rings()
{
std::vector<Ring> rings;
size_t startidx = 0;
while ((startidx = search_start_cell(startidx)) < m_tags.size()) {
size_t startidx = 0;
while ((startidx = search_start_cell(startidx)) < m_gridlen) {
Ring ring;
size_t idx = startidx;
Dir prev = Dir::none, next = next_dir(prev, get_tag(idx));
while (next != Dir::none && !is_visited(idx, prev)) {
size_t idx = startidx;
Dir next = next_dir(idx);
do {
// We should never touch a cell with no remaining directions
// until we get back to the start cell.
assert(next != Dir::none);
Coord ringvertex{long(idx), long(next)};
ring.emplace_back(ringvertex);
set_visited(idx, prev);
idx = seq(step(coord(idx), next));
prev = next;
next = next_dir(next, get_tag(idx));
}
// To prevent infinite loops in case of degenerate input
if (next == Dir::none) m_tags[startidx] = _t(SquareTag::none);
if (ring.size() > 1) {
ring.pop_back();
clr_dirs(idx, next);
idx = stepidx(idx, next);
next = next_dir(idx, next);
assert(idx < m_gridlen);
} while (idx != startidx);
// The start cell on returning should either have its directions
// cleared or have the second ambiguous direction.
assert(next == Dir::none || next == Dir::left || next == Dir::up);
if (ring.size() > 1)
rings.emplace_back(ring);
}
}
return rings;
}
// Calculate the exact raster position from the cells which store the
// sequantial index of the square and the next direction
template<class ExecutionPolicy>
void interpolate_rings(ExecutionPolicy && policy,
std::vector<Ring> &rings,
TRasterValue<Rst> isov)
// sequential index of the square and the next direction
template<class ExecutionPolicy> void interpolate_rings(ExecutionPolicy&& policy, std::vector<Ring>& rings, TRasterValue<Rst> isov)
{
for_each(std::forward<ExecutionPolicy>(policy),
rings.begin(), rings.end(), [this, isov] (Ring &ring, size_t)
{
for (Coord &ringvertex : ring) {
Edge e = edge(ringvertex);
CellIt found = std::lower_bound(e.from, e.to, isov);
ringvertex = found.crd;
}
for_each_idx(std::forward<ExecutionPolicy>(policy), rings.begin(), rings.end(), [this, isov](Ring& ring, size_t) {
for (Coord& e : ring)
interpolate_edge(e, isov);
});
}
std::ostream& streamtags(std::ostream& os)
{
os << " :";
for (auto c = 0; c < m_gridsize.c; c++)
os << (c % 10);
os << "\n";
for (auto r = 0; r < m_gridsize.r; r++) {
os << std::setw(3) << r << ":";
for (auto c = 0; c < m_gridsize.c; c++)
os << ((get_tags9(seq(Coord(r, c))) & 1) ? "H" : ".");
os << "\n";
}
return os;
}
std::ostream& streamdirs(std::ostream& os)
{
os << " :";
for (auto c = 0; c < m_gridsize.c; c++)
os << (c % 10);
os << "\n";
for (auto r = 0; r < m_gridsize.r; r++) {
os << std::setw(3) << r << ":";
for (auto c = 0; c < m_gridsize.c; c++)
os << get_dirs(seq(Coord(r, c)));
os << std::endl;
}
return os;
}
};
template<class Raster, class ExecutionPolicy>
std::vector<marchsq::Ring> execute_with_policy(ExecutionPolicy && policy,
const Raster & raster,
std::vector<marchsq::Ring> execute_with_policy(ExecutionPolicy&& policy,
const Raster& raster,
TRasterValue<Raster> isoval,
Coord windowsize = {})
Coord windowsize = {})
{
if (!rows(raster) || !cols(raster)) return {};
if (!rows(raster) || !cols(raster))
return {};
size_t ratio = cols(raster) / rows(raster);
if (!windowsize.r) windowsize.r = 2;
if (!windowsize.r)
windowsize.r = 2;
if (!windowsize.c)
windowsize.c = std::max(2l, long(windowsize.r * ratio));
Coord overlap{1};
Grid<Raster> grid{raster, windowsize, overlap};
Grid<Raster> grid{raster, windowsize};
grid.tag_grid(std::forward<ExecutionPolicy>(policy), isoval);
std::vector<marchsq::Ring> rings = grid.scan_rings();
grid.interpolate_rings(std::forward<ExecutionPolicy>(policy), rings, isoval);
return rings;
}
template<class Raster>
std::vector<marchsq::Ring> execute(const Raster &raster,
TRasterValue<Raster> isoval,
Coord windowsize = {})
template<class Raster> std::vector<marchsq::Ring> execute(const Raster& raster, TRasterValue<Raster> isoval, Coord windowsize = {})
{
return execute_with_policy(nullptr, raster, isoval, windowsize);
}

44
src/libslic3r/SLA/RasterToPolygons.cpp
View File

@@ -10,13 +10,13 @@ namespace marchsq {
// Specialize this struct to register a raster type for the Marching squares alg
template<> struct _RasterTraits<Slic3r::sla::RasterGrayscaleAA> {
using Rst = Slic3r::sla::RasterGrayscaleAA;
// The type of pixel cell in the raster
using ValueType = uint8_t;
// Value at a given position
static uint8_t get(const Rst &rst, size_t row, size_t col) { return rst.read_pixel(col, row); }
// Number of rows and cols of the raster
static size_t rows(const Rst &rst) { return rst.resolution().height_px; }
static size_t cols(const Rst &rst) { return rst.resolution().width_px; }
@@ -34,57 +34,55 @@ template<class Fn> void foreach_vertex(ExPolygon &poly, Fn &&fn)
}
ExPolygons raster_to_polygons(const RasterGrayscaleAA &rst, Vec2i32 windowsize)
{
{
size_t rows = rst.resolution().height_px, cols = rst.resolution().width_px;
if (rows < 2 || cols < 2) return {};
Polygons polys;
long w_rows = std::max(2l, long(windowsize.y()));
long w_cols = std::max(2l, long(windowsize.x()));
long w_rows = std::max(1l, long(windowsize.y()));
long w_cols = std::max(1l, long(windowsize.x()));
std::vector<marchsq::Ring> rings =
marchsq::execute(rst, 128, {w_rows, w_cols});
polys.reserve(rings.size());
auto pxd = rst.pixel_dimensions();
pxd.w_mm = (rst.resolution().width_px * pxd.w_mm) / (rst.resolution().width_px - 1);
pxd.h_mm = (rst.resolution().height_px * pxd.h_mm) / (rst.resolution().height_px - 1);
for (const marchsq::Ring &ring : rings) {
Polygon poly; Points &pts = poly.points;
pts.reserve(ring.size());
for (const marchsq::Coord &crd : ring)
pts.emplace_back(scaled(crd.c * pxd.w_mm), scaled(crd.r * pxd.h_mm));
polys.emplace_back(poly);
}
// reverse the raster transformations
ExPolygons unioned = union_ex(polys);
coord_t width = scaled(cols * pxd.h_mm), height = scaled(rows * pxd.w_mm);
auto tr = rst.trafo();
for (ExPolygon &expoly : unioned) {
if (tr.mirror_y)
foreach_vertex(expoly, [height](Point &p) {p.y() = height - p.y(); });
if (tr.mirror_x)
foreach_vertex(expoly, [width](Point &p) {p.x() = width - p.x(); });
expoly.translate(-tr.center_x, -tr.center_y);
if (tr.flipXY)
foreach_vertex(expoly, [](Point &p) { std::swap(p.x(), p.y()); });
if ((tr.mirror_x + tr.mirror_y + tr.flipXY) % 2) {
expoly.contour.reverse();
for (auto &h : expoly.holes) h.reverse();
}
}
return unioned;
}

2
src/libslic3r/SLA/RasterToPolygons.hpp
View File

@@ -8,7 +8,7 @@ namespace sla {
class RasterGrayscaleAA;
ExPolygons raster_to_polygons(const RasterGrayscaleAA &rst, Vec2i32 windowsize = {2, 2});
ExPolygons raster_to_polygons(const RasterGrayscaleAA &rst, Vec2i32 windowsize = {1, 1});
}} // namespace Slic3r::sla

60
src/libslic3r/StreamUtils.hpp Normal file
View File

@@ -0,0 +1,60 @@
#ifndef slic3r_StreamUtils_hpp_
#define slic3r_StreamUtils_hpp_
#include "Point.hpp"
#include "libslic3r.h"
#include "Polygon.hpp"
#include "Polyline.hpp"
#include "ExPolygon.hpp"
#include <sstream>
#include <vector>
namespace Slic3r {
inline std::ostream& operator<<(std::ostream& os, const Points& pts)
{
os << "[" << pts.size() << "]:";
for (Point p : pts)
os << " (" << p << ")";
os << "\n";
return os;
}
inline std::ostream& operator<<(std::ostream& os, const MultiPoint& mpts)
{
os << "Multipoint" << mpts.points;
return os;
}
inline std::ostream& operator<<(std::ostream& os, const Polygon& poly)
{
os << "Polygon" << poly.points;
return os;
}
inline std::ostream& operator<<(std::ostream& os, const Polygons& polys)
{
os << "Polygons[" << polys.size() << "]:" << "\n";
for (Polygon p : polys)
os << " " << p;
return os;
}
inline std::ostream& operator<<(std::ostream& os, const ExPolygon& epoly)
{
os << "ExPolygon:\n";
os << " contour: " << epoly.contour;
os << " holes: " << epoly.holes;
return os;
}
inline std::ostream& operator<<(std::ostream& os, const ExPolygons& epolys)
{
os << "ExPolygons[" << epolys.size() << "]:" << "\n";
for (ExPolygon p : epolys)
os << " " << p;
return os;
}
} // namespace Slic3r
#endif // slic3r_StreamUtils_hpp_