Godrak
43d9166382
nomralize weights of points before curve fitting and fitted value interpolation
2022-04-25 12:42:51 +02:00
PavelMikus
68cf4db58e
interpolate fitted and original position during b spline alignment -
...
push points with large weight more towards their original position
2022-04-25 12:42:51 +02:00
PavelMikus
137fa35238
Postpone seam picking for spNearest configuration to the place_seam method.
2022-04-25 12:42:51 +02:00
PavelMikus
83f3ca27dc
reworked and improved the find_next_seam_in_layer method:
...
Now uses find nearby_points with radius to save some computations
if all points are far.
From the nearby points, it finds the nearest and best point, and tries
to use them in this order (preivously only nearest was considered).
This helps to snap the alignment to nearby sharp corners if present.
2022-04-25 12:42:51 +02:00
PavelMikus
04d4a0d4f7
when searching for central enforcer (for alignment purposes), find properly the first
...
enforced segment. Fixed issue where if the enforced segment was painted over the start/end of the
perimeter, part of the enforced points was not considered.
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
2dfabb7e69
Fixing missing include.
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
399b7f79e8
Little more refactoring of SeamPlacer.
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
8c2e6aba79
On MSVC, std::deque degenerates to a list of pointers, which defeats
...
its purpose of reducing allocator load and memory fragmentation.
https://github.com/microsoft/STL/issues/147#issuecomment-1090148740
Slic3r::deque<> compiles to boost::container::deque<> on Windows,
to std::deque<> on other systems.
SeamPlacer newly uses Slic3r::deque<>.
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
adb467286f
Documented the seam placement corner penalty with an image.
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
853b8adf80
Little refactoring of SeamPlacer.
...
Moved color mapping functions to Color.hpp
Removed the "extern" keyword from Color.hpp
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
f5709345ad
GCode export: Replaced std::strings in G-code comments with string_views.
...
Seam placer: Detecting perimeter by the pointer and size of the
comment_perimeter string_view, only placing seams for perimeters.
2022-04-25 12:42:51 +02:00
PavelMikus
fb2621c03c
Negative volumes raycasting fix - normal was flipped between iterations, incorrect algorithm for hit detection
...
Debug files export fix after refactoring
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
8ce36e9137
Refactoring of SeamPlacer:
...
Replaced shared_ptr<> with deque.
Merged multiple vectors into one.
Refactoring using common helper functions (prev/next_idx_modulo(),
angle(), ...)
AABBTreeIndirect::intersect_ray_all_hits(): Reuse memory of the hits cache.
2022-04-25 12:42:51 +02:00
PavelMikus
3b8cfc62da
fixed drawing seams on multipart objects
...
removed oversampling for blockers
improved drawing seams over sharp features
2022-04-25 12:42:51 +02:00
PavelMikus
47fc39a4ba
fixed misleanding or wrong comments of the fitting function
2022-04-25 12:42:51 +02:00
PavelMikus
ae89d65e3e
added description for the parameter count increase
2022-04-25 12:42:51 +02:00
PavelMikus
396d3215bd
Refactoring of curve fitting algorithm:
...
removal of artificial extension at the ends of the curve
removal of observation points normalization
added clamping of parameter index which compensates for under-represented spline segments
added parameter for level of freedom at the ends of the curve
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
c19770189f
Follow-up to 1c9ba291fe32bc4a4c78cabbab0639b0c164f23f
...
Refactoring of Curves.hpp for better memory management and vectorization
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
42e802c1b8
Refactoring of Curves.hpp for better memory management and vectorization
...
(replaced vector of vectors with Eigen 2D matrices).
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
bd8ce6fabd
Follow-up to 9cb51caead9dee5c4d12a4ef5f925b5efeb00b75
...
Fixing compilation on GCC
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
e95d19b560
WIP, FIXME: Worked around a Perl integration test failure
...
in t/perimeters.t, see the comment
FIXME skip the 1st layer in the test 'loops start on concave point if any'
The failure is due to the seam smoothing, which brings the seam too far
away from the external perimeter at the 1st layer, which uses significantly
wider extrusion rate than the other layers. Moving the seam point from
the external perimeter causes an error when projecting the seam point
back to the external perimeter, where the projection lands not exactly
at a corner point, creating a short segment at the end of a polygon loop.
2022-04-25 12:42:51 +02:00
Vojtech Bubnik
7d02647ebf
Removed various Point::ccw() and Point::ccw_angle() methods, they were
...
provided for Perl bindings and their semantic was confusing.
Implemented free function angle() to measure angle between two vectors.
Reworked Polygon::convex/concave_points(), changed the meaning of their
angle threshold parameter.
Removed some unused methods from Perl bindings and tests.
Reworked the "wipe inside at the external perimeter" function
after Point::ccw_angle() was removed.
2022-04-25 12:42:51 +02:00
PavelMikus
156a60017d
fixed ExPolygons dealocation while using EdgeGrid
...
fixed warnings in Bicubic.h file
2022-04-25 12:42:51 +02:00
PavelMikus
965803822e
remove invalid comment
2022-04-25 12:42:51 +02:00
PavelMikus
1164449d4e
compute overhang distance using SDF
...
detect embedded (inner) perimeter points and prefer them for seam placement
2022-04-25 12:42:51 +02:00
PavelMikus
191e788aa0
make Random seams disaligned
2022-04-25 12:42:51 +02:00
PavelMikus
15135ef2ed
fixes, central enforced point preference
2022-04-25 12:42:51 +02:00
PavelMikus
5c23d471de
BSplines, Polynomial fitting
2022-04-25 12:42:51 +02:00
PavelMikus
bbcd6be250
Implemented piecewise data (curve) fitting with variable kernels
2022-04-25 12:42:51 +02:00
PavelMikus
bb89b630d9
implemented occlusion estimation for objects with negative volumes
2022-04-25 12:42:51 +02:00
PavelMikus
ca259caf33
fix ccw angle computation in Point.cpp
2022-04-25 12:42:51 +02:00
PavelMikus
177a1fd54a
finalize integration into GCode.cpp export functions, remove unused
...
edge grids
2022-04-25 12:42:51 +02:00
PavelMikus
6dbc7149be
parameter fixes, alignemnt for enforcers simplified
2022-04-25 12:42:51 +02:00
PavelMikus
c640fb854f
bug fix: using trafo() instead of trafo_centred() caused
...
misalignment between occlusion mesh and seam candidates
2022-04-25 12:42:51 +02:00
PavelMikus
2274965079
alignment from best candidate
2022-04-25 12:42:51 +02:00
PavelMikus
eccf1c1553
refactored raycasting - inverted direction of raycasting - now each face is tested
...
fixed bug with custom seam drawings - square distance parameter named incorrectly
2022-04-25 12:42:51 +02:00
PavelMikus
a9f5330ad2
using gauss function to smoothen criteria skips
2022-04-25 12:42:51 +02:00
Godrak
ab3c8d0fe8
implemented smooth angle penalty function
2022-04-25 12:42:51 +02:00
PavelMikus
a92d5038bd
debug export svg info,
...
fixing problem with weird seam placement caused by disconnected scoring function
2022-04-25 12:42:51 +02:00
PavelMikus
962282c9ef
split occlusion and enforcers/blockers into separate functions
...
added weights to polynomial fitting
2022-04-25 12:42:51 +02:00
PavelMikus
f837759928
fixed problem with multipart objects
...
fixed bug : model volume trafo was not considered
2022-04-25 12:42:51 +02:00
PavelMikus
a775bf2978
improved enforcers blockers oversampling, renamed parameter
2022-04-25 12:42:51 +02:00
PavelMikus
8f7b86915a
oversample polygons where necessary, due to seam enforcers & blockers
2022-04-25 12:42:51 +02:00
PavelMikus
87c276b7a4
comments and bugfix
2022-04-25 12:42:51 +02:00
PavelMikus
ad819850f9
tweaked parameters,
...
fixed minor bugs
2022-04-25 12:42:51 +02:00
PavelMikus
c72687c96c
computing smooth score instead of binary decision when picking seams
...
some basic documentation
2022-04-25 12:42:51 +02:00
PavelMikus
105b67c9a7
presorting seams before alignemnt
...
mesh decimation for speed up
2022-04-25 12:42:51 +02:00
PavelMikus
f018160e72
implemented polynomial alignment,
...
however, initital seam placement is not ideal
- hard to balance visual cues and angle information
2022-04-25 12:42:51 +02:00
PavelMikus
ffc7452d9e
improved visibility calculation - it now considers normals and
...
accordingly counts only hits which have similar normal
2022-04-25 12:42:51 +02:00
PavelMikus
38a9d870c0
implemented seam alignment using exponential smoothing
2022-04-25 12:42:51 +02:00
