blender
/
blender
134
399
Fork 581
Code Issues 6.3k Pull Requests 757 Projects 19 Wiki Activity

WIP: The SLIM UV unwrap algorithm implementation for Blender 4.x #114545

Draft
Lukasz Czyz wants to merge 168 commits from glukoz/blender:uv_unwrapping_slim_algorithm_v5_update into main
pull from: glukoz/blender:uv_unwrapping_slim_algorithm_v5_update
merge into: blender:main

When changing the target branch, be careful to rebase the branch in your fork to match. See documentation.
Conversation 32 Commits 168 Files Changed 44 +4916 -155
Lukasz Czyz commented 2023-11-06 17:50:22 +01:00
First-time contributor
Copy Link

SUMMARY

The patch integrates an existing implementation of the SLIM unwrapping algorithm into Blender. More info about SLIM here: https://igl.ethz.ch/projects/slim/. The patch is based on the integration code written by Aurel Gruber for Blender 2.7x (not finished and never actually merged with the master branch). To learn more about SLIM, watch the following presentation by Aurel: https://www.youtube.com/watch?v=PLUoLQUK3-s

In this PR the original Aurel's code was rebased to a recent main commit and a bunch of further improvements were introduced.

HANDLING DOUBLED VERTICES AND ZERO TRIANGLES

First of all I used the original p_chart_simplify disabled code to implement  a new function: p_chart_collapse_doubles. The collapsing code used by p_chart_collapse_doubles is almost identical to p_chart_simplify - the criteria of collapsing is different though: as the name suggests it collapses edges only to remove doubled vertices.

Thanks to the simplified collapse criteria, it seems that we can avoid explicit uncollapsing after the operation is done - instead we can easily flush UV coords to collapsed vertices just before applying the result. It is the approach the current code uses (check the p_chart_flush_collapsed_uvs function).

To sum up (together with changes from the previous update): currently the mesh preprocessing before SLIM unwrap consists of two phases:

p_chart_collapse_doubles - removing doubled vertices
p_chart_correct_zero_angles - correcting zero area triangles.

WARNING - at the current stage it is not guaranteed that the above functions will work properly together with the live unwrap functionality. They may require some improvements. The point is that I cannot risk spending much time on improving and polishing these solutions without a confirmation that such an approach is acceptable by Blender devs on the general level.

So at this moment I need a confirmation (or rejection) of this general approach in order to continue work on this task.

### SUMMARY The patch integrates an existing implementation of the SLIM unwrapping algorithm into Blender. More info about SLIM here: https://igl.ethz.ch/projects/slim/. The patch is based on the integration code written by Aurel Gruber for Blender 2.7x (not finished and never actually merged with the master branch). To learn more about SLIM, watch the following presentation by Aurel: https://www.youtube.com/watch?v=PLUoLQUK3-s In this PR the original Aurel's code was rebased to a recent main commit and a bunch of further improvements were introduced. ### HANDLING DOUBLED VERTICES AND ZERO TRIANGLES First of all I used the original `p_chart_simplify` disabled code to implement  a new function: `p_chart_collapse_doubles`. The collapsing code used by `p_chart_collapse_doubles` is almost identical to `p_chart_simplify` - the criteria of collapsing is different though: as the name suggests it collapses edges only to remove doubled vertices. Thanks to the simplified collapse criteria, it seems that we can avoid explicit uncollapsing after the operation is done - instead we can easily flush UV coords to collapsed vertices just before applying the result. It is the approach the current code uses (check the `p_chart_flush_collapsed_uvs` function). To sum up (together with changes from the previous update): currently the mesh preprocessing before SLIM unwrap consists of two phases: `p_chart_collapse_doubles` - removing doubled vertices `p_chart_correct_zero_angles` - correcting zero area triangles. WARNING - at the current stage it is not guaranteed that the above functions will work properly together with the live unwrap functionality. They may require some improvements. The point is that I cannot risk spending much time on improving and polishing these solutions without a confirmation that such an approach is acceptable by Blender devs on the general level. **So at this moment I need a confirmation (or rejection) of this general approach in order to continue work on this task.**
👍 2 ❤️ 5 🎉 3 🚀 5
Lukasz Czyz added 104 commits 2023-11-06 17:50:39 +01:00
bee988b99c Merge branch 'master' into uv_unwrapping_slim_algorithm_v3.2 
# Conflicts:
#	source/blender/editors/uvedit/uvedit_parametrizer.h
#	source/blender/editors/uvedit/uvedit_unwrap_ops.c
#	source/blender/geometry/intern/uv_parametrizer.c
4a5cc96a20 Merge branch 'master' of git://git.blender.org/blender into uv_unwrapping_slim_algorithm_v4 
# Conflicts:
#	intern/CMakeLists.txt
#	source/blender/blenloader/intern/versioning_300.c
#	source/blender/editors/uvedit/uvedit_unwrap_ops.c
#	source/blender/geometry/intern/uv_parametrizer.cc
c3d3918bef Merge branch 'main' of https://projects.blender.org/blender/blender into uv_unwrapping_slim_algorithm_v5 
# Conflicts:
#	release/datafiles/locale
#	release/scripts/addons
#	source/blender/blenloader/intern/versioning_260.cc
#	source/blender/blenloader/intern/versioning_300.cc
#	source/blender/editors/uvedit/CMakeLists.txt
#	source/blender/editors/uvedit/uvedit_unwrap_ops.cc
#	source/blender/geometry/CMakeLists.txt
#	source/blender/geometry/GEO_uv_parametrizer.h
#	source/blender/geometry/intern/uv_parametrizer.cc
#	source/blender/makesdna/DNA_scene_types.h
#	source/blender/nodes/geometry/CMakeLists.txt
#	source/blender/nodes/geometry/nodes/node_geo_uv_pack_islands.cc
#	source/blender/nodes/geometry/nodes/node_geo_uv_unwrap.cc
#	source/blender/windowmanager/WM_api.h
#	source/tools
0c339e6e40 Merge branch 'uv_unwrapping_slim_algorithm_v4_liveunwrap' into uv_unwrapping_slim_algorithm_v5_liveunwrap 
# Conflicts:
#	source/blender/editors/include/ED_uvedit.h
#	source/blender/editors/transform/transform.cc
#	source/blender/editors/uvedit/uvedit_unwrap_ops.cc
Lukasz Czyz changed title from The SLIM UV unwrap algorithm implementation for Blender 4.x to WIP: The SLIM UV unwrap algorithm implementation for Blender 4.x 2023-11-06 17:50:58 +01:00
Lukasz Czyz requested review from Brecht Van Lommel 2023-11-06 17:54:37 +01:00
Brecht Van Lommel commented 2023-11-08 19:44:25 +01:00
Owner
Copy Link

Thanks, I will find some time to look into this again. Hopefully next week.

Thanks, I will find some time to look into this again. Hopefully next week.
Brecht Van Lommel added 28 commits 2023-11-17 19:32:20 +01:00
Brecht Van Lommel commented 2023-11-17 19:35:49 +01:00
Owner
Copy Link

I'm still working my way through the code, but pushed a bunch of changes to cut the size of this PR in half. And to use proper SPDX headers and align closer to our code style.

I'm still working my way through the code, but pushed a bunch of changes to cut the size of this PR in half. And to use proper SPDX headers and align closer to our code style.
Brecht Van Lommel commented 2023-11-17 19:37:29 +01:00
Owner
Copy Link

I see various warnings when building this, should show on the buildbot too:
@blender-bot build

I see various warnings when building this, should show on the buildbot too: @blender-bot build
Lukasz Czyz commented 2023-11-23 14:09:47 +01:00
Author
First-time contributor
Copy Link

I pulled your changes and after that building fails. Is that expect or is it an issue with my environment?

I pulled your changes and after that building fails. Is that expect or is it an issue with my environment?
Brecht Van Lommel commented 2023-11-23 14:23:00 +01:00
Owner
Copy Link

It's not expected. Maybe try to fix the errors, and if you can't show the error log?

It's not expected. Maybe try to fix the errors, and if you can't show the error log?
Lukasz Czyz added 1 commit 2023-11-23 15:14:03 +01:00
Lukasz Czyz commented 2023-11-23 15:15:08 +01:00
Author
First-time contributor
Copy Link

Ok, I fixed compilation and all warnings related to this PR.

Ok, I fixed compilation and all warnings related to this PR.
Brecht Van Lommel added 2 commits 2023-11-23 17:25:24 +01:00
buildbot/vexp-code-patch-coordinator Build done. Details
335ea29db0 Cleanup: follow code style
Brecht Van Lommel commented 2023-11-23 17:30:11 +01:00
Owner
Copy Link

There were more warnings in the buildbot logs. I fixed them, except these:

source/blender/geometry/intern/uv_parametrizer.cc:4964:10: warning: variable ‘select’ set but not used [-Wunused-but-set-variable]
source/blender/geometry/intern/uv_parametrizer.cc:4964:26: warning: variable ‘deselect’ set but not used [-Wunused-but-set-variable]

I guess certain charts need to be skipped based on these, similar to p_chart_lscm_begin.

@blender-bot build

There were more warnings in the buildbot logs. I fixed them, except these: ``` source/blender/geometry/intern/uv_parametrizer.cc:4964:10: warning: variable ‘select’ set but not used [-Wunused-but-set-variable] source/blender/geometry/intern/uv_parametrizer.cc:4964:26: warning: variable ‘deselect’ set but not used [-Wunused-but-set-variable] ``` I guess certain charts need to be skipped based on these, similar to `p_chart_lscm_begin`. @blender-bot build
Brecht Van Lommel added 3 commits 2023-11-23 20:45:08 +01:00
Brecht Van Lommel requested changes 2023-11-23 20:50:22 +01:00
source/blender/editors/uvedit/uvedit_unwrap_ops.cc Outdated
@ -222,0 +263,4 @@
}
else {
/* We use the properties from the last unwrap operator for subsequent
* live unwrap and minize stretch operators. */
Brecht Van Lommel commented 2023-11-23 19:16:20 +01:00
Owner
Copy Link

This comment seems to be incorrect, and perhaps this code and WM_operator_current_or_default_properties_alloc can be removed?

  • ED_uvedit_live_unwrap passes in toolsettings, so this code does not run
  • unwrap_exec does not pass in tool settings, but the operator should already remember its properties
  • pack_islands_exec and average_islands_scale_exec do not pass in tool settings, but these properties seems irrelevant to them.
This comment seems to be incorrect, and perhaps this code and `WM_operator_current_or_default_properties_alloc` can be removed? * `ED_uvedit_live_unwrap` passes in toolsettings, so this code does not run * `unwrap_exec` does not pass in tool settings, but the operator should already remember its properties * `pack_islands_exec` and `average_islands_scale_exec` do not pass in tool settings, but these properties seems irrelevant to them.
Lukasz Czyz commented 2023-12-01 17:00:56 +01:00
Author
First-time contributor
Copy Link

The main rationale of this code was to make the minimize stretch operator use the latest options used by the unwrap operator.

The rationale is not reflected in the current state of the implementation because the minimize stretch operator hasn't been ported to use SLIM yet. But it will probably be ported sooner or later so this code may be useful in the future.

The main rationale of this code was to make the minimize stretch operator use the latest options used by the unwrap operator. The rationale is not reflected in the current state of the implementation because the minimize stretch operator hasn't been ported to use SLIM yet. But it will probably be ported sooner or later so this code may be useful in the future.
Brecht Van Lommel commented 2023-12-01 17:38:28 +01:00
Owner
Copy Link

The last used unwrap settings are already stored in the scene tool settings, so minimize stretch should get them from there instead.

The last used unwrap settings are already stored in the scene tool settings, so minimize stretch should get them from there instead.
Lukasz Czyz commented 2023-12-01 19:33:17 +01:00
Author
First-time contributor
Copy Link

I see. So I removed that call.

I see. So I removed that call.
brecht marked this conversation as resolved
source/blender/editors/uvedit/uvedit_unwrap_ops.cc Outdated
@ -2605,0 +2826,4 @@
else {
uiDefAutoButsRNA(
layout, &ptr, unwrap_draw_check_prop_abf, NULL, NULL, UI_BUT_LABEL_ALIGN_NONE, false);
}
Brecht Van Lommel commented 2023-11-23 20:42:47 +01:00
Owner
Copy Link

I think this would benefit from a manual layout now, using uiItemR for individual properties instead of uiDefAUtoButsRNA. That way properties can be grouped better.

For SLIM I think it could be grouped like this:

Method
Iterations
Reflection Mode

Vertex Group
Factor

Use Subdivision Surface

Correct Aspect
Margin Method
Margin
I think this would benefit from a manual layout now, using `uiItemR` for individual properties instead of `uiDefAUtoButsRNA`. That way properties can be grouped better. For SLIM I think it could be grouped like this: ``` Method Iterations Reflection Mode Vertex Group Factor Use Subdivision Surface Correct Aspect Margin Method Margin ```
Lukasz Czyz commented 2023-12-01 19:30:54 +01:00
Author
First-time contributor
Copy Link

Done, though I wasn't able to add separators between prop groups you specified. I was looking for a C++ counterpart of the Python layout.separator() call but with no luck.

Done, though I wasn't able to add separators between prop groups you specified. I was looking for a C++ counterpart of the Python `layout.separator()` call but with no luck.
Brecht Van Lommel commented 2023-12-05 15:05:28 +01:00
Owner
Copy Link

It's uiItemS().

It's `uiItemS()`.
source/blender/editors/uvedit/uvedit_unwrap_ops.cc Outdated
@ -2610,3 +2838,4 @@
{0, nullptr, 0, nullptr, nullptr},
};
static const EnumPropertyItem reflection_items[] = {{0, "ALLOW", 0, "Allow Flips", ""},
Brecht Van Lommel commented 2023-11-23 20:43:32 +01:00
Owner
Copy Link

Could this be made into a allow_flips boolean? Doesn't seem like it should be an enum.

Could this be made into a `allow_flips` boolean? Doesn't seem like it should be an enum.
Lukasz Czyz commented 2023-12-01 19:28:07 +01:00
Author
First-time contributor
Copy Link

Done.

Done.
brecht marked this conversation as resolved
source/blender/geometry/intern/uv_parametrizer.cc Outdated
@ -4286,0 +4961,4 @@
slim::MatrixTransferChart &mt_chart = mt->mt_charts[i];
bool select = false, deselect = false;
int npins = 0;
Brecht Van Lommel commented 2023-11-23 19:23:56 +01:00
Owner
Copy Link

select, deselect and npins are not used. Does this need similar logic to p_chart_lscm_begin, skipping some charts based on these?

Right now when you use live unwrap, and select a pin on one island and move it, it will continue solving other islands. I think this is unexpected, and would cause poor performance when there are many islands but you only want to work on one.

`select`, `deselect` and `npins` are not used. Does this need similar logic to `p_chart_lscm_begin`, skipping some charts based on these? Right now when you use live unwrap, and select a pin on one island and move it, it will continue solving other islands. I think this is unexpected, and would cause poor performance when there are many islands but you only want to work on one.
Lukasz Czyz commented 2023-12-01 19:28:27 +01:00
Author
First-time contributor
Copy Link

Done.

Done.
brecht marked this conversation as resolved
Brecht Van Lommel commented 2023-11-23 20:58:04 +01:00
Owner
Copy Link

HANDLING DOUBLED VERTICES AND ZERO TRIANGLES

First of all I used the original p_chart_simplify disabled code to implement  a new function: p_chart_collapse_doubles. The collapsing code used by p_chart_collapse_doubles is almost identical to p_chart_simplify - the criteria of collapsing is different though: as the name suggests it collapses edges only to remove doubled vertices.

Thanks to the simplified collapse criteria, it seems that we can avoid explicit uncollapsing after the operation is done - instead we can easily flush UV coords to collapsed vertices just before applying the result. It is the approach the current code uses (check the p_chart_flush_collapsed_uvs function).

Is this code in the pull request? I do not see functions with these names.

The approach sounds good to me.

> ### HANDLING DOUBLED VERTICES AND ZERO TRIANGLES > > First of all I used the original `p_chart_simplify` disabled code to implement  a new function: `p_chart_collapse_doubles`. The collapsing code used by `p_chart_collapse_doubles` is almost identical to `p_chart_simplify` - the criteria of collapsing is different though: as the name suggests it collapses edges only to remove doubled vertices. > > Thanks to the simplified collapse criteria, it seems that we can avoid explicit uncollapsing after the operation is done - instead we can easily flush UV coords to collapsed vertices just before applying the result. It is the approach the current code uses (check the `p_chart_flush_collapsed_uvs` function). Is this code in the pull request? I do not see functions with these names. The approach sounds good to me.
Brecht Van Lommel commented 2023-11-23 21:03:51 +01:00
Owner
Copy Link

I was going to open a devtalk topic for user feedback, but it seems the code for doubled vertices and zero triangles is missing? If so I can wait for that. Text was going to be:

There is an improved UV unwrapping method under review, and it would be great to get some user testing. The main advantage of this method is that it can better preserve area, instead of mainly caring about preserving angles like the existing Angle Based method. It can be tested by using Unwrap and selecting "SLIM" in the Adjust Last Operation panel.

Download builds.

Mainly helpful would be:

  • Test various different meshes, with varying complexity, poor topology, thin triangles, .. to find potential bugs.
  • Test if pinning and live unwrapping works well. Unlike Conformal/Angle Based, SLIM is iterative and does not converge to the solution immediately. How is the user interface for this?
  • Test unwrapping with a weight painted map. Did you find this feature useful? How is the user interface?
I was going to open a devtalk topic for user feedback, but it seems the code for doubled vertices and zero triangles is missing? If so I can wait for that. Text was going to be: > There is an improved [UV unwrapping method under review](https://projects.blender.org/blender/blender/pulls/114545), and it would be great to get some user testing. The main advantage of this method is that it can better preserve area, instead of mainly caring about preserving angles like the existing Angle Based method. It can be tested by using Unwrap and selecting "SLIM" in the Adjust Last Operation panel. > > [Download builds](https://builder.blender.org/download/patch/PR114545/). > > Mainly helpful would be: > * Test various different meshes, with varying complexity, poor topology, thin triangles, .. to find potential bugs. > * Test if pinning and live unwrapping works well. Unlike Conformal/Angle Based, SLIM is iterative and does not converge to the solution immediately. How is the user interface for this? > * Test unwrapping with a weight painted map. Did you find this feature useful? How is the user interface?
Lukasz Czyz added 6 commits 2023-11-24 12:12:34 +01:00
8bd362d095 Merge branch 'uv_unwrapping_slim_algorithm_v4_collapse_doubles' into uv_unwrapping_slim_algorithm_v5_collapse_doubles 
# Conflicts:
#	source/blender/editors/uvedit/uvedit_unwrap_ops.cc
#	source/blender/geometry/intern/uv_parametrizer.cc
buildbot/vexp-code-patch-coordinator Build done. Details
17f72b8587 Make format.
Lukasz Czyz commented 2023-11-24 12:15:42 +01:00
Author
First-time contributor
Copy Link

Indeed, I forgot to rebase the collapse_doubles part of the original change. Now it's fixed. I think you can open the devtalk topic now :)

I will look into other change requests in the next week.

Indeed, I forgot to rebase the collapse_doubles part of the original change. Now it's fixed. I think you can open the devtalk topic now :) I will look into other change requests in the next week.
Brecht Van Lommel commented 2023-11-24 17:08:59 +01:00
Owner
Copy Link

@blender-bot package

@blender-bot package
Blender Bot commented 2023-11-24 17:09:02 +01:00
Member
Copy Link

Package build started. Download here when ready.

Package build started. [Download here](https://builder.blender.org/download/patch/PR114545) when ready.
Brecht Van Lommel added 1 commit 2023-11-24 18:15:35 +01:00
31ebcf66da Cleanup: deduplicate collapsing code 
With callback functions for cost/allowed, nearly all the code can
be shared with simplification.
Brecht Van Lommel added 1 commit 2023-11-24 19:16:17 +01:00
Brecht Van Lommel commented 2023-11-24 19:20:17 +01:00
Owner
Copy Link

The double vertex collapsing code looks fine to me. After deduplicating it with the simplify implementation, it's also not much extra code.

The double vertex collapsing code looks fine to me. After deduplicating it with the simplify implementation, it's also not much extra code.
Brecht Van Lommel commented 2023-11-24 21:08:04 +01:00
Owner
Copy Link

User feedback topic:
https://devtalk.blender.org/t/slim-uv-unwrapping/32192

User feedback topic: https://devtalk.blender.org/t/slim-uv-unwrapping/32192
Lukasz Czyz added 6 commits 2023-12-01 19:26:25 +01:00
buildbot/vexp-code-patch-coordinator Build done. Details
35cbc5a796 Make format
Brecht Van Lommel commented 2023-12-05 15:37:39 +01:00
Owner
Copy Link

@glukoz are you following the feedback topic? Would be good if you can reply there too.

I think the most obvious improvements to make are to have two separate entries in the Unwrap menu, and to rename Vertex Group to indicate its purpose. For both the trickiest part is finding good names. We should also consider renaming SLIM to something that's easier to understand along with that, as we did for ABF and LSCM in the past.

There was one example of an unwrap failing, maybe possible to solve by tweaking the degenerate triangle handling?
https://devtalk.blender.org/t/slim-uv-unwrapping/32192/26

Report about pins not working well:
https://devtalk.blender.org/t/slim-uv-unwrapping/32192/27

@glukoz are you following the feedback topic? Would be good if you can reply there too. I think the most obvious improvements to make are to have two separate entries in the Unwrap menu, and to rename Vertex Group to indicate its purpose. For both the trickiest part is finding good names. We should also consider renaming SLIM to something that's easier to understand along with that, as we did for ABF and LSCM in the past. There was one example of an unwrap failing, maybe possible to solve by tweaking the degenerate triangle handling? https://devtalk.blender.org/t/slim-uv-unwrapping/32192/26 Report about pins not working well: https://devtalk.blender.org/t/slim-uv-unwrapping/32192/27
Brecht Van Lommel commented 2023-12-05 15:39:11 +01:00
Owner
Copy Link

@blender-bot package

@blender-bot package
Blender Bot commented 2023-12-05 15:39:58 +01:00
Member
Copy Link

Package build started. Download here when ready.

Package build started. [Download here](https://builder.blender.org/download/patch/PR114545) when ready.
Lukasz Czyz commented 2023-12-08 11:16:22 +01:00
Author
First-time contributor
Copy Link

@glukoz are you following the feedback topic? Would be good if you can reply there too.

Sure, I will take a look on it.

There was one example of an unwrap failing, maybe possible to solve by tweaking the degenerate triangle handling?
https://devtalk.blender.org/t/slim-uv-unwrapping/32192/26

Sure, I can tweak the algorithm so that the given mesh is processed BUT keep in mind that even after tweaking there may be another mesh which will not pass the test. I mean the algorithm does its best to fix a mesh but it cannot guarantee that it will work for every degenerated example. The question is whether you can accept such an approach in the long term.

> @glukoz are you following the feedback topic? Would be good if you can reply there too. Sure, I will take a look on it. > There was one example of an unwrap failing, maybe possible to solve by tweaking the degenerate triangle handling? > https://devtalk.blender.org/t/slim-uv-unwrapping/32192/26 Sure, I can tweak the algorithm so that the given mesh is processed BUT keep in mind that even after tweaking there may be another mesh which will not pass the test. I mean the algorithm does its best to fix a mesh but it cannot guarantee that it will work for every degenerated example. The question is whether you can accept such an approach in the long term.
Brecht Van Lommel commented 2023-12-08 19:46:23 +01:00
Owner
Copy Link

Sure, I can tweak the algorithm so that the given mesh is processed BUT keep in mind that even after tweaking there may be another mesh which will not pass the test. I mean the algorithm does its best to fix a mesh but it cannot guarantee that it will work for every degenerated example. The question is whether you can accept such an approach in the long term.

What I did for ABF and LSCM is to keep investigating and fixing user reports until failures become rare. There's no well defined cut-off point for that. But it's important to look at some user reported ones since they tend to catch things a developer can find in testing.

If it turns out there is a certain class of problems we don't know how to solve, and it's common, then we can decide if some other algorithm is needed to clean up the geometry or if we accept it. But I would expect simple adjustments could go a long way. For example the in uv_initializer.cpp, if that step fails perhaps the angles could be clamped to some minimum value. Something similar really helped for ABF. If some other step fails, maybe similar clamping is possible in the solver.

> Sure, I can tweak the algorithm so that the given mesh is processed BUT keep in mind that even after tweaking there may be another mesh which will not pass the test. I mean the algorithm does its best to fix a mesh but it cannot guarantee that it will work for every degenerated example. The question is whether you can accept such an approach in the long term. What I did for ABF and LSCM is to keep investigating and fixing user reports until failures become rare. There's no well defined cut-off point for that. But it's important to look at some user reported ones since they tend to catch things a developer can find in testing. If it turns out there is a certain class of problems we don't know how to solve, and it's common, then we can decide if some other algorithm is needed to clean up the geometry or if we accept it. But I would expect simple adjustments could go a long way. For example the in `uv_initializer.cpp`, if that step fails perhaps the angles could be clamped to some minimum value. Something similar really helped for ABF. If some other step fails, maybe similar clamping is possible in the solver.
Brecht Van Lommel commented 2023-12-08 19:47:59 +01:00
Owner
Copy Link

What do you think about naming like this?

In the operator properties:
SLIM -> Minimum Stretch
Vertex Group -> Importance Weights

In the unwrap menu add two separate operators:

  • Unwrap Angle Based
  • Unwrap Minimum Stretch

It ties in with the same algorithm being used for minimize stretch later.

What do you think about naming like this? In the operator properties: SLIM -> Minimum Stretch Vertex Group -> Importance Weights In the unwrap menu add two separate operators: * Unwrap Angle Based * Unwrap Minimum Stretch It ties in with the same algorithm being used for minimize stretch later.
Brecht Van Lommel commented 2023-12-08 21:14:23 +01:00
Owner
Copy Link

I ran ABF and SLIM on a random subset of the dataset here. See attachment for script and results.
http://staff.ustc.edu.cn/~fuxm/projects/ProgressivePara/dataset.html

There are a fair amount of models where SLIM fails. Some of them actually succeed when removing the minimum angle and double vertex threshold, though have not checked if the results are good in general.

I ran ABF and SLIM on a random subset of the dataset here. See attachment for script and results. http://staff.ustc.edu.cn/~fuxm/projects/ProgressivePara/dataset.html There are a fair amount of models where SLIM fails. Some of them actually succeed when removing the minimum angle and double vertex threshold, though have not checked if the results are good in general.
auto_unwrap_results.zip
49 MiB
auto_test_unwrap_obj.blend
2.3 MiB
Lukasz Czyz commented 2023-12-13 21:23:36 +01:00
Author
First-time contributor
Copy Link

What do you think about naming like this?

The naming seems ok for me.

I will take a look at the failing examples soon.

> What do you think about naming like this? The naming seems ok for me. I will take a look at the failing examples soon.
Lukasz Czyz added 13 commits 2024-01-02 15:23:22 +01:00
83abe102f4 Merge branch 'main' into uv_unwrapping_slim_algorithm_v5_update 
# Conflicts:
#	doc/license/SPDX-license-identifiers.txt
#	source/blender/makesdna/DNA_scene_types.h
buildbot/vexp-code-patch-lint Build done. Details
buildbot/vexp-code-patch-darwin-arm64 Build done. Details
buildbot/vexp-code-patch-darwin-x86_64 Build done. Details
buildbot/vexp-code-patch-linux-x86_64 Build done. Details
buildbot/vexp-code-patch-windows-amd64 Build done. Details
buildbot/vexp-code-patch-coordinator Build done. Details
e2422af2c4 Make format.
Lukasz Czyz commented 2024-01-02 15:59:30 +01:00
Author
First-time contributor
Copy Link

In this update I worked on improving the function for correcting degenerate triangles. The function name is now p_chart_correct_degenerate_triangles. I reworked the function so it now checks two triangle parameters: area and minimum angle - the function tries to correct both parameters so they are above specified thresholds (note that both parameters are independent from the mathematical point of view).

The p_chart_correct_degenerate_triangles function is now able to deal with double vertices so the question is whether we still want to use the p_chart_collapse_doubles function.

I ran a test on the enrite D2 mesh set and it turned out that the correction function in the current state lowers the failing mesh count from 28 to 18. Note that the entire D2 set consists of 6190 meshes.

I investigated some of the failing 18 meshes and it turns out that SLIM may fail even when the correction function is able to correct all triangles in the mesh (by min area and min angle). Note that I double check that all triangles are actually corrected before the function returns (using asserts). So it looks like the failing meshes may be caused by bugs in the SLIM implementation itself.

Now I will take a look at the issue with pins not being respected by SLIM.

In this update I worked on improving the function for correcting degenerate triangles. The function name is now `p_chart_correct_degenerate_triangles`. I reworked the function so it now checks two triangle parameters: area and minimum angle - the function tries to correct both parameters so they are above specified thresholds (note that both parameters are independent from the mathematical point of view). The `p_chart_correct_degenerate_triangles` function is now able to deal with double vertices so the question is whether we still want to use the `p_chart_collapse_doubles` function. I ran a test on the enrite D2 mesh set and it turned out that the correction function in the current state lowers the failing mesh count from 28 to 18. Note that the entire D2 set consists of 6190 meshes. I investigated some of the failing 18 meshes and it turns out that SLIM may fail even when the correction function is able to correct all triangles in the mesh (by min area and min angle). Note that I double check that all triangles are actually corrected before the function returns (using asserts). So it looks like the failing meshes may be caused by bugs in the SLIM implementation itself. Now I will take a look at the issue with pins not being respected by SLIM.
👍 1
Brecht Van Lommel commented 2024-01-02 19:10:43 +01:00
Owner
Copy Link

@blender-bot package

@blender-bot package
Blender Bot commented 2024-01-02 19:10:47 +01:00
Member
Copy Link

Package build started. Download here when ready.

Package build started. [Download here](https://builder.blender.org/download/patch/PR114545) when ready.
Lukasz Czyz added 3 commits 2024-02-05 17:06:16 +01:00
00f0a1896b Merge remote-tracking branch 'origin/main' into uv_unwrapping_slim_algorithm_v5_update 
# Conflicts:
#	source/blender/blenloader/intern/versioning_300.cc
#	source/blender/nodes/geometry/CMakeLists.txt
buildbot/vexp-code-patch-lint Build done. Details
buildbot/vexp-code-patch-darwin-x86_64 Build done. Details
buildbot/vexp-code-patch-darwin-arm64 Build done. Details
buildbot/vexp-code-patch-linux-x86_64 Build done. Details
buildbot/vexp-code-patch-windows-amd64 Build done. Details
buildbot/vexp-code-patch-coordinator Build done. Details
17dc993248 SLIM - compilation fix.
Lukasz Czyz commented 2024-02-05 17:09:00 +01:00
Author
First-time contributor
Copy Link

I solved the problem with pins not being respected.

Now the main question is whether you still want to merge this algorithm given the fact that SLIM may fail even when the correction function is able to correct all triangles in the mesh (by min area and min angle) - as I explained above.

I solved the problem with pins not being respected. Now the main question is whether you still want to merge this algorithm given the fact that SLIM may fail even when the correction function is able to correct all triangles in the mesh (by min area and min angle) - as I explained above.
❤️ 1 🎉 2
Lukasz Czyz commented 2024-02-07 11:01:32 +01:00
Author
First-time contributor
Copy Link

@brecht Could you start a build with the recent changes? I am not sure whether I can do it on my own.

@brecht Could you start a build with the recent changes? I am not sure whether I can do it on my own.
Brecht Van Lommel commented 2024-02-07 13:31:04 +01:00
Owner
Copy Link

@blender-bot package

@blender-bot package
Blender Bot commented 2024-02-07 13:31:07 +01:00
Member
Copy Link

Package build started. Download here when ready.

Package build started. [Download here](https://builder.blender.org/download/patch/PR114545) when ready.
Brecht Van Lommel commented 2024-02-07 13:32:50 +01:00
Owner
Copy Link

I'd like to have another look to see if I can find some solution for those failures. I don't think it's a blocker if we can't, but would be nice.

Then I guess this is nearly ready to merge for Blender 4.2? Still need to check all the latest changes.

I'd like to have another look to see if I can find some solution for those failures. I don't think it's a blocker if we can't, but would be nice. Then I guess this is nearly ready to merge for Blender 4.2? Still need to check all the latest changes.
🎉 5
Lukasz Czyz commented 2024-02-19 12:07:16 +01:00
Author
First-time contributor
Copy Link

@brecht Could ask people for feedback for the latest build in the dev thread? Unfortunately I cannot add 3 messages in a row there.

@brecht Could ask people for feedback for the latest build in the dev thread? Unfortunately I cannot add 3 messages in a row there.
Brecht Van Lommel commented 2024-02-19 12:56:47 +01:00
Owner
Copy Link

@glukoz I've changed your trust level on devtalk to remove that restriction.

@glukoz I've changed your trust level on devtalk to remove that restriction.
Some checks failed
buildbot/vexp-code-patch-lint Build done.
Details
buildbot/vexp-code-patch-darwin-x86_64 Build done.
Details
buildbot/vexp-code-patch-darwin-arm64 Build done.
Details
buildbot/vexp-code-patch-linux-x86_64 Build done.
Details
buildbot/vexp-code-patch-windows-amd64 Build done.
Details
buildbot/vexp-code-patch-coordinator Build done.
Details
This pull request has changes conflicting with the target branch.
  • source/blender/editors/sculpt_paint/sculpt_uv.cc
  • source/blender/editors/uvedit/uvedit_unwrap_ops.cc
  • source/blender/geometry/intern/uv_parametrizer.cc
View command line instructions.

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u uv_unwrapping_slim_algorithm_v5_update:glukoz-uv_unwrapping_slim_algorithm_v5_update
git checkout glukoz-uv_unwrapping_slim_algorithm_v5_update
Sign in to join this conversation.
Reviewers
No reviewers
Brecht Van Lommel
Labels
Clear labels   
Interest
Alembic

  
Interest
Animation & Rigging

  
Interest
Asset Browser

  
Interest
Asset Browser Project Overview

  
Interest
Audio

  
Interest
Automated Testing

  
Interest
Blender Asset Bundle

  
Interest
BlendFile

  
Interest
Collada

  
Interest
Compatibility
This issue affects/is about backward or forward compatibility

  
Interest
Compositing

  
Interest
Core

  
Interest
Cycles

  
Interest
Dependency Graph

  
Interest
Development Management

  
Interest
EEVEE

  
Interest
EEVEE & Viewport

  
Interest
Freestyle

  
Interest
Geometry Nodes

  
Interest
Grease Pencil

  
Interest
ID Management

  
Interest
Images & Movies

  
Interest
Import Export

  
Interest
Line Art

  
Interest
Masking

  
Interest
Metal

  
Interest
Modeling

  
Interest
Modifiers

  
Interest
Motion Tracking

  
Interest
Nodes & Physics

  
Interest
OpenGL

  
Interest
Overlay

  
Interest
Overrides

  
Interest
Performance

  
Interest
Physics

  
Interest
Pipeline, Assets & IO

  
Interest
Platforms, Builds & Tests

  
Interest
Python API

  
Interest
Render & Cycles

  
Interest
Render Pipeline

  
Interest
Sculpt, Paint & Texture

  
Interest
Text Editor

  
Interest
Translations

  
Interest
Triaging

  
Interest
Undo

  
Interest
USD

  
Interest
User Interface

  
Interest
UV Editing

  
Interest
VFX & Video

  
Interest
Video Sequencer

  
Interest
Virtual Reality

  
Interest
Vulkan

  
Interest
Wayland
Issues relating to Wayland windowing support.

  
Interest
Workbench

  
Interest: X11

Issues relating to Xorg/X11 support.

  
Legacy
Blender 2.8 Project

  
Legacy
Milestone 1: Basic, Local Asset Browser

  
Legacy
OpenGL Error

  
Meta
Good First Issue

  
Meta
Papercut

  
Meta
Retrospective

  
Meta
Security
Issues relating to security: https://wiki.blender.org/wiki/Process/Vulnerability_Reports

  
Module
Animation & Rigging

  
Module
Core

  
Module
Development Management

  
Module
EEVEE & Viewport

  
Module
Grease Pencil

  
Module
Modeling

  
Module
Nodes & Physics

  
Module
Pipeline, Assets & IO

  
Module
Platforms, Builds & Tests

  
Module
Python API

  
Module
Render & Cycles

  
Module
Sculpt, Paint & Texture

  
Module
Triaging

  
Module
User Interface

  
Module
VFX & Video

  
Platform
FreeBSD

  
Platform
Linux

  
Platform
macOS

  
Platform
Windows

  
Priority
High

  
Priority
Low

  
Priority
Normal

  
Priority
Unbreak Now!

  
Status
Archived

  
Status
Confirmed

  
Status
Duplicate

  
Status
Needs Info from Developers

  
Status
Needs Information from User

  
Status
Needs Triage

  
Status
Resolved

  
Type
Bug

  
Type
Design

  
Type
Known Issue

  
Type
Patch

  
Type
Report

  
Type
To Do

No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
Milestone
Clear milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
No Assignees
3 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#114545
No description provided.
Delete Branch "glukoz/blender:uv_unwrapping_slim_algorithm_v5_update"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?