Blender 2.93. crashes with meshes with zero length edges #92484

New Issue

Paolo Furani · 2021-10-25T23:24:50+02:00

Paolo Furani commented

2021-10-25 23:24:50 +02:00

System Information
Operating system: Linux-4.15.0-159-generic-x86_64-with-glibc2.27 64 Bits
Graphics card: Mesa DRI Intel(R) HD Graphics 5500 (BDW GT2) Intel Open Source Technology Center 4.6 (Core Profile) Mesa 20.0.8

Blender Version
Broken: version: 2.93.6 Release Candidate, branch: master, commit date: 2021-10-06 10:52, hash: 0930a70e81
Worked: blender-2.83.19-candidate+v283.edee6f911c81

Short description of error
Hi I thought it was important to report this crash with Blender 2.93, I have previously reported this problem which now seems solved in Blender 3.0 as you can see here , but this is still present in Blender 2.93.6LTS and being a version that will be used for a long time it would be useful to remove this crash

crash report

# Blender 2.93.6, Commit date: 2021-10-06 10:52, Hash 0930a70e8110
File written by newer Blender binary (300.36), expect loss of data!  # Error
bpy.context.space_data.params.filename = "retopology-experiments_2.93.6-lts.blend"  # Property
Saved "retopology-experiments_2.93.6-lts.blend"  # Info
bpy.ops.object.duplicates_make_real()  # Operator
bpy.ops.object.editmode_toggle()  # Operator
bpy.ops.object.editmode_toggle()  # Operator
bpy.ops.object.convert(target='MESH')  # Operator
bpy.context.space_data.context = 'MODIFIER'  # Property
bpy.context.space_data.context = 'TOOL'  # Property
bpy.context.space_data.context = 'RENDER'  # Property
bpy.context.space_data.context = 'DATA'  # Property
bpy.context.object.data.remesh_mode = 'QUAD'  # Property
bpy.data.window_managers["WinMan"].(null) = False  # Property
bpy.data.window_managers["WinMan"].(null) = True  # Property
bpy.data.window_managers["WinMan"].(null) = True  # Property
bpy.ops.object.quadriflow_remesh(use_mesh_symmetry=False, use_preserve_sharp=True, use_preserve_boundary=True)  # Operator

# backtrace
/home/pafu/Software/blender-2.93.6/blender(BLI_system_backtrace+0x20) [0xa594750]
/home/pafu/Software/blender-2.93.6/blender() [0xf6024d]
/lib/x86_64-linux-gnu/libc.so.6(+0x3f040) [0x7f8152e12040]
/home/pafu/Software/blender-2.93.6/blender(AUD_Device_getRate+0x3d) [0x890408d]
/home/pafu/Software/blender-2.93.6/blender(BKE_sound_scene_playing+0x20) [0x104b5d0]
/home/pafu/Software/blender-2.93.6/blender(wm_event_do_handlers+0xa6) [0x1313b36]
/home/pafu/Software/blender-2.93.6/blender(WM_main+0x20) [0x130a6d0]
/home/pafu/Software/blender-2.93.6/blender(main+0x31e) [0xe87bce]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f8152df4bf7]
/home/pafu/Software/blender-2.93.6/blender() [0xf5cbcc]

# Python backtrace

When I execute the Remesh command from the obtained object Blender goes into Crash, while if I recreate the same geometric composition in Blender 2.83 where the booleans with the collections are not supported the problem does not happen and the calculation of the Remesh command is successful and generates the new topology.

The modifier was applied to a copy of Cube with holes by using CTRL+A > Visual Geometry to Mesh

Exact steps for others to reproduce the error

Open .blend file
Select the cube on the left with the modifiers applied   and perform the procedure below
{nav Object Data Properties > Remesh > Quad > QuadriFlow Remesh}
Refer image for the settings

retopology-experiments_2.93.6-lts.blend

This is the same file but I've remade the boolean without collection that is not supported by 2.83LTS and works fine
remesh-B2.83lts.blend

**System Information** Operating system: Linux-4.15.0-159-generic-x86_64-with-glibc2.27 64 Bits Graphics card: Mesa DRI Intel(R) HD Graphics 5500 (BDW GT2) Intel Open Source Technology Center 4.6 (Core Profile) Mesa 20.0.8 **Blender Version** Broken: version: 2.93.6 Release Candidate, branch: master, commit date: 2021-10-06 10:52, hash: `0930a70e81` Worked: blender-2.83.19-candidate+v283.edee6f911c81 **Short description of error** Hi I thought it was important to report this crash with Blender 2.93, I have previously reported this problem which now seems solved in Blender 3.0 as you can see [here ](https://developer.blender.org/T92391) , but this is still present in Blender 2.93.6LTS and being a version that will be used for a long time it would be useful to remove this crash **crash report** ``` # Blender 2.93.6, Commit date: 2021-10-06 10:52, Hash 0930a70e8110 File written by newer Blender binary (300.36), expect loss of data! # Error bpy.context.space_data.params.filename = "retopology-experiments_2.93.6-lts.blend" # Property Saved "retopology-experiments_2.93.6-lts.blend" # Info bpy.ops.object.duplicates_make_real() # Operator bpy.ops.object.editmode_toggle() # Operator bpy.ops.object.editmode_toggle() # Operator bpy.ops.object.convert(target='MESH') # Operator bpy.context.space_data.context = 'MODIFIER' # Property bpy.context.space_data.context = 'TOOL' # Property bpy.context.space_data.context = 'RENDER' # Property bpy.context.space_data.context = 'DATA' # Property bpy.context.object.data.remesh_mode = 'QUAD' # Property bpy.data.window_managers["WinMan"].(null) = False # Property bpy.data.window_managers["WinMan"].(null) = True # Property bpy.data.window_managers["WinMan"].(null) = True # Property bpy.ops.object.quadriflow_remesh(use_mesh_symmetry=False, use_preserve_sharp=True, use_preserve_boundary=True) # Operator # backtrace /home/pafu/Software/blender-2.93.6/blender(BLI_system_backtrace+0x20) [0xa594750] /home/pafu/Software/blender-2.93.6/blender() [0xf6024d] /lib/x86_64-linux-gnu/libc.so.6(+0x3f040) [0x7f8152e12040] /home/pafu/Software/blender-2.93.6/blender(AUD_Device_getRate+0x3d) [0x890408d] /home/pafu/Software/blender-2.93.6/blender(BKE_sound_scene_playing+0x20) [0x104b5d0] /home/pafu/Software/blender-2.93.6/blender(wm_event_do_handlers+0xa6) [0x1313b36] /home/pafu/Software/blender-2.93.6/blender(WM_main+0x20) [0x130a6d0] /home/pafu/Software/blender-2.93.6/blender(main+0x31e) [0xe87bce] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f8152df4bf7] /home/pafu/Software/blender-2.93.6/blender() [0xf5cbcc] # Python backtrace ``` When I execute the Remesh command from the obtained object Blender goes into Crash, while if I recreate the same geometric composition in Blender 2.83 where the booleans with the collections are not supported the problem does not happen and the calculation of the Remesh command is successful and generates the new topology. - The modifier was applied to a copy of Cube with holes by using **CTRL+A** > **Visual Geometry to Mesh** **Exact steps for others to reproduce the error** - Open .blend file - Select the cube on the left with the modifiers applied   and perform the procedure below - {nav Object Data Properties > Remesh > Quad > QuadriFlow Remesh} - Refer image for the settings ![quad-remsh-bug.png](https://archive.blender.org/developer/F11295360/quad-remsh-bug.png) [retopology-experiments_2.93.6-lts.blend](https://archive.blender.org/developer/F11503334/retopology-experiments_2.93.6-lts.blend) This is the same file but I've remade the boolean without collection that is not supported by 2.83LTS and works fine [remesh-B2.83lts.blend](https://archive.blender.org/developer/F11366174/remesh-B2.83lts.blend)

Paolo Furani commented

2021-10-25 23:24:50 +02:00

Added subscriber: @pafurijaz-4

Richard Antalik commented

2021-10-26 05:20:27 +02:00

Added subscriber: @iss

Richard Antalik commented

2021-10-26 05:20:27 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Richard Antalik commented

2021-10-26 05:20:27 +02:00

Can't reproduce crash in 2.93.6

In provided backtrace this crashed in audaspace module. Can you re-check if this happens when you set audio device to None in preferences? If crash doesn;t happen it may be worthwhile to trigger the crash with more specific steps or investigate what hardware is needed to reproduce.

Can't reproduce crash in 2.93.6 In provided backtrace this crashed in audaspace module. Can you re-check if this happens when you set audio device to None in preferences? If crash doesn;t happen it may be worthwhile to trigger the crash with more specific steps or investigate what hardware is needed to reproduce.

Paolo Furani commented

2021-10-26 11:39:07 +02:00

Hi I also tried disabling the sound from the preferences and I also tried with another Linux distribution on a USB stick with PuppyLinux Fossa 20.04 LTS, and with both I have the same problem of anomalous crash below the new reports obtained from these crashes of the two different Ubuntu linux operating systems with the same hardware.

report with audio device to None PuppyLinux 20.04LTS

# Blender 2.93.6, Commit date: 2021-10-06 10:52, Hash 0930a70e8110
bpy.context.space_data.context = 'DATA'  # Property
bpy.context.object.data.remesh_mode = 'QUAD'  # Property
bpy.data.window_managers["WinMan"].(null) = False  # Property
bpy.data.window_managers["WinMan"].(null) = True  # Property
bpy.data.window_managers["WinMan"].(null) = True  # Property
bpy.ops.object.quadriflow_remesh(use_mesh_symmetry=False, use_preserve_sharp=True, use_preserve_boundary=True)  # Operator

# backtrace
/home/software/blender-2.93.6/blender(BLI_system_backtrace+0x20) [0xa594750]
/home/software/blender-2.93.6/blender() [0xf6024d]
/lib/x86_64-linux-gnu/libc.so.6(+0x46210) [0x7fccb01b0210]
/home/software/blender-2.93.6/blender(AUD_Device_getRate+0x3d) [0x890408d]
/home/software/blender-2.93.6/blender(BKE_sound_scene_playing+0x20) [0x104b5d0]
/home/software/blender-2.93.6/blender(wm_event_do_handlers+0xa6) [0x1313b36]
/home/software/blender-2.93.6/blender(WM_main+0x20) [0x130a6d0]
/home/software/blender-2.93.6/blender(main+0x31e) [0xe87bce]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3) [0x7fccb01910b3]
/home/software/blender-2.93.6/blender() [0xf5cbcc]

# Python backtrace

report with audio device to None Ubuntu 18.04LTS

# Blender 2.93.6, Commit date: 2021-10-06 10:52, Hash 0930a70e8110
bpy.context.space_data.context = 'DATA'  # Property
bpy.context.object.data.remesh_mode = 'QUAD'  # Property
bpy.data.window_managers["WinMan"].(null) = False  # Property
bpy.data.window_managers["WinMan"].(null) = True  # Property
bpy.data.window_managers["WinMan"].(null) = True  # Property
bpy.ops.object.quadriflow_remesh(use_mesh_symmetry=False, use_preserve_sharp=True, use_preserve_boundary=True)  # Operator

# backtrace
/home/pafu/Software/blender-2.93.6/blender(BLI_system_backtrace+0x20) [0xa594750]
/home/pafu/Software/blender-2.93.6/blender() [0xf6024d]
/lib/x86_64-linux-gnu/libc.so.6(+0x3f040) [0x7f09cdb8f040]
/home/pafu/Software/blender-2.93.6/blender(AUD_Device_getRate+0x3d) [0x890408d]
/home/pafu/Software/blender-2.93.6/blender(BKE_sound_scene_playing+0x20) [0x104b5d0]
/home/pafu/Software/blender-2.93.6/blender(wm_event_do_handlers+0xa6) [0x1313b36]
/home/pafu/Software/blender-2.93.6/blender(WM_main+0x20) [0x130a6d0]
/home/pafu/Software/blender-2.93.6/blender(main+0x31e) [0xe87bce]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f09cdb71bf7]
/home/pafu/Software/blender-2.93.6/blender() [0xf5cbcc]

# Python backtrace

   I wonder if you have selected the options as described above and visible in this image below.   

The sound problem had already been addressed in this **[report ]]below, then solved with a modification of the QuadRemesh code visible[ https:*developer.blender.org/rB10fb5cc58d210592ef0ae410e305be906ad5ce51 | here .  ** 

Greetings

Hi I also tried disabling the sound from the preferences and I also tried with another Linux distribution on a USB stick with PuppyLinux Fossa 20.04 LTS, and with both I have the same problem of anomalous crash below the new reports obtained from these crashes of the two different Ubuntu linux operating systems with the same hardware. **report with audio device to None PuppyLinux 20.04LTS** ``` # Blender 2.93.6, Commit date: 2021-10-06 10:52, Hash 0930a70e8110 bpy.context.space_data.context = 'DATA' # Property bpy.context.object.data.remesh_mode = 'QUAD' # Property bpy.data.window_managers["WinMan"].(null) = False # Property bpy.data.window_managers["WinMan"].(null) = True # Property bpy.data.window_managers["WinMan"].(null) = True # Property bpy.ops.object.quadriflow_remesh(use_mesh_symmetry=False, use_preserve_sharp=True, use_preserve_boundary=True) # Operator # backtrace /home/software/blender-2.93.6/blender(BLI_system_backtrace+0x20) [0xa594750] /home/software/blender-2.93.6/blender() [0xf6024d] /lib/x86_64-linux-gnu/libc.so.6(+0x46210) [0x7fccb01b0210] /home/software/blender-2.93.6/blender(AUD_Device_getRate+0x3d) [0x890408d] /home/software/blender-2.93.6/blender(BKE_sound_scene_playing+0x20) [0x104b5d0] /home/software/blender-2.93.6/blender(wm_event_do_handlers+0xa6) [0x1313b36] /home/software/blender-2.93.6/blender(WM_main+0x20) [0x130a6d0] /home/software/blender-2.93.6/blender(main+0x31e) [0xe87bce] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3) [0x7fccb01910b3] /home/software/blender-2.93.6/blender() [0xf5cbcc] # Python backtrace ``` **report with audio device to None Ubuntu 18.04LTS** ``` # Blender 2.93.6, Commit date: 2021-10-06 10:52, Hash 0930a70e8110 bpy.context.space_data.context = 'DATA' # Property bpy.context.object.data.remesh_mode = 'QUAD' # Property bpy.data.window_managers["WinMan"].(null) = False # Property bpy.data.window_managers["WinMan"].(null) = True # Property bpy.data.window_managers["WinMan"].(null) = True # Property bpy.ops.object.quadriflow_remesh(use_mesh_symmetry=False, use_preserve_sharp=True, use_preserve_boundary=True) # Operator # backtrace /home/pafu/Software/blender-2.93.6/blender(BLI_system_backtrace+0x20) [0xa594750] /home/pafu/Software/blender-2.93.6/blender() [0xf6024d] /lib/x86_64-linux-gnu/libc.so.6(+0x3f040) [0x7f09cdb8f040] /home/pafu/Software/blender-2.93.6/blender(AUD_Device_getRate+0x3d) [0x890408d] /home/pafu/Software/blender-2.93.6/blender(BKE_sound_scene_playing+0x20) [0x104b5d0] /home/pafu/Software/blender-2.93.6/blender(wm_event_do_handlers+0xa6) [0x1313b36] /home/pafu/Software/blender-2.93.6/blender(WM_main+0x20) [0x130a6d0] /home/pafu/Software/blender-2.93.6/blender(main+0x31e) [0xe87bce] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f09cdb71bf7] /home/pafu/Software/blender-2.93.6/blender() [0xf5cbcc] # Python backtrace ```    I wonder if you have selected the options as described above and visible in this image below.    ![Options-quadR.png](https://archive.blender.org/developer/F11518863/Options-quadR.png) The sound problem had already been addressed in this **[report ]]**below, then solved with a modification of the QuadRemesh code visible**[[ https:*developer.blender.org/rB10fb5cc58d210592ef0ae410e305be906ad5ce51 | here ](https:*developer.blender.org/T92355).  **  Greetings

Paolo Furani commented

2021-10-26 20:12:35 +02:00

Changed status from 'Needs User Info' to: 'Needs Triage'

Richard Antalik commented

2021-10-27 04:31:02 +02:00

I did follow the steps as described.

Now checked with CTRL+A > Visual Geometry to Mesh on second cube and I can reproduce some issue. Not sure if this can be resolved by backporting 10fb5cc58d. I got confused by seeing audaspace in backtrace.

But remeshing cube on left as described did not cause crash here.

I did follow the steps as described. Now checked with CTRL+A > Visual Geometry to Mesh on second cube and I can reproduce some issue. Not sure if this can be resolved by backporting 10fb5cc58d21. I got confused by seeing audaspace in backtrace. But remeshing cube on left as described did not cause crash here.

Paolo Furani commented

2021-10-27 11:27:52 +02:00

In #92484#1242048, @iss wrote:
I did follow the steps as described.

Now checked with CTRL+A > Visual Geometry to Mesh on second cube and I can reproduce some issue. Not sure if this can be resolved by backporting 10fb5cc58d. I got confused by seeing audaspace in backtrace.

Hi, I don't understand what you mean by "I can able to reproduce some issue".   

Can you be clearer in the problem, and have you tested it on a Linux operating system?

   However this is a problem that occurs in versions after 2.83 LTS, this problem does not occur, and the solution made in version 3.0 adds a check that was not needed before, I will do further checks on the collections and how they process the meshes with the modifier of the booleans.  

Perhaps the problems are related to this, namely how collections are processed with booleans and generate non-manifold meshes.

> In #92484#1242048, @iss wrote: > I did follow the steps as described. > > Now checked with CTRL+A > Visual Geometry to Mesh on second cube and I can reproduce some issue. Not sure if this can be resolved by backporting 10fb5cc58d21. I got confused by seeing audaspace in backtrace. Hi, I don't understand what you mean by "I can able to reproduce some issue".    Can you be clearer in the problem, and have you tested it on a Linux operating system?    However this is a problem that occurs in versions after 2.83 LTS, this problem does not occur, and the solution made in version 3.0 adds a check that was not needed before, I will do further checks on the collections and how they process the meshes with the modifier of the booleans.   Perhaps the problems are related to this, namely how collections are processed with booleans and generate non-manifold meshes.

Richard Antalik commented

2021-10-28 10:38:17 +02:00

Added subscribers: @ZedDB, @lichtwerk

Richard Antalik commented

2021-10-28 10:38:17 +02:00

I am testing on Windows, which shouldn't matter. I was able to cause Blender to freeze (at least it looked like freeze), but not sure if this has anything to do with this problem.
Checking again with debugger, it could be infinite loop or just very very slow, so likely is not the same issue.

@ZedDB, @lichtwerk can you check if you can reproduce?

I am testing on Windows, which shouldn't matter. I was able to cause Blender to freeze (at least it looked like freeze), but not sure if this has anything to do with this problem. Checking again with debugger, it could be infinite loop or just very very slow, so likely is not the same issue. @ZedDB, @lichtwerk can you check if you can reproduce?

Paolo Furani commented

2021-10-28 11:15:53 +02:00

I did a test both with and without collections, but on Blender 2.8 it works because the boolean operations do the automatic merge of the vertices which no longer happens with Blender 2.93 and blender 3.0, I do not know the reason but this is the problem that it no longer allows you to run booleans without having meshes getting corrupted, now I will probably be told it should work like that, but this didn't happen before so all the problems are related for this part of the code that has changed.

Sebastian Parborg commented

2021-10-28 12:43:45 +02:00

Closed as duplicate of #92355

Sebastian Parborg closed this issue

2021-10-28 12:43:45 +02:00

Paolo Furani commented

2021-10-28 12:49:23 +02:00

@ZedDB But this happen also in the 2.93. and the problem is related to boolean issue that make error?
Why you closed this?
How I can report this problem generated by error and this is also solved for blender 2.93.6 LTS?

@ZedDB But this happen also in the 2.93. and the problem is related to boolean issue that make error? Why you closed this? How I can report this problem generated by error and this is also solved for blender 2.93.6 LTS?

Sebastian Parborg commented

2021-10-28 12:49:40 +02:00

Added subscriber: @howardt

Sebastian Parborg commented

2021-10-28 12:49:40 +02:00

Thanks for pointing out that I forgot to merge this fix into the 2.93 branch.

However you have now filed the same bug three times. I must ask you to not file out the same bug multiple times.
You are spamming the bug tracker and wasting a lot of time for everyone involved.

If you had any additional issues related to the first report that I fixed and closed, reply in that task, do not open a new one. (For example after I closed it as fixed, you should have replied there telling me that it isn't fix in the 2.93 release)

In addition to this, you didn't do what I told you to do.
The reason this happens is that the boolean re-meshing that is in 2.93 and 3.0 now produces non manifold geometry where in blender 2.8 it didn't.
You should file out a report about that and "@" me and @howardt in it.

Thanks for pointing out that I forgot to merge this fix into the 2.93 branch. However you have now filed the same bug three times. I must ask you to not file out the same bug multiple times. You are spamming the bug tracker and wasting a lot of time for everyone involved. If you had any additional issues related to the first report that I fixed and closed, reply in that task, do not open a new one. (For example after I closed it as fixed, you should have replied there telling me that it isn't fix in the 2.93 release) In addition to this, you didn't do what I told you to do. The reason this happens is that the boolean re-meshing that is in 2.93 and 3.0 now produces non manifold geometry where in blender 2.8 it didn't. You should file out a report about that and "@" me and @howardt in it.

Paolo Furani commented

2021-10-28 12:59:22 +02:00

it was not my intention to spam but I don't really know how to report this problem but I also pointed out that this was not happening in version 2.83 vertices are merged automatically.
OK I create the report for non-mainfold meshes if someone hasn't done this already.

it was not my intention to spam but I don't really know how to report this problem but I also pointed out that this was not happening in version 2.83 vertices are merged automatically. OK I create the report for non-mainfold meshes if someone hasn't done this already.

Sebastian Parborg commented

2021-10-28 13:07:30 +02:00

The general idea is that you file out one report for all blender versions that are affected.

In this case it was my fault that I didn't fix it in the 2.93 branch as well.
So no matter how many versions of blender that are affected, only one report should be filed.
If something as not fixed even if it was closed, then you should not open a new report but reply in the closed report.

If you get no reply in 2-3 days, you can go to blender.chat and ask in the #blender-coders or #blender-triagers channel to see if you can poke someone.

The general idea is that you file out one report for all blender versions that are affected. In this case it was my fault that I didn't fix it in the 2.93 branch as well. So no matter how many versions of blender that are affected, only one report should be filed. If something as not fixed even if it was closed, then you should not open a new report but reply in the closed report. If you get no reply in 2-3 days, you can go to blender.chat and ask in the `#blender-coders` or `#blender-triagers` channel to see if you can poke someone.

Paolo Furani commented

2021-10-28 13:10:16 +02:00

OK thanks.

Sebastian Parborg commented

2021-10-28 13:21:20 +02:00

Also note that I will merge the fix into the 2.93 branch, so this should be in the next 2.93 release.
If that wasn't clear. Thanks again for pointing this out.

Also note that I will merge the fix into the 2.93 branch, so this should be in the next 2.93 release. If that wasn't clear. Thanks again for pointing this out.

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Blender 2.93. crashes with meshes with zero length edges #92484