Regression: Exact Boolean Fails On Simple Model #89330

New Issue

Philip Holzmann · 2021-06-21T10:39:21+02:00

Philip Holzmann commented

2021-06-21 10:39:21 +02:00

System Information
Operating system: Windows 10 64 bit
Graphics card: Nvidia GTX 750 Ti

Blender Version
Broken: 2.93.0 release, 2.93.1 candidate baa7a53974, 3.0.0 alpha 6afafc46f6,
Worked: 2.91.0 release, 2.92.0 release

Short description of error
Two subsequent union operations result in an empty mesh. If the order of the operations is reversed, or the "Self" flag is enabled, it succeeds. However, there do not seem to be any self intersections in either the input meshes or the intermediate result. (At least by visual inspection.)

Turning the "Self" flag on is not viable though, since it is too memory hungry for the meshes I use.

Triangulating the intermediate result also fixes the issue.

No warnings are output, which makes it easy to have a small part of a model missing in the end.

Exact steps for others to reproduce the error
Open the file.

293_boolean_empty_result.blend

Object A has two boolean modifiers, one to union with B and one to union with C.
As you can see, they result in an empty mesh.
Moving either object can resolve the issue, but the margin seems to be comparatively big.

**System Information** Operating system: Windows 10 64 bit Graphics card: Nvidia GTX 750 Ti **Blender Version** Broken: 2.93.0 release, 2.93.1 candidate baa7a5397486, 3.0.0 alpha 6afafc46f639, Worked: 2.91.0 release, 2.92.0 release **Short description of error** Two subsequent *union* operations result in an empty mesh. If the order of the operations is reversed, or the "Self" flag is enabled, it succeeds. However, there do not seem to be any self intersections in either the input meshes or the intermediate result. (At least by visual inspection.) Turning the "Self" flag on is not viable though, since it is too memory hungry for the meshes I use. Triangulating the intermediate result also fixes the issue. No warnings are output, which makes it easy to have a small part of a model missing in the end. **Exact steps for others to reproduce the error** Open the file. [293_boolean_empty_result.blend](https://archive.blender.org/developer/F10185619/293_boolean_empty_result.blend) Object A has two boolean modifiers, one to union with B and one to union with C. As you can see, they result in an empty mesh. Moving either object can resolve the issue, but the margin seems to be comparatively big.

Philip Holzmann commented

2021-06-21 10:39:21 +02:00

Added subscriber: @Foaly

Philip Holzmann commented

2021-06-21 10:39:47 +02:00

Added subscriber: @howardt

Philip Holzmann commented

2021-06-21 10:39:47 +02:00

@howardt I think this would be for you.

Richard Antalik commented

2021-06-21 14:33:45 +02:00

Changed status from 'Needs Triage' to: 'Confirmed'

Fen commented

2021-06-21 18:11:23 +02:00

Added subscriber: @chemicalcrux

Fen commented

2021-06-21 18:11:23 +02:00

It looks good on the commit before a3f091d7ce ... but I'm having weird problems with conflicting function declarations, and can't see if that commit is responsible or not! Perhaps someone else could try building that?

~~Nevermind that; I got a later commit to work and it's behaving correctly at that point.~~

Turns out I wasn't building Blender with GMP support, so it wasn't even using the exact Boolean modifier (:

a3f091d7ce is, indeed, the first commit where the problem appears.

~~It looks good on the commit before a3f091d7ce ... but I'm having weird problems with conflicting function declarations, and can't see if that commit is responsible or not! Perhaps someone else could try building that?~~ ~~Nevermind that; I got a later commit to work and it's behaving correctly at that point.~~ Turns out I wasn't building Blender with GMP support, so it wasn't even using the exact Boolean modifier (: a3f091d7ce is, indeed, the first commit where the problem appears.

Nico Traber commented

2021-06-23 09:21:49 +02:00

Added subscriber: @Nico-Traber

Philip Holzmann commented

2021-06-23 09:30:52 +02:00

Changing priority to high, since this is a recent regression.

Changing priority to *high*, since this is a recent regression.

Howard Trickey self-assigned this 2021-06-23 14:46:36 +02:00

Howard Trickey commented

2021-06-23 15:23:37 +02:00

For some reason my code is only finding two patches and one cell, which is wrong. Need to investigate more.

Philip Holzmann commented

2021-06-24 09:42:16 +02:00

I think I may have found a simpler case, with just 3 cubes. Maybe that helps?
293_boolean_empty_result_02.blend

2.92
2.93

I think I may have found a simpler case, with just 3 cubes. Maybe that helps? [293_boolean_empty_result_02.blend](https://archive.blender.org/developer/F10192828/293_boolean_empty_result_02.blend) | **2.92** | ![293_boolean_empty_result_02_2.92.png](https://archive.blender.org/developer/F10192831/293_boolean_empty_result_02_2.92.png) | -- | -- | | **2.93** | ![293_boolean_empty_result_02_2.93.png](https://archive.blender.org/developer/F10192832/293_boolean_empty_result_02_2.93.png)

Howard Trickey commented

2021-06-24 14:12:26 +02:00

I think it will help a lot, thanks! It was tedious to track where things went wrong with so many faces in that original file.

blender-admin commented

2021-06-26 23:49:52 +02:00

This issue was referenced by 7223a0348f

This issue was referenced by 7223a0348f53df7e19677f05aaef89852c0f0187

Howard Trickey commented

2021-06-26 23:51:12 +02:00

Changed status from 'Confirmed' to: 'Resolved'

Howard Trickey closed this issue

2021-06-26 23:51:12 +02:00

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

Regression: Exact Boolean Fails On Simple Model #89330