Boolean modifier sometimes produces a hole at the intersection point. #46869

New Issue

Sebastian Parborg · 2015-11-25T12:21:03+01:00

Sebastian Parborg commented

2015-11-25 12:21:03 +01:00

System Information
Operating system and graphics card
Gentoo Linux, GPU: AMD 290x (using the open source radeon driver)

Blender Version
Broken: 95bc8389a9
Worked: I haven't check with older versions than git master.

Short description of error
Sometimes when using the boolean modifier, it produces a hole in the mesh instead of filling in the intersection point with faces. I'm guessing this only triggers on certain advanced geometry configurations. I suspect this because I'm using the boolean modifier on a mesh I've got from a brain scan (I did some clean up with the decimate and subdiv mod).

It also seems like there are also certain "sweet spots" for where these holes tend to appear. I've attached a file with an example of one such spot.

Exact steps for others to reproduce the error

Download brain_bool_test.blend
Open it and notice that there is a hole in the mesh
Step the timeline back and forward (notice that the hole dissapears and comes back on the initial keyframe sometimes (more often than not in this case).

**System Information** Operating system and graphics card Gentoo Linux, GPU: AMD 290x (using the open source radeon driver) **Blender Version** Broken: 95bc8389a983cac85eecf9cbc527ed678783e817 Worked: I haven't check with older versions than git master. **Short description of error** Sometimes when using the boolean modifier, it produces a hole in the mesh instead of filling in the intersection point with faces. I'm guessing this only triggers on certain advanced geometry configurations. I suspect this because I'm using the boolean modifier on a mesh I've got from a brain scan (I did some clean up with the decimate and subdiv mod). It also seems like there are also certain "sweet spots" for where these holes tend to appear. I've attached a file with an example of one such spot. **Exact steps for others to reproduce the error** 1. Download [brain_bool_test.blend](https://archive.blender.org/developer/F259102/brain_bool_test.blend) 2. Open it and notice that there is a hole in the mesh 3. Step the timeline back and forward (notice that the hole dissapears and comes back on the initial keyframe sometimes (more often than not in this case).

Sebastian Parborg commented

2015-11-25 12:21:03 +01:00

Changed status to: 'Open'

Sebastian Parborg commented

2015-11-25 12:21:03 +01:00

Added subscriber: @ZedDB

Sergey Sharybin commented

2015-11-25 13:30:16 +01:00

Added subscriber: @Sergey

Sergey Sharybin commented

2015-11-25 13:30:16 +01:00

I don't really see any holes in the mesh actually.

Does it happen with 2.76b official release?
Your meshes aren't really manifolds. Please make sure you don't have duplicated vertices and try interseciton with a box instead of the plane.
Attach a screenshot of the problem if removing doubles and using box instead of the plane doesn't help.

I don't really see any holes in the mesh actually. - Does it happen with 2.76b official release? - Your meshes aren't really manifolds. Please make sure you don't have duplicated vertices and try interseciton with a box instead of the plane. - Attach a screenshot of the problem if removing doubles and using box instead of the plane doesn't help.

Sebastian Parborg commented

2015-11-25 14:49:34 +01:00

Ah, I had multiple problems with the boolean operator it seems... The one I attached were because of doubles. Or at least it seems to be the case as I didn't manage to get it to bug out anymore.

However the reason that I noticed that something was wrong were because when I tried to render the moving intersection plane with freestyle as a video, the lines would vanish on certain frames. It was really noticable on the bigger part of the brain. But because the smaller one (the one I attached previously) had the hole problem, and it were much faster when stepping between keyframes, I thought that was the main cause of the problem.

But when I tried to remove doubles on the big brain, none were found and removed. I should have stated previously that I have tried with a box shape instead of a plane, but it didn't help. Thankfully I did manage to find the cause for why the lines didn't render sometimes on certain frames. It seems like the boolean operation fails sometimes. So the brain doesn't get sliced at all sometimes on certain frames.

I've attached a new file with one of those frames (frame 69). Step to frame 68/70 and then back to 69. For me the boolean operation seems to work sometimes and sometimes it doesn't. Note that this time it is intersecting with a box shape...

brain_bool_test.blend

Ah, I had multiple problems with the boolean operator it seems... The one I attached were because of doubles. Or at least it seems to be the case as I didn't manage to get it to bug out anymore. However the reason that I noticed that something was wrong were because when I tried to render the moving intersection plane with freestyle as a video, the lines would vanish on certain frames. It was really noticable on the bigger part of the brain. But because the smaller one (the one I attached previously) had the hole problem, and it were much faster when stepping between keyframes, I thought that was the main cause of the problem. But when I tried to remove doubles on the big brain, none were found and removed. I should have stated previously that I have tried with a box shape instead of a plane, but it didn't help. Thankfully I did manage to find the cause for why the lines didn't render sometimes on certain frames. It seems like the boolean operation fails sometimes. So the brain doesn't get sliced at all sometimes on certain frames. I've attached a new file with one of those frames (frame 69). Step to frame 68/70 and then back to 69. For me the boolean operation seems to work sometimes and sometimes it doesn't. Note that this time it is intersecting with a box shape... [brain_bool_test.blend](https://archive.blender.org/developer/F259192/brain_bool_test.blend)

Sebastian Parborg commented

2015-11-25 14:54:18 +01:00

I can attach a screenshot and try earlier versions, but I thought because the new error doesn't seem to be because of doubles, I'll just upload a blend file again to see if you can reproduce it...

Sergey Sharybin commented

2015-11-25 15:00:42 +01:00

Can see the issue with the attached model. Will look later in possible fixes.

Sebastian Parborg commented

2015-12-12 01:36:34 +01:00

Added subscriber: @ideasman42

Sebastian Parborg commented

2015-12-12 01:36:34 +01:00

I guess the solution to this problem will be the new bmesh boolean stuff? I mean there is no point in trying to fix the problem with carve if it's going the get phased out soon anyways, right?

I saw that the bmesh stuff got merged into the master branch recently. I'm guessing that it is still a WIP as it fails to produce any good results even with a simple cube on cube boolean modifier action.

However, it seems to be much faster than the carve based boolean modifer for me. Now I can playback the (corrupt) brain animation at 3fps! Where it takes serveral seconds to step one frame with the carve lib.

@ideasman42 Just in case Sergey didn't mention this bug to you, the brain mesh could be a good test case for the new boolean operator. Because if it manages to construct a mesh where the carve lib failed, it would be a nice improvement (besides the potential speed gain).

Edit: Gah, I can't edit the task to update the attached blend file with the most recent one. Be sure to use the one in the comments :/

I guess the solution to this problem will be the new bmesh boolean stuff? I mean there is no point in trying to fix the problem with carve if it's going the get phased out soon anyways, right? I saw that the bmesh stuff got merged into the master branch recently. I'm guessing that it is still a WIP as it fails to produce any good results even with a simple cube on cube boolean modifier action. However, it seems to be much faster than the carve based boolean modifer for me. Now I can playback the (corrupt) brain animation at 3fps! Where it takes serveral seconds to step one frame with the carve lib. @ideasman42 Just in case Sergey didn't mention this bug to you, the brain mesh could be a good test case for the new boolean operator. Because if it manages to construct a mesh where the carve lib failed, it would be a nice improvement (besides the potential speed gain). Edit: Gah, I can't edit the task to update the attached blend file with the most recent one. Be sure to use the one in the comments :/

Bastien Montagne commented

2016-01-22 15:33:41 +01:00

Added subscriber: @mont29

Sergey Sharybin was assigned by Bastien Montagne

2016-01-22 15:33:41 +01:00

Bastien Montagne commented

2016-01-22 15:33:41 +01:00

Assigning to you @Sergey for now, but maybe we could close it as TODO since we aim at replacing carve with new bmesh code?

Sergey Sharybin commented

2016-01-24 20:34:56 +01:00

Changed status from 'Open' to: 'Archived'

Sergey Sharybin closed this issue

2016-01-24 20:34:56 +01:00

Sergey Sharybin commented

2016-01-24 20:34:56 +01:00

Think this was already on Campbell's radars, and he've got an own task now to gather all the complicated cases for the boolean operation.

Personally, would rather not spend time on trying to fix Carve now, but maybe help Campbell to investigate BMesh boolean remaining TODOs instead. Hence, would consider a BMesh Boolean TODO actually.

Think this was already on Campbell's radars, and he've got an own task now to gather all the complicated cases for the boolean operation. Personally, would rather not spend time on trying to fix Carve now, but maybe help Campbell to investigate BMesh boolean remaining TODOs instead. Hence, would consider a BMesh Boolean TODO actually.

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

Boolean modifier sometimes produces a hole at the intersection point. #46869