Few bugs with Custom Normals modifiers. #44004

New Issue

Wolthera van Hövell · 2015-03-15T16:14:16+01:00

2015-03-15 16:14:16 +01:00

System Information
Ubuntu 14.04, radeon graphics driver.

Blender Version
Broken: 2.74 RC1

Short description of error
Strange triangulation on the set split normal modifier, and weirdness with edge normals using the transfer mesh data modifier.

okay, I'll add a blend file with both in a minute.

problem(first scene): When having a set split normal modifier on a mesh, and the view mode set to texture+having something else besides the affected mesh selected, results in the affected mesh looking triangulated. In meterial viewmode, it looks normal.
Second problem(second scene.) When having a data transfer of normals from another mesh onto a smooth shaded mesh, deactivating the modifier results in the edge-normals being hard again. Going into edit mode and setting the edges to smooth fixes this. But then, if we go out of edit mode, and toggle the mesh-transfer modifier again, and unhide it again, and the edges are hard again.
So, the mesh transfer modifier is sort of destructive at the moment, and if it's not a bug, I think it would be really helpful if it told users that it's sorta destructive :)

I hope this helps!

**System Information** Ubuntu 14.04, radeon graphics driver. **Blender Version** Broken: 2.74 RC1 **Short description of error** Strange triangulation on the set split normal modifier, and weirdness with edge normals using the transfer mesh data modifier. okay, I'll add a blend file with both in a minute. 1. problem(first scene): When having a set split normal modifier on a mesh, and the view mode set to texture+having something else besides the affected mesh selected, results in the affected mesh looking triangulated. In meterial viewmode, it looks normal. 2. Second problem(second scene.) When having a data transfer of normals from another mesh onto a smooth shaded mesh, deactivating the modifier results in the edge-normals being hard again. Going into edit mode and setting the edges to smooth fixes this. But then, if we go out of edit mode, and toggle the mesh-transfer modifier again, and unhide it again, and the edges are hard again. So, the mesh transfer modifier is sort of destructive at the moment, and if it's not a bug, I think it would be really helpful if it told users that it's sorta destructive :) I hope this helps!

Wolthera van Hövell commented

2015-03-15 16:14:16 +01:00

Changed status to: 'Open'

Bastien Montagne was assigned by Wolthera van Hövell

2015-03-15 16:14:16 +01:00

Wolthera van Hövell commented

2015-03-15 16:14:16 +01:00

Added subscriber: @wolthera

Wolthera van Hövell commented

2015-03-15 16:17:18 +01:00

http://www.pasteall.org/blend/35038 <blend

And this is what I see.

http://www.pasteall.org/blend/35038 <blend ![Screenshot_from_2015-03-15_16:05:25.png](https://archive.blender.org/developer/F151750/Screenshot_from_2015-03-15_16_05_25.png) And this is what I see.

Bastien Montagne commented

2015-03-15 16:29:27 +01:00

Meeeh… I cannot confirm any of those two issues…

This could really be an OpenGL bug, here it works like a charm, can you please try to tweak your OGL settings (like VBO on/off, etc.), and/or try to start in software OpenGL mode (use the blender-softwaregl distributed with official blender, will be slow, but will show whether this is a driver issue or not)?
Cannot reproduce that with current master… I did fix that (or a very similar) bug some weeks ago, are you sure this happens with 2.74-RC1?

Meeeh… I cannot confirm any of those two issues… 1) This could really be an OpenGL bug, here it works like a charm, can you please try to tweak your OGL settings (like VBO on/off, etc.), and/or try to start in software OpenGL mode (use the `blender-softwaregl` distributed with official blender, will be slow, but will show whether this is a driver issue or not)? 2) Cannot reproduce that with current master… I did fix that (or a very similar) bug some weeks ago, are you sure this happens with 2.74-RC1?

Bastien Montagne commented

2015-03-15 16:33:06 +01:00

Attaching the compressed file here, btw. bugswithustomnormals.blend

Attaching the compressed file here, btw. [bugswithustomnormals.blend](https://archive.blender.org/developer/F151763/bugswithustomnormals.blend)

Wolthera van Hövell commented

2015-03-15 16:48:18 +01:00

yes, RC1 straight from the website this morning.

I wouldn't doubt it, however, I just realised that the main difference between material and texture is that material is GLSL and texture is multtexture, perhaps that's the cause? (setting texture display to glsl fixes the issue)
If it's fixed with current master, then that's good. If RC2 comes out, I'll test it again. Maybe the patch slipped between the cracks :)

yes, RC1 straight from the website this morning. 1) I wouldn't doubt it, however, I just realised that the main difference between material and texture is that material is GLSL and texture is multtexture, perhaps that's the cause? (setting texture display to glsl fixes the issue) 2) If it's fixed with current master, then that's good. If RC2 comes out, I'll test it again. Maybe the patch slipped between the cracks :)

Bastien Montagne commented

2015-03-15 17:01:48 +01:00

Just downloaded official 2.74-RC1 to be 100% sure.

Yes, but that still remains a GPU/driver bug imho, we cannot do much about that.
Definitively cannot reproduce that using your file, even on official linux64 2.74-rc1 release…

Just downloaded official 2.74-RC1 to be 100% sure. 1) Yes, but that still remains a GPU/driver bug imho, we cannot do much about that. 2) Definitively cannot reproduce that using your file, even on official linux64 2.74-rc1 release…

Wolthera van Hövell commented

2015-03-15 17:50:10 +01:00

Hm, I do have videocard issues due to darned hybrid graphics...

Ah well, I can work around it. At the very least it's documented now, and if other people have the same issue, they can reopen this.

Hm, I do have videocard issues due to darned hybrid graphics... Ah well, I can work around it. At the very least it's documented now, and if other people have the same issue, they can reopen this.

Campbell Barton commented

2015-03-15 22:57:53 +01:00

Added subscribers: @mont29, @ideasman42

Campbell Barton commented

2015-03-15 22:57:53 +01:00

@mont29, The file looks fine in release builds for me, but this file gives an assert mesh_evaluate.c:390 - BLI_assert(nbr > 2);

The mesh is valid (validate runs OK), so this assert should be fixed?

@mont29, The file looks fine in release builds for me, but this file gives an assert `mesh_evaluate.c:390` - `BLI_assert(nbr > 2);` The mesh is valid (validate runs OK), so this assert should be fixed?

Bastien Montagne commented

2015-03-16 10:42:37 +01:00

Actually, it was the assert that was wrong, fixed in 9542cf041a.

Actually, it was the assert that was wrong, fixed in 9542cf041a.

Julian Eisel commented

2015-03-22 15:55:49 +01:00

Added subscriber: @JulianEisel