Bevel Modifier: Inconsistent vertex colors with odd-numbered segment counts #73966

New Issue

Jani Kahrama · 2020-02-18T11:23:54+01:00

Jani Kahrama commented

2020-02-18 11:23:54 +01:00

System Information
Operating system: Windows-10-10.0.18362-SP0 64 Bits
Graphics card: GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19

Blender Version
Broken: version: 2.82 (sub 7), branch: master, commit date: 2020-02-12 16:20, hash: 77d23b0bd7

Short description of error
When using the bevel modifier with a vertex-colored object, using odd-numbered segment counts creates unusable results.

Image 0: Base mesh

Image 1: One-segment bevel, with broken color regions

Image 2: Two-segment bevel, clean regions

Image 3: Three-segment bevel, broken regions

Exact steps for others to reproduce the error
Provided blend file illustrates the issue. Simply adjust the segment count on the simple cylinder that has some vertex colors and bevel weights.

bevelcolors_bug.blend

Preferred result would be for color regions to remain clean with odd-numbered segment counts.

**System Information** Operating system: Windows-10-10.0.18362-SP0 64 Bits Graphics card: GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19 **Blender Version** Broken: version: 2.82 (sub 7), branch: master, commit date: 2020-02-12 16:20, hash: `77d23b0bd7` **Short description of error** When using the bevel modifier with a vertex-colored object, using odd-numbered segment counts creates unusable results. Image 0: Base mesh ![bevel_0_disabled.PNG](https://archive.blender.org/developer/F8347199/bevel_0_disabled.PNG) Image 1: One-segment bevel, with broken color regions ![bevel_1_one_segment.PNG](https://archive.blender.org/developer/F8347202/bevel_1_one_segment.PNG) Image 2: Two-segment bevel, clean regions ![bevel_2_two_segment.PNG](https://archive.blender.org/developer/F8347205/bevel_2_two_segment.PNG) Image 3: Three-segment bevel, broken regions ![bevel_3_three_segment.PNG](https://archive.blender.org/developer/F8347207/bevel_3_three_segment.PNG) **Exact steps for others to reproduce the error** Provided blend file illustrates the issue. Simply adjust the segment count on the simple cylinder that has some vertex colors and bevel weights. [bevelcolors_bug.blend](https://archive.blender.org/developer/F8347211/bevelcolors_bug.blend) Preferred result would be for color regions to remain clean with odd-numbered segment counts.

Jani Kahrama commented

2020-02-18 11:23:54 +01:00

Added subscriber: @Frand

blender-admin commented

2020-02-18 11:23:54 +01:00

#77188 was marked as duplicate of this issue

Jacques Lucke commented

2020-02-18 12:25:16 +01:00

Added subscribers: @howardt, @JacquesLucke

Jacques Lucke commented

2020-02-18 12:25:16 +01:00

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

Jacques Lucke commented

2020-02-18 12:25:16 +01:00

I think this is working as expected. Which color each face has after the bevel is somewhat arbitrary when the number of segments is odd.

@howardt I think we had a similar issue at some point. Maybe you want to give your opinion here. I'll mark it as known issue for now.

I think this is working as expected. Which color each face has after the bevel is somewhat arbitrary when the number of segments is odd. @howardt I think we had a similar issue at some point. Maybe you want to give your opinion here. I'll mark it as known issue for now.

Howard Trickey commented

2020-02-18 12:39:44 +01:00

This is similar to the problem of UV maps with bevels of odd numbers of vertices. It is indeed arbitrary what happens at in such cases. I have some ideas on how to make it feel at least a little more consistent for the UV case, and it is good to have this report to remind me that the same solution should be used for vertex properties, including color, also.

Jani Kahrama commented

2020-02-18 12:39:54 +01:00

The odd-numbered results are regrettably unusable.
An even-numbered bevel in turn generates much unwanted geometry in a pipeline where meshes are subdivided post-bevel. And we would wish to avoid the situation where it's necessary to apply the modifier and fix these color errors manually.

Suggestion: Could the face color results from the next even-numbered bevel be used to determine the colors for the odd-numbered bevels? If you sample the most frequent color for each face region from the next even bevel, the results would probably be much improved. EDIT: No, on closer examination that would not work. Here's hoping another solution can be found.

The odd-numbered results are regrettably unusable. An even-numbered bevel in turn generates much unwanted geometry in a pipeline where meshes are subdivided post-bevel. And we would wish to avoid the situation where it's necessary to apply the modifier and fix these color errors manually. Suggestion: Could the face color results from the next even-numbered bevel be used to determine the colors for the odd-numbered bevels? If you sample the most frequent color for each face region from the next even bevel, the results would probably be much improved. EDIT: No, on closer examination that would not work. Here's hoping another solution can be found.

Alaska commented

2020-05-30 05:15:07 +02:00

Added subscribers: @Stonebana, @Alaska

Howard Trickey commented

2020-06-26 12:53:18 +02:00

Changed status from 'Confirmed' to: 'Resolved'

Howard Trickey closed this issue

2020-06-26 12:53:18 +02:00

Howard Trickey self-assigned this 2020-06-26 12:53:18 +02:00

Howard Trickey commented

2020-06-26 12:53:18 +02:00

I just committed change e3a420c477 which uses some disambiguation rules to help solve this problem.
I will say it fixes this problem, even though the results still do not look great on the example blend: that example has both a UV map and a Loop color assignment. Those two together make for a harder case to solve (there is a detection of "connected islands" that has to pay attention to both of those maps). If one deletes the apparently unused uv map from the example, the result looks good, in my opinion.

I just committed change e3a420c477 which uses some disambiguation rules to help solve this problem. I will say it fixes this problem, even though the results still do not look great on the example blend: that example has both a UV map and a Loop color assignment. Those two together make for a harder case to solve (there is a detection of "connected islands" that has to pay attention to both of those maps). If one deletes the apparently unused uv map from the example, the result looks good, in my opinion.

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

Bevel Modifier: Inconsistent vertex colors with odd-numbered segment counts #73966