Weird bake results with non integer color values #47727

New Issue

System Information
Win7 64
Geforce GTX 960

Blender Version
Broken: (example: 2.77 RC2)

Short description of error
Baking Diffuse and Glossy channels we get weird results when your color need to use the float precision.

Example:
Baking with color (1,0,0) in Diffuse BSDF Node:
http://puu.sh/nzvPw/952908aecc.jpg

Baking with color (1,0.1,.01) in Diffuse BSDF Node:
http://puu.sh/nzvPw/952908aecc.jpg

That bug seams to happen too with the Glossy pass using Glossy BSDF Node.

Exact steps for others to reproduce the error
Change the current active node conected to the output
Change the pass that you want to bake.
Check the results

BugDiffuse.blend

**System Information** Win7 64 Geforce GTX 960 **Blender Version** Broken: (example: 2.77 RC2) **Short description of error** Baking Diffuse and Glossy channels we get weird results when your color need to use the float precision. Example: Baking with color (1,0,0) in Diffuse BSDF Node: http://puu.sh/nzvPw/952908aecc.jpg Baking with color (1,0.1,.01) in Diffuse BSDF Node: http://puu.sh/nzvPw/952908aecc.jpg That bug seams to happen too with the Glossy pass using Glossy BSDF Node. **Exact steps for others to reproduce the error** Change the current active node conected to the output Change the pass that you want to bake. Check the results [BugDiffuse.blend](https://archive.blender.org/developer/F289071/BugDiffuse.blend)

Changed status to: 'Open'

Added subscriber: @vitorbalbio-3

The link to the wrong result:
http://puu.sh/nzwf2/7a5db52513.jpg

The link to the wrong result: http://puu.sh/nzwf2/7a5db52513.jpg

Added subscriber: @dfelinto

I can't reproduce the issue on OSX ec99778 (master - own build)

I can't reproduce the issue on OSX ec99778 (master - own build)

I'm still getting this problem here with both GPU and CPU render.
Win7 x64
GTX 960 4GB
rev 02cabda

I'm still getting this problem here with both GPU and CPU render. Win7 x64 GTX 960 4GB rev 02cabda

Another .blend example with this problem. Just change the Diffuse Node, bake and check the results. I could confirm that this problem was happening in Blender 2.76 also.
Bug.blend

Another .blend example with this problem. Just change the Diffuse Node, bake and check the results. I could confirm that this problem was happening in Blender 2.76 also. [Bug.blend](https://archive.blender.org/developer/F289523/Bug.blend)

Does it happen every single time? Or is it only eventually?

Added subscriber: @mont29

Every single time.
Also why that was marked as "Incomplete"? It seams a very completed report to me.
Bye!

Every single time. Also why that was marked as "Incomplete"? It seams a very completed report to me. Bye!

Incomplete also means 'waiting further info from reporter'…

Added subscriber: @Sergey

You are only baking indirect light component which is not supposed to have any color information in it. So your expectation of wrong and correct baking results are inverted.

Will commit fix which makes sure light pass gets properly divided by color shortly.

You are only baking indirect light component which is not supposed to have any color information in it. So your expectation of wrong and correct baking results are inverted. Will commit fix which makes sure light pass gets properly divided by color shortly.

This issue was referenced by f74c7fcca2

This issue was referenced by f74c7fcca2f9e2603b426fb5c0400f30b4b9ff86

Changed status from 'Open' to: 'Resolved'

Added subscribers: @brecht, @robertm

I can reproduce it on Debian.
But the patch doesn't fix the problem.
Before the patch the direct light was buggy.
After the patch the indirect light is buggy.

@Sergey As I understand it indirect light has to carry color information from surface to surface.
About direct light I'm not sure. It was something like object color + AO, as I remember.
I'm sure @brecht will correct me.

I can reproduce it on Debian. But the patch doesn't fix the problem. Before the patch the direct light was buggy. After the patch the indirect light is buggy. @Sergey As I understand it indirect light has to carry color information from surface to surface. About direct light I'm not sure. It was something like object color + AO, as I remember. I'm sure @brecht will correct me.

Indirect looks correct to me. It's still red, but that's because it's only showing light reflected off red surfaces. If you place the object in an environment with different colored surfaces those colors will show up in the indirect pass too. Direct light can have colors too if you use colored lamps.

Seems that baking produces wrong results only for the RGB(1; 0; 0) diffuse color case.
The intermediate values are ok.

Seems that baking produces wrong results only for the RGB(1; 0; 0) diffuse color case. The intermediate values are ok.

That's what Sergey's commit is supposed to fix, are you sure you're using a build with the fix included? The ones on builder.blender.org don't include the fix yet.

Yes I compiled it myself, hash 89df672.
Hmm, thats strange: To produce correct indirect diffuse color, all RGB channels have to be non-zero.
If only one channel is non-zero then the result is always grey. (although there are different colored objects that should reflect their light onto the baked object)
Can someone confirm this?

Yes I compiled it myself, hash 89df672. Hmm, thats strange: To produce correct indirect diffuse color, all RGB channels have to be non-zero. If only one channel is non-zero then the result is always grey. (although there are different colored objects that should reflect their light onto the baked object) Can someone confirm this?

Sorry for the noise. I came to the conclusion that my understanding of correct and incorrect is wrong.

This issue was referenced by blender/cycles@d3ca7f1eae

This issue was referenced by blender/cycles@d3ca7f1eaeea906bdf32dba6b49d87cf150c48c0

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Weird bake results with non integer color values #47727