Indirect Specular pass problem

Use this forum for general user support and related questions.
Forum rules
Please upload a testscene that allows developers to reproduce the problem, and attach some images.
User avatar
B.Y.O.B.
Developer
Developer
Posts: 4146
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: Indirect Specular pass problem

Post by B.Y.O.B. »

Pixel filtering is now disabled by default, and forced disabled if either
a) bidir and denoising are combined
b) path and light tracing and denoising are combined
steveernst117
Posts: 11
Joined: Sat Apr 04, 2020 5:21 pm

Re: Indirect Specular pass problem

Post by steveernst117 »

Great, I think I understand what was happening with the caustics pass now. Thanks for clarifying, and for pretty much fixing it to solve any confusion in the future!

I still have the problem with the indirect specular pass, however. and the plot has thickened. Here are a few comparison images to better illustrate what's happening.

Combined render (standard view transform color management):
Combined render
Combined render
Indirect Specular pass (standard view transform color management):
Indirect Specular pass
Indirect Specular pass
Assembled passes final result (standard view transform color management):
Assembled passes final result
Assembled passes final result

You can see how there's strange blueness showing up in the ind spec pass that shouldn't be there. I specified that it was using standard view transform color because if you have Filmic color on, your exported png of the third example image will look like this for some reason:
Assembled passes final result (Filmic)
Assembled passes final result (Filmic)

Safe to say, I have no idea why any of this is going on, but I would love to know if it's something I'm doing wrong (and then how to do it right), or if it's really a bug. It's frustrating not being able to use all these passes as expected. Thanks!
User avatar
B.Y.O.B.
Developer
Developer
Posts: 4146
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: Indirect Specular pass problem

Post by B.Y.O.B. »

Please upload a simplified testscene that shows the problem.
steveernst117
Posts: 11
Joined: Sat Apr 04, 2020 5:21 pm

Re: Indirect Specular pass problem

Post by steveernst117 »

It's the same test scene I uploaded in the original post, but here it is again.
luxcore_caustic_tests.blend
(753.47 KiB) Downloaded 197 times
steveernst117
Posts: 11
Joined: Sat Apr 04, 2020 5:21 pm

Re: Indirect Specular pass problem

Post by steveernst117 »

Any news on this issue yet?
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Indirect Specular pass problem

Post by Dade »

steveernst117 wrote: Mon Feb 22, 2021 1:10 am Any news on this issue yet?
It is cause by NaNs. LuxCore film has a safety net to discard NaN samples for the main frame buffer but not for AOVs so you end with a valid complete render but with AOVs containing NaNs.

I'm extending the safety net to AOVs too.
Support LuxCoreRender project with salts and bounties
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Indirect Specular pass problem

Post by Dade »

Actually, it was CPU-only problem because GPU code had already the complete safety net. It should be fixed in next automatic build.
Support LuxCoreRender project with salts and bounties
steveernst117
Posts: 11
Joined: Sat Apr 04, 2020 5:21 pm

Re: Indirect Specular pass problem

Post by steveernst117 »

I didn't understand all of that, but fantastic news! Thanks so much for looking into this stuff.
User avatar
B.Y.O.B.
Developer
Developer
Posts: 4146
Joined: Mon Dec 04, 2017 10:08 pm
Location: Germany
Contact:

Re: Indirect Specular pass problem

Post by B.Y.O.B. »

What was causing the NaNs?
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Indirect Specular pass problem

Post by Dade »

B.Y.O.B. wrote: Tue Feb 23, 2021 8:54 am What was causing the NaNs?
It was caused by the math of homogeneous volume transmittance over an infinite segment (and the infinite segment is caused by having the bottom of the cube and the ground plane overlapped).
Support LuxCoreRender project with salts and bounties
Post Reply