Jump to content

PBR Artifacts


Recommended Posts

Hi guys,

I'm currently working on a project which highly depends on indirect diffuse illumination and therefore I thought PBR would be the perfect match for this project. Unfortunately I'm getting strange shading artifacts when using PBR. However, when I'm using raytracing with VEX GI set to Path Tracing i'm not getting any artifacts. Does anyone know how I might solve that problem?

What exactly does the Sampling Quality on the VEX GI node do when set to Path Tracing? Is it essentially the same as upping the min ray samples when rendering with PBR but only for the indirect illumination?

I've also noticed that the color limit parameter doesn't work in my current build of Houdini (10.0.465) on a Mac. Has anybody experienced the same issue?

I've attached two images and the scene file. Thanks.

Merry Christmas everyone,

Dennis

post-4481-126156274021_thumb.pngpost-4481-126156299236_thumb.png

PBR_RT_Artifacts.hip.zip

Link to comment
Share on other sites

This is just a guess, but perhaps the PBR Ray Bias is different to the ray bias in RT mode? ie, perhaps the difference in look is really a ray self-intersection artifact? Perhaps increase the bias in PBR tab.

Hi Jason,

unfortunately that didn't help.

Other ideas or suggestions are highly appreciated.

Greetings,

Dennis

Link to comment
Share on other sites

Actually it wasn't rendered as subd. But unfortunately the subd version looks just the same.

Cheers,

Dennis

post-4481-12618218665_thumb.png

Call me... whatever, but for me it looks like specular/glossy reflections of PBR (obviously absent in GI/pathtrace mode). Are you sure, your shader in PBR performs equally diff/spec as in RT?

ps Unfortunately I can't check your scene now to qualify my suspicions...

Edited by SYmek
Link to comment
Share on other sites

Call me... whatever, but for me it looks like specular/glossy reflections of PBR (obviously absent in GI/pathtrace mode). Are you sure, your shader in PBR performs equally diff/spec as in RT?

ps Unfortunately I can't check your scene now to qualify my suspicions...

That's actually what i thought too. But everything has just a basic clay shader applied at the moment so there shouldn't be any glossy reflections. For those who can't check the scene it might be interesting to know that the whole scene is just lit by luminous geometry.

In another version I tried it with a big area light at the back and instanced spherical area lights for the floor and it resulted in the same issue.

I noticed that i had flipped normals on my walls, but reversing them didn't help either.

Link to comment
Share on other sites

  • 2 weeks later...

I had some artifacts in PBR and SESI said that I should turn down ray samples and use pixel samples to up the quality ... ?

I don't really understand it but it worked.

Thanks, I will try that. For now I went for a different approach using just raytracing. I will post here as soon as I can verify your suggestion. Hope this will work.

Link to comment
Share on other sites

I played with your scene the other day and I couldn't figure out what was causing the problem. When I changed the self-illuminated geo to an area light the problem went away though (and noise was cut too of course).

Hmmm, thats strange. Did you just use a big rectangular area light at the same spot? I tried that in another scene and it seemed to not work for me. Maybe I was doing something wrong. Have to check it again.

As I understood it PBR shoots one diffuse ray at a random light, another random diffuse ray and one specular ray whose propability is weighed by the specular function (hope this makes sense) into the scene. I had the impression, that the random light ray very rarely hit my area light as most of it is occluded by geometry. But maybe I have to check that again. Could you post the modified scene so I can have a look at the setup and verify your results?

Just had a play, pixel samples of 12x12 and ray samples of 16 seems to have fixed it. I'm too impatient to render the whole thing .... laptop isn't very quick ...

I just checked that and indeed it seems to have improved the artifacts. But if you increase the brightness they are still there. The strange thing is, that you used a total of 2304 samples. If I up the pixel samples to 45x45 and use a ray sampels value of 1 which gives me 2025 sampels I get no artifacts at all. I wonder if that is due to the unbiased way pathtracing works, because with more pixel samples I get more more randomness per pixel and thus a more correct result.

Greetings,

Dennis

post-4481-126312279175_thumb.pngpost-4481-126312283818_thumb.png

Edited by Dennis
Link to comment
Share on other sites

Hmmm, thats strange. Did you just use a big rectangular area light at the same spot? I tried that in another scene and it seemed to not work for me. Maybe I was doing something wrong. Have to check it again.

As I understood it PBR shoots one diffuse ray at a random light, another random diffuse ray and one specular ray whose propability is weighed by the specular function (hope this makes sense) into the scene. I had the impression, that the random light ray very rarely hit my area light as most of it is occluded by geometry. But maybe I have to check that again. Could you post the modified scene so I can have a look at the setup and verify your results?

I just checked that and indeed it seems to have improved the artifacts. But if you increase the brightness they are still there. The strange thing is, that you used a total of 2304 samples. If I up the pixel samples to 45x45 and use a ray sampels value of 1 which gives me 2025 sampels I get no artifacts at all. I wonder if that is due to the unbiased way pathtracing works, because with more pixel samples I get more more randomness per pixel and thus a more correct result.

Greetings,

Dennis

post-4481-126312279175_thumb.pngpost-4481-126312283818_thumb.png

A quick question. How long did it take to render those frames?

Rogier

Link to comment
Share on other sites

A quick question. How long did it take to render those frames?

Rogier

The cleaner version with 45x45 pixel samples and min ray samples set to 1 took 23 minutes on my Mac Pro with 16 virtual cores @ 2.26 GHz.

I should add, that the diffuse limit is set to 3 for these renders.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...