Universal settings (updated?)

[peterguthrie]

I just noticed in the help docs the DMC Sampler adaptive amount has changed from 1.0 to 0.9

http://www.spot3d.com/vray/help/200R...nisettings.htm

Is this correct? why the change?

[joconnell]

Vlado mentioned in a few things that having a fully adaptive workflow might not give vray enough initial samples to work with and cause problems. I asked him in another thread what he'd normally leave adaptive amount at and he said he'd never go over 0.9. If you go over it you might be cutting the absolute minimum samples and not give vray enough sampling information to make good judgements.

[vlado]

Is this correct? why the change?

When I wrote the original tutorial, we didn't have the VRayExtraTex element into which people could plug f.e. a VRayDirt texture. The 1.0 adaptive amount worked well for the RGB image, but it was not enough to get good results in the additional VRayDirt texture - no matter how many subdivs you put in it, it still remained noisy.

Best regards,
Vlado

[3LP]

Yeah right, good to know

What about the first pass as brute force, It looks like it introduce a lot of noise in the image, and we have to push the settings really high to suppress that noise, even if we put the brute force at 24 or 64 in subdivision (it help) but it's still not as smooth as what IR is doing...

Universal settings are great when we are working, but for final images, Universal settings are to "expensive", what is the typical way you guys are using for final renders? revert back to IR or use Brute Force and just wait for hours for a noise free image?

Thanks

Stan

[joconnell]

24 or 64 might not be enough, as you might find that your anti aliasing is limiting how many samples your brute force actually uses. It's a problem of using per-pixel / ray traced / brute force GI that it's making a totally different light simulation for every pixel in your image, so it's always going to come back grainy. Irmap is doing so many cheats and gradual refinements by comparison, and it's also designed to be smooth in the end rather than 100% accurate to the geometry.

Same thing as usual, there's plenty of options in there for what looks good enough to you for the quality you seek to get.

Just curious, what type of main light source are you typically using in your scenes? What type of aa, noise threshold and brute force sampling amounts are you using?

[sandykoufax]

I didn't noticed that so far. thanks for the info.

[3LP]

Well as long as we are going in the Universal Settings, I thought that we shoudn't use any AA Filtering, so for me it's just off.
I'm using mainly Vraydomelight and VrayIES.
For my final renders, I just tried 64 subdiv in bruteforce and 4-12 DMC with noise at 0.005
This give me a descent render but takes ages.

When you are speaking about the aliasing limit, do you mean DMC Min/Max or Aliasing Filtering?
I didn't knew that the DMC or the AAFilter would affect the amount of sampling that is used for the Brute Force, or is that because the adaptive amount is at 1?

What are your thoughts?

Thanks

Stan

24 or 64 might not be enough, as you might find that your anti aliasing is limiting how many samples your brute force actually uses. It's a problem of using per-pixel / ray traced / brute force GI that it's making a totally different light simulation for every pixel in your image, so it's always going to come back grainy. Irmap is doing so many cheats and gradual refinements by comparison, and it's also designed to be smooth in the end rather than 100% accurate to the geometry.

Same thing as usual, there's plenty of options in there for what looks good enough to you for the quality you seek to get.

Just curious, what type of main light source are you typically using in your scenes? What type of aa, noise threshold and brute force sampling amounts are you using?

[joconnell]

Yep - the more max AA samples you give vray to work with, the more the AA sampler tries to do. If you check out the sample rate render element you can get a good idea how many of your min (blue) and red (max) AA samples are getting used in the scene. If you're using more AA samples than you need, your light and material sampling get reduced which can lead to worse results. Say for example you've got a scene with one big glossy material with 40 reflection samples and you're using aa min 1 max 8 in vray, if your AA is trying to clean up the scene instead of your material sampler, it might be doing a bad job of it rather than using less AA and more material samples. It's almost as if the AA sampler is trying to fix everything, and just doing a bad job of it

[zeronove]

here you will find some useful infos about this issue

http://www.interstation3d.com/tutori...yfing_dmc.html

[peterguthrie]

here you will find some useful infos about this issue

http://www.interstation3d.com/tutori...yfing_dmc.html

read it many times, but it's still mostly a mystery

[zeronove]

me too ...and totally agree with you, but it helps ..someway

[joconnell]

I thiiiink I'm starting to get it. I want to get a few concrete examples proving my thoughts, but using the sample rate element and material settings I'm getting decent render times on film res frames with no noise in about 20 / 25 minutes for large elements. Gonna do a few more experiments but the main part of the idea is that the aa sampler is totally bullying all the other samplers and trying to fix your render issues itself, and in a lot of cases it's doing a bad job of it. You can definitely get clean results using only AA but as vlado mentioned on the original universal method, it's a simple way to guarantee good clean results, but it might not be the quickest results in every scene.

[rivoli]

You can definitely get clean results using only AA but as vlado mentioned on the original universal method, it's a simple way to guarantee good clean results, but it might not be the quickest results in every scene.

in my experience is never the quickest result.. looking forward to read more about your findings on this.

[tolgahan]

Deleted post.