Heya
I keep reading and reading about it. I also went over few “tutorials” and tips from users on forum as to how to set it up but on all occasions I failed.
Could some1 be so kind and post for me a maya scene with it set up please?
Here you go. Once the scene is set up in this way, you can use the noise threshold for the “Adaptive” sampler to control the quality (and the light cache subdivs, if needed).
Best regards,
Vlado
univ_settings.zip (7.75 KB)
Thanks Vlado !
you can use the noise threshold for the “Adaptive” sampler to control the quality
So do you mean AA Adaptive DMC Threshold or DMC Sampler threshold ?
And am I right that I should not increase any of the material/lights/etc samples right? Just threshold? And that I can control AA threshold separately to fine tune AA/edges right? OR should that be “Lock Threshold to dmc sampler threshold” on ?
I’m pretty much re-asking what I think I read in past but I’m not sure if that’s 100% correct tbh.
Here is a render with dmc sampler threshold set to 0.0001… I must be doing something wrong because it took long to render and its noisy :- (
The one in the image sampler section.
And am I right that I should not increase any of the material/lights/etc samples right? Just threshold? And that I can control AA threshold separately to fine tune AA/edges right? OR should that be “Lock Threshold to dmc sampler threshold” on ?Don’t touch anything else.
Here is a render with dmc sampler threshold set to 0.0001… I must be doing something wrong because it took long to render and its noisy :- (You have double gamma on your image - it makes it look a lot noisier than it actually is.
Btw this was done with the latest 3.0 nightly build; if you used an older build, it might not come out quite right (there were some changes in the color mapping settings).
Best regards,
Vlado
Oh yea crap the double gamma has me :o. I just rendered ur scene straight away. So now that confuses me !
Proper gamma workflow - jezus I cant believe I’m asking for this…
Save as EXR
VFB = Display color in SRGB space
Color mapping
Gamma 2.2
Dont affect colors, only adaption.
Incoming textures any 16/8 bit enably vray gamma in file node.
Is this correct or I’m wrong?
I’ve also updated vray to today’s version
Hmm, the today’s 3.0 builds do not have the “Don’t affect colors” option; it has been renamed to “Mode” and its default value is “Color mapping only, no gamma”. I just verified and it saves correctly an .exr file that looks fine.
Best regards,
Vlado
Vlado, using these settings are you putting gamma correct nodes on your shaders for straight diffuse colors (not derived from external images) when rendering with V-Ray for Maya? Why or why not?
Eh, do we have to get into that 
The universal settings as such have nothing to do with the gamma workflow, they are just for image sampling. In general, the gamma nodes will slow down the rendering - they take a bit of time to compute even for a solid color, so I would avoid using them even if only for that reason. I hope that Maya will implement a better color management solution soon.
Best regards,
Vlado
Sorry, I just saw the linear workflow post above and wanted to cover all the bases. Thanks for the feedback.
Just use in file node the extra vray attribute to correct your gamma. Dont use the mayas gama node thini. None the less back to topic.
I had to roll back to 2.4 for a while and over ther it seem to work very well. Actually I’m quite stunned how good it works and how well it scales from 1k resolution all the way up to 8k. Before I had to redo a lot of materials as the render ended up being 24h long where my expectations vere 20 min. None the less need more testing but so far, well curse me for being old fashion and trying to control all settings 
I do use this for image files but you can’t do this for colors set directly in the diffuse swatch. Not trying to threadjack, just sayin…
Is there a file like the one Vlado attached but for MAX?
I’m sorry if there is one in the forum but I don’t know how to search for it.
Edit: Oh, this is Maya only forum. Sorry!
For flat colors there should be vray node called VRayUserColor. It should give you the right gamma accurate colour.
You also have VRayUserScalar for alphas.
VRayUserColor is not for this. It does not give you the right gamma, this is wrong.
If you want the good and optimized color , create a color correct node from VRay texture, set your color, set the gamma to 2.2 (not exactly srgb but…). Now set the same color into your material color (diffuse color, spec color…) from the swatch of your color correct (the swatch is gamma corrected).
Delete the color correct. Done.
Ok back to Universal sampling topic…
I did some hi res tests and so on. All and all I can say that I’m quite impressed. Works very well with scalability from 1k prev to 8k final render.
The problems I run in to are:
1. Utilities. When I try to get diffuse pass and refl_Filter pass. They are very noisy and bad. I have to jank up min AA to 5 to render them smooth. Is that a proper workflow?
2. Is there a way to control quality per err type? Atm we have AA threshold which control all but can I somehow control quality of Lights separately to one on reflections and refractions? I find that most of my tests produce acceptable light source BUT reflection pass is about 50% below where I need it to be.
So for point 2 should I just increase materials refl glossy samples to 16 which would double in effect their quality in final render? Would that work?
No GI tests so far so no comment on that.
Universal settings control the image quality via single parameter - Image Sampler Threshold. Hence controlling the separate properties of the scene is not the main aim of this approach.
In general if something is noisier you have to just decrease Threshold value until you get the noise-levels you are looking after.
However it is still possible to use DMC Sampler and secondary rays subdivisions in order to decrease the amount of noise locally but you have to have in mind that the number of the subdivisions should be much greater (usually x10) comparing non-universal approach due to the fact that by default Vray subdivides secondary rays subdivisions by the max AA subdivs. In Vray 3.x we’ve introduced an option to disable that internal subdivision called “Divide Shading Subdivs” which will force Vray to use the user input subdivs without subdividing them, so you don’t have to bump them to extreme values.
If this option is activated which is by default there will be almost no difference between 8 and 16 glossy subdivisions since they both will be subdivided by 100 - in order to get a desired noise reduction you should use value like 160 or more depending on the rest DMC Sampler settings like Adaptive Amount / Adaptive Threshold / Adaptive Min Samples and Min Shading Rate(in version 3.x).
Another quick approach to increase the quality of the DMC rays globally would be to use Subdivs Multiplier into DMC Sampler Settings.
Vlado, whats about this universal settings :
AA min : 1
AA max : 10
AA Threshold : 0.010
DMC Sampler subdivs mult : 0
Min shading rate : 16
It works great (with BF + LC GI) !
isn’t AA MAX supposed to be 100?[quote=“bigbossfr, username:bigbossfr”]
Vlado, whats about this universal settings :
AA min : 1
AA max : 10
AA Threshold : 0.010
DMC Sampler subdivs mult : 0
Min shading rate : 16
It works great (with BF + LC GI) !
[/quote]
I chose 100 because it is a high enough value to cover most cases. But in practice many people lower that.
Best regards,
Vlado
Humh
Will need to re-check my tests. But I think that when I had noise threshold set to 0.0075 I was getting acceptable Light pass but not reflection pass. Then I went to 0.002 and reflection was better-I think I actually settled on it and light was way over the edge with quality. So I just wondered if I can somehow manually boost per reflection quality.
In any case you are saying that I will need big numbers due to subdivision or not if I were to use the Divide Shading Subdivs. I pretty much have 3 materials in my scene now. So the one causing reflection Issues could be set to 200 to double+ the samples on it and in return it should produce better results without wasting time on light samples via threshold approach.
Will test it. Thanks!