About noise in general

[joconnell]

If you've no red, it means your AA max value is too high for your scene. In theory this shouldn't be a problem, but it also could mean that your high AA setting is also reducing your material and light samples and causing noise and slowdowns. I like having a nice range in the sample rate as it makes the rest of the scene more predictable to me.

[Guest]

If you've no red, it means your AA max value is too high for your scene. In theory this shouldn't be a problem, but it also could mean that your high AA setting is also reducing your material and light samples and causing noise and slowdowns. I like having a nice range in the sample rate as it makes the rest of the scene more predictable to me.

So this could be the fundamental flaw of the universal settings? by over sampling you end up under sampling in some areas but in the majority of cases it is acceptable.

[zeronove]

I agree with this conlusion as well; otherwise it's not so unusual that an high amount of AA samples is required to solve object edges or highlights. Maybe I'm wrong but sometimes I've found that raising clr threshold instead of lowering is the way to go to limit AA sampling only to these issues.

[joconnell]

So this could be the fundamental flaw of the universal settings? by over sampling you end up under sampling in some areas but in the majority of cases it is acceptable.

Yeah I think so. In theory AA can clean up everything but as vlado said right from the off, universal was a method designed to be simple and not necessarily the best in terms of render time. Whatever about the sample rate being red or blue, the issue is that if you've got two parts of an equation making the final render, you're screwed if you don't know what at least one of those is. With a good sample rate element you can at least make a good guess whether the AA is high, middle or low for certain pixels. If you know that, you can judge a reflection or a lights graininess now knowing what AA might be adding or taking away from it's sampling.

[joconnell]

I agree with this conlusion as well; otherwise it's not so unusual that an high amount of AA samples is required to solve object edges or highlights. Maybe I'm wrong but sometimes I've found that raising clr threshold instead of lowering is the way to go to limit AA sampling only to these issues.

Correct - AA is for textures, highlights and geometric edges and nothing else. You could have a really simple scene from a geometric point of view and a high colour threshold of 0.03 or 0.05 might be totally fine to get the properly quality in these areas. If you unlink your AA threshold from the DMC one, you can set your AA to what's needed for your geometry and textures, and let the main dmc control the quality of the material and shading sampling.

[windowlicker]

..universal was a method designed to be simple and not necessarily the best in terms of render time

I don't know if I'm missing something when it comes to the universal setting, but whenever I've tried it I've trippled my render times to get more or less the same quality as I get with IM/LC. Every time I've tried it, I have failed to see the benefit of it other than the simplicity in having one slider to adjust the quality of the render (which is great for testing purposes, but then again we have RT). I'd love to have it as a working option in some projects, but so far I simply cannot produce a decent looking render in a reasonably short amount of time.

[joconnell]

True, but it seems to have worked fine back in vray 1.5. 2 introduced a lot of changes which have made the method less effective. I personally only briefly used the 1 / 100 approach but as vlado says it was never intended as the optimal settings for ever scene.

[jasonhuang1115]

Right. I didn't want to write anything until I was total certain of whether my approach works or not, but I reckon the sample rate is the key to it. AA messes with everything else, so you've got to find what AA you need first, and that's via the sample rate element and getting a nice balance with only red pixels on the edges, and fine texture detail. Once you've got your AA set, then start looking at your light and material samples. If you're constantly changing both, you're chasing your tail. 32 samples could be okay on materials but some will need more, some will need less. What could be another good guide is the mtlglossiness element, so you can look at your render and the grey values tie in with how glossy a material is, the darker, the more blurry. What you can do is swap between your reflection pass and the MTL glossiness pass, and see what materials are noisy and clean. If all of your 0.6 glossy materials are clean, then 32 is likely okay, if not, up the samples.

Great tip! Is the MTL glossines element available in Maya? Or it is a Max only feature. I didn't see this render element in Vray 2.30.01.

[pengo]

Vlado I'm sending a basic scene (to support@chaosgroup.com) I rendered with a problematic Irrad map GI splotchiness and with that "micronoise" that only lowering the clr thresholdfrom 0,007 (in the first picture) to something like 0,003 (in the second picture) or lower (0,002) is reduced, instead lowering noise threshold value has no effects or visible effects on this grainy effect.
If u have time I'd like to have a response on wich can be the cause of those problems. Sure in my workflow is the problem and with it I suppose some way to create materials and so on.
The scene has only one light a vraydomelight with an exr map in it
Settings:
IM: preset high hsph subds 75 interp sampes 35
Lc: 3500 subds screen size 0,0070 storedirect light prefilter (switched) 50 - use light cache for glossy rays (switched) - retrace threshold 1 (switched) - interp samples 5

Dmc sampler settings: adaptive amount 0,65 - Noise threshold: 0,01 - Min samples: 8 - global subds multiplier: 1
Image sampler settings: 1/100 clr threshold:0,007 adaptive dmc type no filter used
nothing in the environment slot as map and no other than 64 subds to the vraydomelight (this light is not stored to irrad map) to get its shadows clean - Material subds 16
Thanks in advance

[Paul Oblomov]

Just turn on some raw channels, and you will see the problem. And may now decide where to start - from rising shadow subdivs or glossy or both. Or maybe downscale/blur .exr map or something like that

[joconnell]

Could you render a crop with the following elements:

SampleRate
RawLighting
Reflection

On paul's note about the raw channels, I reckon that rawreflection is very misleading from the point of optimization / judging noise since if you've got a weak reflection component on a material, vray boosts those pixels massively to make the raw channel - it'll always look bad.

What I reckon this is, is that the 100 aa samples are dividing down your material samples hugely, so it's not giving them enough material samples to sample the bright parts of your light source, or it could be the aa samples reducing the light samples hugely causing the same issue. I reckon it's a combination of noise in your light and your materials.

Actually since this is such a simple scene, perhaps it'd be worth uploading so the forum can have a look? There can't be that many materials so it's an easy one to fiddle with!

[pengo]

Hi paul thanks for the reply I have the vraysamplerate on and everything is blu dark blu about globall illumination I get those artifacts onto white walls at the upper left corner. I suppose is the vraydomelight option full dome active, I'll try to deactivate/unchek it and I'll see. Could it be the colorcorrection map? Because I used it under 0,4545 to put the map in linear mode. Should it be too much and as in other case some noise or splotches was given by the active curve option in the bitmap options rollout with this map in color correct slot. Thanks Paul and joconnell for the interest/support.

[joconnell]

I'd say your AA is choking all the other samplers - if you're using 1/100, try going to 1/50 and then 1/25 for starters.

[pengo]

I tryed 1/24 with 200 subds for materials and for shadows (24 AA max rate X 8=192) but in vray sampler I got greenish and slowest was the render. So I tested the pseudo-universal method (because the pure universal method use qmc in primary bounces not Irrad map as I did) I re-tested it (AA 1/24 with above mentioned subds 200 for lights and hsph and material subds) with clr thresh 0,01 to start to tests again. The previous test with those parameters was using another map. I have to tell the GI is more clean but the grainy effect a lot stronger (in Vrayglobalillumination no grain in the rgb color channel a lot more than before) is it suggested to use an area filter to smooth this problem? Have I to crank up the light subds that shoud correspond to something like AA max rate x 16 or more? The times are reasonable with clr threshold as default but that grain ggrrrrr!!
I'm testing with vraylightselect channel and the vraydomelight is really grainy, instead as told above the vrayglobalillumination is clear without GI noise too. No way cranking up the light subs to 400.No difference in grain a little more light in the solution but about the noise no effect. So I think the only way is to work by noise threshold or clr threshold values. Wich is better to use in your opinion?
Thanks Joconnell