I have a problem with the DMC. I am trying to render a scene in BF + LC. A sort of universal settings.
I have noticed that when I use min 1 max100 in general the image is darker or anyway notably different then when I use min 1 max 4. I will post an image when I can.
The difference in time is not big but I assume the 1-100 should be the correct one.
I would like to understand why do I get such difference in lighting.
Anyone can help?
I have already read the doc from Tony Bratincevic and altought it helps a lot I am still a little confused about it.
It looks to me as if you unhided some lights inside the draw/changed shadow on matterials with glass or made the glass self illuminating. There is something more going on than sampler.
Yup but as you can see in the draw inside there is shadow or reflection changing or something. So maybe between tests you changed material or something. Or maybe u disabled ambient occlusion in GI.
Not really.
It is exactly the same scene. I have just reopened the scene. Removed everything and render it over a new plane and I get the same result.
1-100 much darker than 1-4
The Difference you see in the draw and shelf is given by the sampling.
The Vraylightmaterial is mapped with a vraydirt (inverted) and its result is much more visible with 1-100.
When set to 1-4 it’s blowing out.
Hmm - it kind of makes sense that it’s darker at 1/100 - it might be a more correct light solution. 1/4 might not have enough samples to be able to deal with the blowout on your light material.
Had similar situation some time ago, I had subpixelmapping==on and higher AA would darken the image, while low AA gave me similar brightness as when subpixelmapping was off.
i think with subpixel mapping off is correct and/or the one with lower AA max,if subpixel is on and you have that high AA,the (unreal) effect of subpixel mapping will be multiplied more than with only 4x. so either turn off subpixel mapping or reduce AA max, (i have found AA around 32x is a good balance, fast and not so much negative effect if subpixelmapping in on)
sorry for the late reply, I have been busy with some crazy deadline.
I’ll try to send the file as things slow down a bit.
If I don’t go wrong I have both the clamp and subpixel mapping on
Another problem I have noticed with the dmc is that sometimes raising the max subdivs increase the noise instead of clearing it.
I can’t understand why this happen unconsistently.
I remember the thread from bertrand benoit and the article from tony bratincevic but to behonest I got lost at a certain point.
Is there anyone who has a simple answer to how exactly behave the dmc sampler on the latest vray versions and how to tune it?
Thank you very much for your scene.
We would like to inform you that this is expected behavior when Sub-Pixel Mapping is ON and Brute Force is used.
The generated Indirect Light from Brute Force when Sub-Pixel mapping is ON is always with less intensity than when Sub-Pixel is OFF - and with the increasing of the Image Sampler subdivisions the illumination becomes even less. This is very obvious if Linear Work Flow setup is used and it is not locked to Adaptive DMC Sampler, the same result will be generated from Fixed or Adaptive Subdivision Samplers.
You can avoid such differences in your renders if you either turn off Sub-Pixel Mapping, or by using Light Cache or Irradiance Map engines instead of Brute Force.
This is also expected and it is because when you tell Vray to shoot more Camera Rays , it starts to fire less DMC rays (material / light / dof / motion blur etc).
And it is developed to work this way because otherwise when Image Sampler subdivisions are increased the final render-time won’t be acceptable at all.
Since with every Camera Ray at least one DMC ray is fired - when Vray shoots a lot of Camera Rays it doesn’t really need to fire all DMC rays specified in material/lights and etc subdivisions.
There are two ways to fix this: we have to either decrease Clr.Threshold value in DMC sampler , or increase material/lights subdivisions.
In most cases both parameters should be tweaked.
