For the past two versions of VRay (since the QMC threshold was raised by a factor of 10) word around the office has changed and arguing has become commonplace as to what the a “bad” DMC noise threshold would be an what a “good” one would be.
Some around the office swear by a noise threshold of 0.1 being a good quality setting and something around 0.17 being a draft setting while others (like myself) hold to 0.01 being a good quality setting and .04 being more of a draft setting. (This hypothesis is based on the default values being so low).
Clearly this is a large difference, and honestly quality tests are pretty inconclusive…Will someone please set us straight!
Depends on the rest of your scene and what the settings are, its a cog in the machine rather than a controller of the whole thing.
I tend to leave it around .01 for renders, or if its causing problems lower it to .005
I understand that if my AA is set to Adaptive Subdivision, and my GI set to IMAP/LC (which it usually is), the noise threshold and adaptive amount can be quite a bit higher than the default values.
It’s true that you can use higher values when using adaptive subdivision AA, you notice the effect of the noise value a lot more when you use adaptive qmc AA. But the noise treshold also affects your ir map quality, keep that in mind too!
I usually use noise of 0.2 or higher for test renders, and 0.01 or 0.02 for finals (with adaptive qmc aa)
I use different settings for IRR/LC pass and final pass.
0.85/0.01 for IRR and 1.0/0.1 for final.
Since IRR map is already calculated, there’s no need to have such low noise threshold, and rendetimes and MUCH shorter. Of course there can be problems with AA, but it is solvable with AA settings.
This rule is especially usefull for arch-animations, for stills I tend to use lower settings for final pass.
Lower as in lower values = higher quality
But I never go below 0.01 - since I’m a time freak, and if I have to wait
for a render longer than one hour I’m sick…
