Adaptive DMC Image Sampler: Sampling Criteria (Vray 1.5)

[Lehm_2001]

Hey All. I haven't worked V-Ray in awhile and I thought it was time to refresh my skills. I'm using V-Ray 1.5sp6 as that's what I have available. I've been experimenting with the DMC ImageSampler trying to get faster results when I noticed something odd. At the corner of a wall where there was light on one side and not the other (high contrast edge) it had very few samples and looked jagged. However a nearby edge of the wall that had much lower contrast had many more samples. I couldn't figure out what was going on. That was until I looked at just the diffuse channel. There was very little contrast on the first edge once your removed the lighting.

So my question is, does the sampler ignore lighting information when deciding to do more samples? If that is not the case what does it use to determine? All I could find was it uses the intensity of the pixel.



I've deliberately raised the DMC threshold to 0.1 to illustrate the issue.

[Lehm_2001]

Nevermind determined why that corner wasn't sampling correctly. It looks much higher contrast when the buffer is set to display sRGB like I have it. If instead I set the color mapping gamma to 2.2 and disable sRGB in the buffer it results it much more even sampling.