Sampling problem (blotches)

[Tim Saunders]

I have never run into this and I can't seem to get rid of it. See the blotches in the attached cropped image. I am using Irradience map with the high preset and 50hsv subdivs and 20 interp samples. then qmc as secondary with defalt settings. I can see the areas that will blotch when it runs the calc phases, but I don't know whay I am getting htese dark areas. Any ideas?

[kippu]

sorrry ...tim, never seen these ones before...just 2 things come to my mind...shadow settings , secondary bias... i am just shooting off top of my head though

[MrCAD]

using a high preset should not mess up like that...are you using shadow maps ? if yes then that could an issue....in that case need to increase the mapsize to a much higher value. What kinda lights and its values have been used ? ...anything small could be possible to generate this error...and which version of vray ?

 

Meher

http://www.mr-cad.com

[MrCAD]

I did a few tests...and you can see the images

the filnames are according to the Hsph. Subdiv-Interp. samples-high.jpg

and I think you can simply higher the Hsph. Subdiv = 60-80 and Interp. samples = 25-30 on your current scene and it should loose some blotches

cheers

Meher

http://www.mr-cad.com

[Tim Saunders]

Thanks for the replies, and taking the time to experiment. As far as the light types, I am using standard direct lights with vray shadows. I kept the defalt settings all except for the area shadows. I have some size going on with the uvw, but everything worked fine with the other camera angles. I did upgrade from viz 2005 to 2007, and in doing that I had to tak a step back in my version of vray from 1.47.03 to 1.46.15. I have rendered scenes like this in the past with way lower settings like 30 hsv subdivs and all was well. I guess I have to crank it up and see how it goes. Thanks again.

[brianzajac]

Tim,

I'm confused with why a certain camera would have the blotches and not the others if that's what you meant. Nonetheless, it shouldn't be too hard to nail down the problem deductively. I would start by rendering to region to make the test renders go really fast.

 

Take the materials out of the equation and set the material overide and use a simple gray vray material.

 

Take QMC out of the equation and don't use. In fact, test it without any 2nd bounce.

 

Try sec. rays bias at 0.001 (global switches)

 

Render to G Buffer output (GI, raw GI, ray lighting) to see what you can deduce.

 

Remove the light and render with default to take the light out of the equation.