How light Cache density affects Brute Force quality

[3DMK]

Hi All,

After some recent issues & testing I now understand another part of how Vray works..[only another 5000 parts to go]

Below are examples of how the density [total number of samples per image] affects the final quality of the image when using BF as first bounce & LC as secondary bounce.
As the BF & LC gi values are added together to get the final gi value. Therefor It makes sense that more dense/detailed LC settings will give a more detailed result.

First image is using LC with sample size of .06 [screen]
First image is using LC with sample size of .01 [screen]

See the results for yourselves below [Download them for a closer inspection if you like]








Hope this helps!!

[yyk]

Thanks, I find this stuff interesting.. light cache is so easy to overlook when preparing to render in high quality!

[zeronove]

thanks for sharing, even for me some LC settings are a bit unclear

For example how to manage density increasing the render size?

Using IM we can go 1 step down for every double resolution, but what about LC?

4 times for every double resolution? it seems to me a bit too much, what about you?

[vlado]

You really shouldn't change the light cache sample size when you change the resolution, it works right away.

Best regards,
Vlado

[^Lele^]

The Lc sample size is expressed as percentage of MPx of your final image resolution, so as Vlado says, there's no need to change that following a resolution change.
You can use the LC Calculator in my script pack to see/change the "Shading Rate" of your LC.
Say you have an image of 1000 x 1000 pixels (1 MPx), and the subdivs in your LC are set to 1000 (which would result in 1 million samples taken).
That's a shading rate of 1, as in other words you have 1 LC sample per pixel.
The calculator does the math for you, and you can vary the shading rate and preview the LC settings before applying them.

[zeronove]

fine,

that the way I actually manage LC, without any alteration increasing resolution

it was only a doubt coming from some reading on web

thanks

[^Lele^]

Ale, i am NOT sure the idea of Shading rate is actually correct.
All i know is that it's something that by being kept consinstent gives me consistent results.
As for sample size, since the number in the VRay LC panel is expressed as normalised (ie. it goes from 0 to 1), leaving it at 0.02 means you have a sample size which is 2% of your total sqrt(MPx).
So changing that isn't needed when changing resolution (ie. it will always stay proportional to the image size, leading to identical GI quality for different image sizes, given an identical number of samples per pixel).

I am replying to the post i read in mail, which you likely edited in the while

Edit:http://www.chaosgroup.com/forums/vbu...25&postcount=4