Second bounce in interiors: Lightcache vs. Bruteforce

[duke2]

I don't know what the rendertime differences are, but i've had far greater success so far using Bruteforce for secondary rays vs. Lightcache. With LC I have to bump the samples up to 2000 to get anything near useable, but with BF I only need around 16 samples and 8 bounces to get a clean result. I don't know how BF behaves in other situations (for example, LC emits light from vraylight materials but BF doesn't) but so far its proving more reliable than LC.



Settings: http://dukecg.net/BFvsLC_settings.jpg

[vlado]

The difference is way more obvious in areas dominated by indirect light, where the brute force method has difficulty finding the light sources.

2000 subdivs for the light cache is not that much, considering that they are spread over the entire image. Indeed in some cases I've had to go up to 3000 or 4000.

Best regards,
Vlado

[m_hinks]

Yeah, I normally go up around 3500 for a 5K image.

I would imaging while the quality of the BF might be better, the render time with 8 bounces will be BIG! LC is really good for interiors where the infinate bounce will create a much more correct light, and quickly.

I tend to use BF for exteriors as the bounce count can be kept low, and with an exr. render, you can simply re-expose the interiors and mask them in to fake the extra light.