Caistocs question
Posted: Sat Feb 18, 2023 10:22 am
Hi everybody. I would to better understand caustics behavior with path tracing.
I made a simple scene with a open box with scatter (random walk) and a cube with displacer, a plane with two simple wall and alone area light, as you can see in the attachment.
The risult is quite good and clean but this render is 6400x3600 with 524288 samples, no denoise and no adaptive sampling...and 32 hours with two 3080ti strix (1250mhz oc on ram).
No adaptive sampling because it makes holes in the caustics area: why?
So many samples to have clean and precise caustics and refelctions/refraction.
This resolution because if i cranke up samples to maximum (1000000) with a lower resolution, caustic are ugly. Why?
PMC has weak power caustics as compared to PT and photon tracing doesn't have enough photon depth yet to calculate deeper caustic and it take to much time to render shading sample as compared to PT.
Maybe BDPT could make better than normal PT? Do you have some news about the implementation of it in octane?
Thanks for your help
I made a simple scene with a open box with scatter (random walk) and a cube with displacer, a plane with two simple wall and alone area light, as you can see in the attachment.
The risult is quite good and clean but this render is 6400x3600 with 524288 samples, no denoise and no adaptive sampling...and 32 hours with two 3080ti strix (1250mhz oc on ram).
No adaptive sampling because it makes holes in the caustics area: why?
So many samples to have clean and precise caustics and refelctions/refraction.
This resolution because if i cranke up samples to maximum (1000000) with a lower resolution, caustic are ugly. Why?
PMC has weak power caustics as compared to PT and photon tracing doesn't have enough photon depth yet to calculate deeper caustic and it take to much time to render shading sample as compared to PT.
Maybe BDPT could make better than normal PT? Do you have some news about the implementation of it in octane?
Thanks for your help