lacilaci wrote: ↑Fri Jan 11, 2019 5:45 am
I know corona devs are working on a way to render caustics efficiently. I think I would be more interested into their solution or similar rather than pushing too hard for bidir rendering which seems to be always too slow for most real world production uses.
The upcoming GI cache renders caustics for free even with Path tracing. It can render SDS paths too, something even BiDir can not render (it requires vertex merging with BiDirVM).
Oh boy... that's what I'm talking about! RIP bidir Won't it suffer from similar problems like DLCS?(Splotches)
Dade wrote: ↑Fri Jan 11, 2019 10:23 am
The upcoming GI cache renders caustics for free even with Path tracing. It can render SDS paths too, something even BiDir can not render (it requires vertex merging with BiDirVM).
lacilaci wrote: ↑Fri Jan 11, 2019 11:40 am
Oh boy... that's what I'm talking about! RIP bidir Won't it suffer from similar problems like DLCS?(Splotches)
It is Vray-like photon mapping so "splotches" can happen but it is usually the result of not having traced enough photons, etc.
Support LuxCoreRender project with salts and bounties
lacilaci wrote: ↑Fri Jan 11, 2019 11:40 am
Oh boy... that's what I'm talking about! RIP bidir Won't it suffer from similar problems like DLCS?(Splotches)
It is Vray-like photon mapping so "splotches" can happen but it is usually the result of not having traced enough photons, etc.
Dade wrote: ↑Fri Jan 11, 2019 12:47 pm
It is Vray-like photon mapping
Very interesting!
I thought GI cache was storing information about the brightness on surfaces during the rendering when samples happen to hit there.
But do you mean it's more like real photon mapping where photons are actively shot from the light sources in order to establish a map with the bright areas first and so give the path tracing a head start where to find them instead of waiting passively till they emerge during the render?
(Maybe there is a paper somewhere with explanations?)
Dade wrote: ↑Fri Jan 11, 2019 12:47 pm
It is Vray-like photon mapping
Very interesting!
I thought GI cache was storing information about the brightness on surfaces during the rendering when samples happen to hit there.
But do you mean it's more like real photon mapping where photons are actively shot from the light sources in order to establish a map with the bright areas first and so give the path tracing a head start where to find them instead of waiting passively till they emerge during the render?
(Maybe there is a paper somewhere with explanations?)
Well I hope it's not too much like photon mapping in mental ray with all the light leaks and loss of small objects in shadows... also, memory...