Hi, no not initially, and we probably won't be able to render all those passes before GPU'S are available with a lot of memory to keep all those in memory while rendering. I think in version 2.0 we will work on these areas, by then you will be able to get , although expensive, fermi GPUs with lots more memory.ekanS wrote:I have a few questions:
- with the rib import (in 1.03 beta?) will octane understand TCL scripts? (Like RIB-BOX in new renderman for maya 3)
- will octane render (and save) multiple passes - like caustic, shadow, reflection, refraction - in separate files, or multi layered image file, like EXR or PSD?
sorry if my english is too bad
ekanS
If you would have to render these all separately it will take you a lot of time,redmotion wrote:radiance, I'm curious that if its possible to tweak individual settings on shaders such as the specular strength/textures/for example, why it would not be easier just to separate this information into channels (and render each channel to a separate file). So like the clay rendering button at the top - but say, for specular - couldn't you just have everything switch off except the specular component? I can kind of do this already by just setting diffuse to black.
I'l have to ask for a second mortage on my house just to be able to pay for these.... lol I you put 4 wheels on top I could drive to work on it, and then maybe sell my car to pay at least one of those....radiance wrote:but i'm hoping quadro GPUs with fermi chips and 6GB ram etc will be of help here in the near future.
every pass needs raytracing in a physically based raytracer, we don't implement reyes rendering...ekanS wrote:I'm think the pass requires only remains two times more memory, but lot of rendering time. I think it becouse I found this expression in renderman DOCS:
result = SpecularDirect - SpecularDirectShadow + SpecularIndirect +
SpecularEnvironment + Ambient +
DiffuseDirect + Translucence - DiffuseDirectShadow +
DiffuseIndirect + DiffuseEnvironment + Backscattering +
Subsurface + Rim + Refraction + Incandescence
As I see, you need one piece of memory for the current pass, and a equal piece for the step-by-step incremented result.
And I'm think not every pass needs raytracing, but maybe I'm wrong.
ekanS
