h13 difference between raytrace and pbr

[MENOZ]

hello, I'm studying h13, but i can't really find what the diferences are now between using raytracing and pbr.

when should I use pbr and when raytracing? they seems really similar...

thank you

[bandini]

Generally, I am using PBR for everything except quick test renders. If you want light bounces and realistic spec/area light reflections, it is the way to go. I guess it depends on the level of realism you want in your renders vs. the time it takes to render without a lot of grain. The trend seems to be toward PBR all the way, with most rendering enhancements from SideFX going into PBR.

That said, if I am just rendering out smoke or dense flame-style particles, I will just use raytrace.

[symek]

I think MENOZ is asking because in H13 there isn't much difference between ray-trace and PBR. All four engines evaluate bsdf(), so that all support multiply bounces, and since raytrace uses primary rays the same as PBR, difference evaporated (or became more esoteric).

[MENOZ]

yes symek, tank you for the better explanation.

I don't really know what is changed and how it handles the rays...I don't know where I could find an explanation of those things, can you help me?

[dennis.albus]

I just tested this with a really simple test case and as far as I can tell there is absolutely no difference between raytrace and pbr mode. The images are exactly the same as well as the rendertimes.

Or is there something we are missing here?

-dennis

[fathom]

yeah, i have noticed that the same limitations of pbr have seemingly crept into rt mode now. i find this to be VERY disturbing since pbr's limitations are pretty significant. in particular, i've noticed the atmosphere shading no longer occurs on secondary rays (well, it occurs but adjusting Cf in an atmo shader has no meaning for pbr and so now it doesn't for rt either). it also seems the attentuation calculation on the built-in shader model's refraction no longer does anything. i'm guessing this is due to the rt shader giving up on actual ray tracing and going with the bsdf calc instead.

presumably the internals of the ray trace engine are still active and this is just the way they coded the latest version of the shader model vop.

[dennis.albus]

yeah, i have noticed that the same limitations of pbr have seemingly crept into rt mode now. i find this to be VERY disturbing since pbr's limitations are pretty significant. in particular, i've noticed the atmosphere shading no longer occurs on secondary rays (well, it occurs but adjusting Cf in an atmo shader has no meaning for pbr and so now it doesn't for rt either). it also seems the attentuation calculation on the built-in shader model's refraction no longer does anything. i'm guessing this is due to the rt shader giving up on actual ray tracing and going with the bsdf calc instead.

presumably the internals of the ray trace engine are still active and this is just the way they coded the latest version of the shader model vop.

With RT atmosphere shading should still be possible as the whole shader is still using Cf as an output instead of F. The BSDFs are calculated inside the shader and return the color to the shader for you to further edit and use it. So it's all good. It seems that PBR inside of the raytrace engine is actually the way to go as it gives us the power of PBR without some of its limitations.

Also it is still possible to build your own shader using illuminance loops and all the legacy workflows that you know. If you want a very specific shading effect and are not into writing your own bsdf this is still a very good and valid way to go.

Edited November 11, 2013 by dennis.weil

[fathom]

With RT atmosphere shading should still be possible as the whole shader is still using Cf as an output instead of F. The BSDFs are calculated inside the shader and return the color to the shader for you to further edit and use it. So it's all good. It seems that PBR inside of the raytrace engine is actually the way to go as it gives us the power of PBR without some of its limitations.

Also it is still possible to build your own shader using illuminance loops and all the legacy workflows that you know. If you want a very specific shading effect and are not into writing your own bsdf this is still a very good and valid way to go.

yeah, i agree it should still be possible. but in the build of 13 i was using, it didn't work (using the standard surface model). i think they're using a bsdf sample to calculate the indirect lighting rather than calling trace, but i haven't investigated too deeply yet. my concern tho is that the surface model is generally pretty good and i prefer to use it when i can.