mental ray at Siggraph 2012

[Remydrh]

could guys get us updated about any mental ray / iray presentation if any ?

Hope to beginning of next week unless Bart posts here sooner.

[bart]

For a video on Phil Miller's Advanced Rendering and GPU Raytracing talk, see http://nvidia.fullviewmedia.com/sigg...and/SS107.html

In this talk, Phil talks about mental ray 3.11 soon in beta, and iray 3. He also shows a short video by one of our devs about our Light Path Expressions available in iray 3.

The 3.11 feature he shows is the Ambient Occlusion pass that can be calculated in parallel on the GPU while the CPU is rendering the rest of the scene. This is enabled by a string option.

We are developing some layered shaders for mental ray, which will help lead us toward using layered bsdfs. The layered shaders will come with 3.11 but can be tested in 3.10. Of significance in these shaders, is an sss component layer that does not require a light map. They also have a builtin user framebuffer pass mechanism, encouraging proper additive work flow.

In 3.11, along with Light Importance Sampling (IS) and Multiple Importance Sampling string options, we have supplied a new shader based method for creating object lighting. This should make it easier for immediate use in integrations.

We also have enabled multiple motion transforms for more complex curvy motion paths, but that one might require a little more work with the integrators. Pipelines that already process the mi files, could add it, of course.

We will also make available iray 3 with Kepler support for mental ray, but final packaging decisions of course depend on the integrations.

[Kzin]

great news. light is and layered bsdf's will speed up the rendering alot.

[RickToxik]

I have read earlier that Nvidia "pulled out" mental ray from maya to make it work as a normal -external- plugin (in maya 2013).

1. Will mental ray 3.11 show the IRay options and others (progressive rendering, unified sampling, etc) natively in maya?
2. Will it be possible to upgrade to mental ray 3.11 without upgrading maya?

[1armedScissor]

...We are developing some layered shaders for mental ray, which will help lead us toward using layered bsdfs. The layered shaders will come with 3.11 but can be tested in 3.10. Of significance in these shaders, is an sss component layer that does not require a light map. They also have a builtin user framebuffer pass mechanism, encouraging proper additive work flow...

Will the combined output of a layered shader(s) be energy conserving (obviously not expected if one of the input shaders isn't energy conserving, but if they are)?

[JJHop]

The 3.11 feature he shows is the Ambient Occlusion pass that can be calculated in parallel on the GPU while the CPU is rendering the rest of the scene. This is enabled by a string option.

We are developing some layered shaders for mental ray, which will help lead us toward using layered bsdfs. The layered shaders will come with 3.11 but can be tested in 3.10. Of significance in these shaders, is an sss component layer that does not require a light map. They also have a builtin user framebuffer pass mechanism, encouraging proper additive work flow.

In 3.11, along with Light Importance Sampling (IS) and Multiple Importance Sampling string options, we have supplied a new shader based method for creating object lighting. This should make it easier for immediate use in integrations.

We also have enabled multiple motion transforms for more complex curvy motion paths, but that one might require a little more work with the integrators. Pipelines that already process the mi files, could add it, of course.

We will also make available iray 3 with Kepler support for mental ray, but final packaging decisions of course depend on the integrations.

Hi, few questions here:

Both Light importance sampling and multiple importance sampling has anything to do with how Mental ray will calculate GI?
I mean, will those new shaders/lights help users to have a better illumination or they are just shading techniques, meaning we will still have to use Final Gathering?

Another question is how comes you decided to switch the Ambient Occlusion calculation into GPU and not something else, per say motion blur/dof/or even final gather photon emission, heavy things to calculate. Displacement.
I feel GPU calculation is pretty wasted on Ambient Occlusion only, despite we all hope we will have a better GI solution someday with mental ray so we should just stop relying into AO all together.

Speaking of global illumination i have a simple question, after all the complains and problem and see how MR at the moment sadly is far behind other renderers (especially on bruteforce approach on GI solutions), do you guys took into consideration to fix this at all? Because its an issue brought to attention so many times and it is a bit frustrating to see you workin in totally different sections of the render engine (shading/lighting).

Got another question here as well. Trying to use the bsdf that comes with Maya and 3ds max (tried even Xsi) with the irradiance particles + importons + IBL, and basically they dont take into account the irradiance particles bounces. Surfaces are rendered almost black, only the IBL is taken into account so its like the IP+Importons are switched off.
Is this a normal behaviour? I thought the bsdf shaders would benefit from ibl+ip+unified sampling but apparently there is a bug?

Thank you.

JJ

[Remydrh]

I can imagine they are, from Bart's comment about a "toward" a BSDF model.

While non-photorealistic is always possible, photo-realism is still the most often asked-for look. This means easy physical correctness by default. You can also see his comment in the mental mill section:

Depends on your level of coding interest. We have evolved our solutions to reduce coding needs for common use. Are you interested in mixing layers of pre-defined materials? Have you seen the iray layered bsdfs?

[rlevene]

Really good to see this presentation. All looks and sounds very promising. Just hope the integration will do all the features justice.

Thanks,

Richard

[Saturn]

For a video on Phil Miller's Advanced Rendering and GPU Raytracing talk, see http://nvidia.fullviewmedia.com/sigg...and/SS107.html

In this talk, Phil talks about mental ray 3.11 soon in beta, and iray 3. He also shows a short video by one of our devs about our Light Path Expressions available in iray 3.

The 3.11 feature he shows is the Ambient Occlusion pass that can be calculated in parallel on the GPU while the CPU is rendering the rest of the scene. This is enabled by a string option.

We are developing some layered shaders for mental ray, which will help lead us toward using layered bsdfs. The layered shaders will come with 3.11 but can be tested in 3.10. Of significance in these shaders, is an sss component layer that does not require a light map. They also have a builtin user framebuffer pass mechanism, encouraging proper additive work flow.

In 3.11, along with Light Importance Sampling (IS) and Multiple Importance Sampling string options, we have supplied a new shader based method for creating object lighting. This should make it easier for immediate use in integrations.

We also have enabled multiple motion transforms for more complex curvy motion paths, but that one might require a little more work with the integrators. Pipelines that already process the mi files, could add it, of course.

We will also make available iray 3 with Kepler support for mental ray, but final packaging decisions of course depend on the integrations.

thanks Bart.

I was wondering about Light Path expression. It was not clear. I understood that you could separate indirect Spec/ diffuse from direct
but I was wondering if that will allow to you categorize light.
I mean having a framebuffer per light contribution ( including bounce and caustics ) so you could do some relighting in post.

[1armedScissor]

I can imagine they are, from Bart's comment about a "toward" a BSDF model.

While non-photorealistic is always possible, photo-realism is still the most often asked-for look. This means easy physical correctness by default. You can also see his comment in the mental mill section:

I asked the question because assuming each individual shader is a black box unto itself, how can a layered shader possibly know how to mix them together to provide an output that is energy conserving?