I'm not in a position to make a comparison, but I think a big drawback of cycles is it's very noisy. It's one of those things where you maybe get an OK looking render in 30s, but it takes >5m to get a render that's actually noise free. Which is workable for still frames, but painful for animation.
For hobbysists? Likely not much, other than familiarity with an industry standard renderer (though not as popular as it once was).
For professionals, you get quite a few things but speed optimizations (resolves faster in my experience), render quality (usually fewer fireflies in my experience) and some more advanced transmission models to name just a few.
The only time I had access to Renderman documentation was during my thesis, porting a particle based visualization engine from Objective-C/NeXT Cube into a plain Windows/OpenGL 1.1/MFC.
Reading that, alongside "The Renderman Companion: A Programmer's Guide to Realistic Computer Graphics" was eye opening how far behind regular PCs were from such graphical workstations.
Naturally that was about 20 something years ago, very interessing to see how much Renderman has moved beyond that, and the evolution of programmable GPGPU as well.
There is a slightly dated (2018) paper [1] on their (offline) renderer. Although many details are far beyond my comprehension, it's still fascinating to see how artistic demands lead the technical development pushing the limit. You'll see why some people are so passionate about computer graphics.
"Researchers, individual artists, and students creating non-commercial projects will have access to Free Non-Commercial RenderMan the same day as the commercial release."
This is primarily so people can learn the renderer in their free time and for educational use. It's akin to what used to be called the Personal Learning Editions in the industry.
I too would like it to be changed to the fetaure link. AFAIK this page was not available when I posted the release PR statement.
Also several of my partners shots are in their product announcement page, which is nice to see. (To be clear I don't work for Pixar and they aren't in the rendering team so this post wasn't a plug)
As someone who worked with RenderMan at an animation studio 20 years ago and hasn't really looked at it since, the first thing that struck me was how much the price has come down.
Renderman was alone in its category. REYES approach brings things other renderers couldn't render; fast motion blur, per pixel displacement, aka microsurface, etc. No other renderer provide such fast implementation of this.
Renderman was hard to learn but the promise was here: It was fast. With time the REYES approach starts to show its age (need a lot of preparation work) bad raytrace performances due to “combining” REYES and ray tracing in the same engine, etc. forcing it to fall back in the same competition line (and price) than it's competitors. Full feature animation movie is not the only market.
For example Vray for Maya (used on Tron Legacy) was a good combination between ease of use inside Maya and the technical director features (features that need some dedication and knowledge, like python pre execution directly on render scene) make it a joy to use.
>RenderMan XPU™
Pixar’s solution for CPU + GPU rendering uses heterogeneous computing resources to accelerate production path-tracing. Starting from a system architecture that can leverage the latest developments in many-core CPU and GPU hardware, separately or in combination, RenderMan XPU™ is being built to handle the scale and complexity of Pixar’s feature animation projects. Its first use in the studio has been focused on detailed look-development tasks, with more roles and capabilities to come.
>RenderMan XPU is only available to commercial customers of RenderMan.
>For GPU acceleration, RenderMan XPU™ supports NVIDIA graphics cards from the Quadro, Tesla or Data Center GPU ranges, with the Maxwell architecture or later.
Go XPU is Nvidia only. Not really surprised as I assume no one uses AMD in those production environment. But I wonder how hard would it be to port to something like Metal or Apple Silicon.
I can’t seem to find it but I believe there was a Otoy forums post explaining how Octane supports AMD (and Apple GPUs) via Metal. So Octane runs under MacOS with an AMD GPU but not Windows on that same Mac. They said that they are waiting for AMD’s Vulkan support to improve.
Edit 2: Also interesting is this part:
“I must thank the great folks at Apple who helped us with the hardest parts, some of whom we have been working with for the better part of a decade on this. There is basically no way Octane has any future on Apple's hardware going forward without us doing this work, and we needed their help to make it happen this year.”
So it sounds like if Apple really wants you on their platform, they will provide significant help. Given the close relationship between Pixar and Apple, I wouldn’t be surprised to see Pixar support Metal in some capacity.
Being nVidia specific suggests the use of CUDA and possibly direct PTX code generation, which pretty much means a total rewrite if you want to make it run on underpowered mobile gpu or one that doesn't support CUDA.
23 comments
[ 4.2 ms ] story [ 63.2 ms ] threadFor professionals, you get quite a few things but speed optimizations (resolves faster in my experience), render quality (usually fewer fireflies in my experience) and some more advanced transmission models to name just a few.
Reading that, alongside "The Renderman Companion: A Programmer's Guide to Realistic Computer Graphics" was eye opening how far behind regular PCs were from such graphical workstations.
Naturally that was about 20 something years ago, very interessing to see how much Renderman has moved beyond that, and the evolution of programmable GPGPU as well.
https://www.amazon.com/-/en/Steve-Upstill/dp/0201508680
Looking forward to the paper on this XPU thing.
[1] https://graphics.pixar.com/library/RendermanTog2018/paper.pd...
https://renderman.pixar.com/whats-new
Also several of my partners shots are in their product announcement page, which is nice to see. (To be clear I don't work for Pixar and they aren't in the rendering team so this post wasn't a plug)
Renderman was hard to learn but the promise was here: It was fast. With time the REYES approach starts to show its age (need a lot of preparation work) bad raytrace performances due to “combining” REYES and ray tracing in the same engine, etc. forcing it to fall back in the same competition line (and price) than it's competitors. Full feature animation movie is not the only market.
For example Vray for Maya (used on Tron Legacy) was a good combination between ease of use inside Maya and the technical director features (features that need some dedication and knowledge, like python pre execution directly on render scene) make it a joy to use.
>RenderMan XPU is only available to commercial customers of RenderMan.
>For GPU acceleration, RenderMan XPU™ supports NVIDIA graphics cards from the Quadro, Tesla or Data Center GPU ranges, with the Maxwell architecture or later.
Go XPU is Nvidia only. Not really surprised as I assume no one uses AMD in those production environment. But I wonder how hard would it be to port to something like Metal or Apple Silicon.
https://home.otoy.com/octane-x-pr1/
Edit: post is here: https://render.otoy.com/forum/viewtopic.php?f=7&t=75411
Edit 2: Also interesting is this part: “I must thank the great folks at Apple who helped us with the hardest parts, some of whom we have been working with for the better part of a decade on this. There is basically no way Octane has any future on Apple's hardware going forward without us doing this work, and we needed their help to make it happen this year.”
So it sounds like if Apple really wants you on their platform, they will provide significant help. Given the close relationship between Pixar and Apple, I wouldn’t be surprised to see Pixar support Metal in some capacity.