Well that's a surprise. Pixar Renderman, the software used and developed by Pixar to turn computer-represented geometry into wonderful images, is now free for non-commercial use. This is not quite as free as Unreal Engine 4, which does not require royalties for rendered audio/video at all because the content is viewed without the engine, but free for non-commercial is still a big deal considering what Renderman is. While Pixar is known for their movies, they are very much software engineers.
Thumbs up all around.
Image Credit: Pixar via Animation Magazine
Currently the rendering package integrates with Autodesk Maya and KATANA by The Foundry. Pixar has Cinema4D and Houdini listed as “in development”. They also claim to be interested in 3D Studio Max, because of course they are, as well as Modo, Rhino, Lightwave, and Blender.
The above list only considers dedicated plug-ins. Pixar Renderman can also be run as a standalone application that accepts their file format, “RIB”, from a command-line interface. There has been many of these for Blender and other 3D suites over the last basically forever. A plug-in is nicer for artists however, and it is good to see Pixar is not afraid of open-source suites to pair with their proprietary rendering package. Also, "Blenderman" is a hilarious name.
Yay Blender, looking forward
Yay Blender, looking forward to more Blender benchmarks on some of the newer GPU hardware. I wonder if the new Vulkan graphics API/LLVM will improve Blender’s performance, and if there will be newer python builds targeting the SPIR-V and the Vulkan LLVM. It will probably take a while but things are looking up for Open Source, and graphics/GPGPU abilities.
From my experience, while in
From my experience, while in the edit view,the GPU is rarely the bottleneck.
Overall, I am hoping to one day see the quality and ease of use of mentalray, but with full GPU acceleration (including for ray tracing), without stupid limitations such as being only for quadro cards.
But there are some mesh
But there are some mesh editing commands that could use some GPU acceleration, as well a multithreading on the CPU, so being able to take OpenGL/OpenCL and run the mesh editing commands via the SPIR-V LLVM would help. Try doing a Mesh relax on the current Blender 3d for 50,000+ polygons/100,000 or 150,000 vertices and it will take some time, and mesh editing is where a lot of the time is spent, and the mesh editing commands, and workflow needs the most acceleration, with rendering the user does not have to attend the PC/Laptop while the rendering is being done. I spend most of my Blender time editing.
For sure whenever the Blender source code, and python based editing, and underlying Python runtime code can take advantage of Vulkan’s API and the SPIR-V LLVM then maybe things like accelerated Ray tracing, and cycles rendering will not be limited to mostly Nvidia GPUs. But in the future it would be great if dedicated ray tracing hardware was available on the GPU, for sure the PowerVR wizard with the dedicated ray tracing hardware, has me hoping that both AMD, and Nvidia will start to add dedicated ray tracing circuitry to their GPUs, But OpenCL ray tracing would sure help also. Those Quadro cards cost too much, but so do the Xeon workstation SKUs, and even the fastest Xeon SKU with the most processing cores can take some time to render a ray traced scene, especially if the ray sample rates are turned up to the high/max settings. Just think about the savings for graphics workstation hardware, if all that was needed for a graphics workstation was the lowest cost Xeon SKU, with the GPU’s dedicated ray tracing hardware doing the ray interaction calculations on some massively parallel GPU cores.
Even if it will be a while before dedicated Ray Tracing hardware makes an appearance on discrete GPUs, I’m looking at some of those ARMv8 ISA based Server SKUs, the ones with more than 16 cores, that Cavium 48 core ThunderX could probably do some Ray tracing workloads, and hopefully it will be less costly than a Xeon based system, or even a Xeon pi based accelerator card. AMDs future Opteron based server SKUs both ARM ISA based, and x86 based are going to be interesting if AMD can take their K12 custom ARM core and make a future server/workstation SKU with loads of cores for ray tracing/rendering. I’m looking at any workstation systems that may ship with an AMD SkyBridge motherboard, if the system can come with a 16+ core ARM based server SKU, with the option of moving up to an x86 based SKU with the swap of CPU/APU chip at a later time if more processing power is needed.
Both K12, and Zen based AMD server/workstations systems are going to be popular for lower cost professional graphics workstation systems, and with SkyBridge based motherboards users could purchase an ARM based workstation, and upgrade it to an x86 based workstation and not have to purchase a new motherboard.
*cries* their site is not
*cries* their site is not sending the forum activation email. 🙁
I really want to see how fast it can render animated clips compared to mentalray.
use a commercial
use a commercial .com/.net/.org email for better effect. Thye might have issues with free email providers.
In other renderer news, OTOY
In other renderer news, OTOY are planning to make a ‘VR’ version of Octane 2.2 free for commercial and noncommercial user from April – July (as per their GTC presentation), and potentially byond if it’s popular. It’s a superset of Octane 2.2, so you can use it for regular non-lightfield rendering too.
I’m a lighting and rendering
I’m a lighting and rendering artist and use these types of renderers all the time. I’m glad Renderman is now free for non-commercial use, as it helps artists unfamiliar with the software as now they can learn it. I’ve been doing a lot of testing with the new Renderman, and so far, it isn’t disappointing.
When I purchase new hardware for my personal PC builds, this type of rendering is always my first thought. I’d love to see more reviews on this site have a focus towards these types of workstation machines.
razor512, as far as Mental Ray goes, it’s a little long in the tooth compared to renderers like Arnold and Renderman RIS. I’d definitely recommend spending a little time learning one of these renderers and you’ll have a better time. 🙂
Nice! Thanks for your
Nice! Thanks for your thoughts!
Yay, the activation email
Yay, the activation email finally came in. Did a test render on a random low polygon count object. (just testing performance mostly)
From messing around with it, it is extremely easy to do lighting (though I was lazy and just did 2).
I have not gotten the re-render function to work yet (the option doesn’t seem to even show up, thus it is hard to preview the lighting in real time.
Only issue with it is that it takes a long time to do ray tracing, it is just as bad as mentalray when it comes to render times. though since it renders everything at once, and simply ups the resolution with each pass, you do not have the issues from mentalray where you may end up with a scene rendering for like 20 minutes only to find out that something is messed up.
With this render, you get a decent resolution view within 2-3 seconds, thus you don’t waste time rendering a mistake at full resolution and quality.
Here’s a link to my initial
Here’s a link to my initial impressions.
You can find comparisons of VRay and Arnold on there as well.
Awesome render 🙂 both the
Awesome render 🙂 both the head render, and the VRay (interior design) render look amazing; hate to think of how much work those must have taken.