Domain: renderman.org
Stories and comments across the archive that link to renderman.org.
Comments · 19
-
Re:Attention Procedural Programmers
-
Re:Don't kid yourselves
Like Pixar destroying lives of former employees?
The demise of BMRT & entropy -
Re:What about Pixar's Software Arm?
Actually the name RenderMan refers to the standard. PRMan or Photorealistic RenderMan refers to the actual Pixar product (the renderer) that applies that standard. Sometimes the names are used interchangably, but since there are other RenderMan renderers (Aqsis, RenderDotC, AIR, 3Delight, Pixie) it's better to make the distinction between PRMan and RenderMan.
As far sa what RenderMan as a specification is, it's mainly composed of 2 parts: a scene description part and the shading language. The scene description describes the 3D world and can be called procedutrally via C/C++ calls (or thoer bindings) or via RIB (RenderMan Interface Bytestream whcih can be text or binary). The shading language describes the apperance of objects (what they look like and their interaction with lighting).
As far as the software development, it's actually split. Some of the development (I think most of the Artist Tools but also some of PRMan) is done in Seattle, where Pixar has an office headed by Dana Batali (he is director of RenderMan product development). Some of the stuff is also done in the bay, after many of the people who worked on PRMan since its beginning, like Tony Apodaca, Tom Duff, etc. are at Emeryville. If Pixar got sold it would certainly create some commotion as tons of VFX studios use it. And that's not counting the fact that Lucas had a deal with Pixar that they would get first look at any Pixar technologies and some ILM people have rights to access the PRMan source code. It could get messy.
You can get some RenderMan info here:
http://renderman.org/ -
Re:How do they do this?
Often some combination of Maya on the front-end with lots of custom scripts is used to generate RIB data which is rendered by a RenderMan interface compliant renderer. You won't get PRMan (Pixar's implementation) very cheaply, but there some other good implentations. You may still be able to dig up a copy of the Blue Moon Rendering Tools (BMRT) somewhere. RenderDotC has a resolution limited evaluation version. There's also Aqsis and which is GPL, but I haven't used it so I can't really speak on its quality or how complete it is. Pixie is another GPL renderer which seems pretty cool. All of them can run on Linux.
Part of the reason that RenderMan renderers are so popular is that they let you write custom shader scripts which they interpret to shade and deform the geometry. It's kind of the equivalent of GPU pixel and vertex shaders (which are partly inspired by RenderMan, by the way.)
If you want books, Advanced RenderMan is excellent. (My copy is well worn.) The RenderMan Repository has some examples and some good basic information on RenderMan. They also have a nice collection of PDF's of the course notes from the RenderMan SIGGRAPH courses over the years. Those typically have chapters by folks from the studios describing how they accomplished certain effects. I highly recommend it for getting the flavor of the the thing. Lastly, there's the RenderMan spec itself from Pixar, though I certainly wouldn't recommend it as an introduction to all this.
My suggestion to start trying this stuff out would be too Google for RIB export scripts/plugins for Blender (I know they exists, but I don't use Blender), grab one of the GPL renderers out there and pick up a copy of Advanced RenderMan to start learning the shading language. Good luck! -
Re:How do they do this?
Often some combination of Maya on the front-end with lots of custom scripts is used to generate RIB data which is rendered by a RenderMan interface compliant renderer. You won't get PRMan (Pixar's implementation) very cheaply, but there some other good implentations. You may still be able to dig up a copy of the Blue Moon Rendering Tools (BMRT) somewhere. RenderDotC has a resolution limited evaluation version. There's also Aqsis and which is GPL, but I haven't used it so I can't really speak on its quality or how complete it is. Pixie is another GPL renderer which seems pretty cool. All of them can run on Linux.
Part of the reason that RenderMan renderers are so popular is that they let you write custom shader scripts which they interpret to shade and deform the geometry. It's kind of the equivalent of GPU pixel and vertex shaders (which are partly inspired by RenderMan, by the way.)
If you want books, Advanced RenderMan is excellent. (My copy is well worn.) The RenderMan Repository has some examples and some good basic information on RenderMan. They also have a nice collection of PDF's of the course notes from the RenderMan SIGGRAPH courses over the years. Those typically have chapters by folks from the studios describing how they accomplished certain effects. I highly recommend it for getting the flavor of the the thing. Lastly, there's the RenderMan spec itself from Pixar, though I certainly wouldn't recommend it as an introduction to all this.
My suggestion to start trying this stuff out would be too Google for RIB export scripts/plugins for Blender (I know they exists, but I don't use Blender), grab one of the GPL renderers out there and pick up a copy of Advanced RenderMan to start learning the shading language. Good luck! -
Re:NOT proprietary rendering technology
Renderman is not a renderer it is a specification for interoperability between modeling tools and renderer (like XMI for software engineering tools except that it works.)
Pixar's renderer is actually PRMan.
From Renderman.org:
There are a lot of people when you hear them talking about RenderMan and how great the images are from it, etc. They are most likely really talking about Pixar's PhotoRealistic RenderMan® (PRMan).
RenderMan is actually a technical specification for interfacing between modeling and rendering programs. From 1998 until 2000 the published RenderMan Interface Specification was known as Version 3.1. In 2000 Pixar published a new specification, Version 3.2. Coming soon Version 3.3 -
Re:Open Source Projects?
It was a published paper so yes the math is put in the open. Specific implmentations might be closed. mental ray 3.3 just added a few more SSS capabilities.
But you could study some implementations in the open. Christophe Hery's Stupid RAT trick slides are aailable, and his course notes from last SIGGRAPH are accesible as well:
Stupid Rat Trick (Shadow Buffers)
RenderMan SIGGRAPH course notes (check the ones from 2003)
ZJ's Slim template at DeathfallPotentially the open implementations like Aqsis, Pixie and Toxic are getting capablle enough that you might have free implementations soon enough.
-
Re:Open Source Projects?
It was a published paper so yes the math is put in the open. Specific implmentations might be closed. mental ray 3.3 just added a few more SSS capabilities.
But you could study some implementations in the open. Christophe Hery's Stupid RAT trick slides are aailable, and his course notes from last SIGGRAPH are accesible as well:
Stupid Rat Trick (Shadow Buffers)
RenderMan SIGGRAPH course notes (check the ones from 2003)
ZJ's Slim template at DeathfallPotentially the open implementations like Aqsis, Pixie and Toxic are getting capablle enough that you might have free implementations soon enough.
-
Nice to see some good out of BMRT/Exluna.Renderman.org's summary of Exluna & BMRT describes where much of this technology probably came from.
For those who don't remember, BMRT was a really cool RenderMan based renderer that Pixar had some sort of love/hate relationship with. IIRC, they used it, yet they sued the company. At the end nVidia bought them, though it wasn't clear why at the time.
-
Re:Took them long enough.BMRT/Entropy creators were ex-Pixar employees who stole (eeek, harsh word around here I know) from Renderman.
-
Re:Renderman modelling
...but again doesn't support animation.
Mm. As noted below, not a lot does wrt renderman, at least, not without paying lotsa dosh. I presume you've had a look thru' the RMR links page - there's a couple of possibilities there, but mostly alpha. The other thing to do would be to get in touch with the people who created this and this rather impressive animation and ask how they did it. You'll probably have to email the aqsis site maintainers to get their contact details.
Also, I forgot to link the brand new Renderman Academy. It's an ambitious project, but there's already a lot of good info there. Nothing on animation yet, unfortunately.
I mentioned Blender above - if you are willing to take on the task of learning to use it, you'll find a very powerful animation engine in there. There's a chap integrating renderman support directly into the main program, but that's apparently on hold until the middle of the year. In the meantime, there's a very adequate exporter called Blenderman (also see this). Of course, if what you're wanting is a directly programmable animation setup ala AL and SDL, this won't help.
Good luck with it - if you have any success, see if you can get it into a Slashback.
-
Re:Pixar Renderman
They didn't only use Pixar's renderer, and here is a dark tale of what Pixar did to another small startup Exluna
Notice how that link goes to Nvidia? There's a reason why. Quite a few years ago, a rather genius programmer left Pixar and started up his own company to write a competing renderer called Entropy. Pixar's renderer, while very fast and the basis for many effects and animation piplines throughout the industry, was getting a bit long in the tooth. It didn't have any raytracing abilities (outside of some clever hacks), and completely lacked the global illumination abilities that were neccassary for some believable lighting models.
Why do I bring this up? Because Gollum was almost exclusively rendered on this renderer. Pixar's Renderman was not capable of doing some of the stuff they needed for that beautiful skin shader at the time they developed Gollum.
Pixar didn't take lightly to this. They launched a lawsuit against Exluna saying they were violating certain patents they held regarding some antialiasing algorithms. Never mind that the renderer was far more advanced and was a complete drop in replacement for Pixar's competing product. This was a straight up ploy to get rid of the competition.
To this day, the Exluna developers still say they did not violate those patents and that they would have won in court. However, winning in court would have destroyed the company. Instead, they sold the company to Nvidia, where they are working on some even more advanced stuff, but under the protection of a larger and well financed (and lawyer'd) company.
There are may other Renderman based renderers out there, all of varying capabilities. Pixie, while technically advanced and written by a brilliant graduate student at berkely, has a few rough edges and is missing some important features. Aqusis is progressing nicely, but doesn't have many features that I rely on. Mental Ray, while not renderman compatable, has all the features and more, but you pay for it in speed. Right now, I'm using Pixie for my tests. It's free for me, but I wouldn't trust it in production just yet. For production I would still choose Pixar's Renderman, which has since incorporated much of the lighting features available in other renderers (somewhat pushed by the demands of their clients, but mostly because they used a lot of those special lighting tools in Finding Nemo).
For more information on all available Renderman capable renderers and how to use them, I suggest visiting the Renderman Repository
Alright, back to work for me. I'm supposed to present this skin shader after new years.
Rich -
Re:Blender is getting mature
The founders of Exluna were accused by Pixar, their former employer, of misappropriation of trade secrets, copyright infringement and patent infringement. Exluna was able to settle the lawsuit by ending BMRT.
The Demise of BMRT & entropy
Here's a list of RenderMan-compliant renderers. Some of them, like AQSIS, are open source.
-
radiosity used to be a feature of BMRT/entropyI posted this in response to another thread, but there used to be a (slower) implementation of the RIB-standard scene rendering process called Blue Moon Render Tools. See here:
http://www.dctsystems.freeserve.co.uk/rmanBasics.h tml
It was later commercially expanded into a faster program called 'entropy'. Exluna was a company that Larry Gritz and some coworkers from Pixar (Gritz joined and then left Pixar) founded. Apparently entropy was fast enough for commercial use (eg. LOTR-scale projects that required photorealistic scenes). Pixar did not like this. At all. The sequelae were as documented here:
http://www.renderman.org/RMR/OtherLinks/blackSIGGR APH.html
Now this is probably not relevant to you if you're working at wetafx or ILM or other big shops, but it's still kind of a shame that, when a product came along that WAS able to compete with PRMan, Pixar chose to squash it with lawyers rather than innovation. I'm not claiming that the case was clear-cut, but the original lawsuit apparently lacked legal merit, and Pixar then went after the individual founders of the company in an effort to drain their resources, which is rather unimpressive.
So the point is that, for a time, there WAS an alternative to PRman for big (cinematic) projects, and Pixar used lawsuits to bury it.
D'oh.
-
Used to hav MULTIPLE RenderMan compatible programsbut I looked into what happened with ExLuna/nVidia and Pixar, and here's the scoop...
http://www.renderman.org/RMR/OtherLinks/blackSIGG
R APH.htmlAs you will see on the page, Pixar made BMRT and entropy 'go away' in July of this year. So, it looks like that is why Aqsis is being suggested as the only remaining contender.
-
Re:uhm, no. Wired has the scoop on Pixar's interna
Well if you want more info on internals just look for all the Pixar papers and course notes from SIGGRAPH and the 2 RenderMan books. Besides the renderMan course notes they have also presented on several FX related courses. Here is a start:
RenderMan docsAnd those moviews are not easy, they take an average of 3 to 4 years to make. Next one will take place under the sea.
-
Re:Looking Inside Pixar
No, actually they used the same renderer, Photorealistic RenderMan. Renderman is a spec, and PRMan is an implementation of it. If anything Pixar uses development versions of PRMan but eventually they make it to the next version of PRMan. Square Pictures actually worked closely with Pixar and many of the requests they had were incorporated into PRMan.
What you are referring to is shaders which are "little" programs that can calculate light and shading. Every place will create their own, so the skin shader used in Final Fantasy is different from the one in Monsters. But the rendere is basicly the same.
If you want a more technical discussion of this just check the RenderMan SIGGRAPH 2001 course notes. They have a chapter by Kevin Bjorke, who had several presentations during SIGGRAPH, about what Square did with RenderMan for Final Fantasy: of th
RenderMan SIGGRAPH course notes -
Re:Looking Inside Pixar
This is a common confusion. RenderMan is a rendering interface a spec. Usually it's compared as the Postscript of 3D. You can implement a renderer that follows the RenderMan spec. Pixar's implementation is called Photorealistic RenderMan or PRMan for short. It uses the REYES (which stands for Renders Everything You Ever Saw)architecture for rendering (in gross termsZ buffer scanline renderer). Many people when they say RenderMan they actually mean PRMan. PRMan is sold by Pixar along with the RenderMan Artist Tools or RAT. But there exists many other implementations of the RenderMan spec, including BMRT, Entropy, RenderDotC, AQSIS, and AIR among others.
But when Pixar got started there was barely any adequate off the shelf software, so like many others back then (like PDI, Blue Sky Studios, Abel and Associates, triple I, etc), they had to create their own tools. Actually you can see mention of it at the end of the movie: Marionette is their animation environment, previously referred to as menv.
RenderMan Interface
Exluna (makers of BMRT and Entropy)
AQSIS
RenderDotC
3DLight
AIR
RenderMan Repository -
Re:quite a change since Tron
Well you might be right, but as far as I know no major movie FX or animation have been done over the Net. Not only because techical considerations but also dealings with the studios. Studios want to protect their investement and market the film according to their plans. You can't just transmit stuff over the net without concerns that someone might snoop in. Besides probably doing the transfers to Exabyte and sending them via courier are faster and more secure.
There might be a few cases though. For example DD has a commercials divisions (also does music videos) and used to have an interactive division. Maybe they worked sending stuff over the net for these smaller projects. ILM Commercial Productions seems to be doing this also (but not the main film FX part):
ILMCP Client AccessSo maybe you are right.
The stuff your are mentioninmg is referred to 4:3 pullup or pulldown (depending on direction) about transfering from film at 24 frames per second to video at 30 frames per second and viceversa.
But the other thing about FX is not that the plates are scanned typically at 2K resolution but all the data files involved in the FX elements: models, shaders, texture maps, etc. In the renderMan newsgroup there was a thread some time ago that had some very interesting stats. While a single frame might be around, say 10MB, RIBs and textures alone for models might take dozens to mabybe even hundreds of MBs. I couldn't find the reference but it was probably posted by Tom Duff or Larry Gritz. This past SIGGRAPH some people discussed some info about their projects. There is some info about Shrek out there, stating that on average they had 2 to 3 GB per frame (though FX would be a bit lower since not all the image is 3D). It's usually a lot less number of frames, but the amount of data per frame is huge. Take Pearl Harbor that only had around 200 shots but they occupy like 50 min of screen time, or long seuqnces like the Contact oull back or the frozen NY flyby in AI.
Shrek rendering staristicsYou might want to check the RenderMan course notes from this past SIGGRAPH, also a lot of info. For the Gungan battle of episode 1, all the baked RIBs accounted for 64 GB of data. The smallest RIB for one cycle was 209 MB.
Renderman repositoryAbout Major Damage, besides SIGGRAPH I think it was discussed in the CH_Char mailing list, now web forum. Jeff Lew used (or maybe still) posts around there.
CG Char