Pixar’s encouraging performance – by giving away the good stuff

Pixar is contributing to the open source community by making their interactive friendly, high speed Open SubDiv available to all.

OpenSubdiv is a set of open source libraries that implement high performance subdivision surface (subdiv) evaluation on massively parallel CPU and GPU architectures. The code is optimized for drawing deforming subdivs with static topology at interactive framerates. The resulting limit surface matches Pixar’s RenderMan to “numerical precision”.

OpenSubdiv is free to use for commercial or non-commercial use. This is the same code that Pixar uses internally for animated film production. The posting comments that “our intent is to encourage high performance accurate subdiv drawing by giving away the “good stuff”.

This is the fifth generation subdiv library in use by Pixar, in a lineage that started with code written by Tony DeRose and Tien Truong for the short film Geri’s Game in 1996. Each generation has been a from-scratch rewrite that has built upon Pixar’s real world experience in using subdivision surfaces to make animated films. This code is used by Pixar in their current films and is very much current, and expected to be used in all future films, much like RenderMan. Pixar has said that it will release updates as open source at the same time they are rolled out to Pixar animation production.

The code is based on this paper by Niessner, Loop, Meyer, and DeRose.

Input base mesh (left), subdivision patch structure (center), & final model rendered with the method (from the paper).

In the paper, the authors explain their approach for high-performance GPU based rendering of Catmull-Clark subdivision surfaces. “Unlike previous methods, our algorithm computes the true limit surface up to machine precision, and is capable of rendering surfaces that conform to the full RenderMan specification for Catmull-Clark surfaces. Specifically, our algorithm can accommodate base meshes consisting of arbitrary valence vertices and faces, and the surface can contain any number and arrangement of semi-sharp creases and hierarchically defined detail.”

The code includes ‘feature-adaptive refinement’ logic which is used to adaptively refine coarse topology near features like “extraordinary vertices and creases in order to make the topology amenable to cubic patch evaluation”. At this stage the beta supports uniform subdivision, for the release, OpenSubdiv will support feature adaptive subdivision.

Pixar posted that “we’re seeing best results here with CUDA”, and also stated that they would like help with some further work from the opensource community in particular, Pixar internally is not heavily using Alembic so “Alembic support would be wonderful, but we don’t use Alembic enough internally to do the work.”

Also they commented that “John Lasseter loves looking at film assets in progress on an iPad. If anyone were to get this working on iOS he’d be looking at your code, and the Apple geeks in all of us would smile.”

Pixar is targeting release for end of year 2012, “hopefully earlier than that. We have the patch code working in a very rough implementation but need to rewrite that in a development branch for release-ready code. Let us know if you’re interested in contributing to that effort!” Pixar’s web site for the OpenSubDiv.

OpenSubdiv is entering open beta for SIGGRAPH 2012.

Pixar at SIGGRAPH 2012

Teapot giveaways each day at midday from the Pixar booth – always popular. The RenderMan User Group is on Wednesday night at 6pm. Pixar is also giving many technical papers covering key technology from the film Brave, (Thursday afternoon 2pm) and production talks earlier in the week (Monday 3:30pm). These are part of the full conference.

There is a Pixar lecture series on the show floor (open to people without full conference passes) with great speakers including our fxphd Prof, Christos Obretenov, along with ILM’s Jeff White on Avengers, and Martin Hill from Weta.

Here are just a few key talks from that schedule:

Exploration in Physically Plausible Shading
Christos Obretenov – Co-founder/Developer, LollipopShaders.com and instructor at fxphd

Christos Obretenov. Van model – Jason Gagnon/Pixels&Polygons, HDRI Map – Arkell Rasiah

Presentation – Christos will begin by examining a rendering scene with professionally captured unclipped HDRI maps and a traditional shaders & lights setup in PRMan, and see where it leads to artifacts and inconsistent material settings across different lighting environments. We then render the same scene but in our Physically Plausible Shading environment in PRMan16/17, demonstrate how our previous issues have been resolved, and introduce the physically plausible surface and light shaders. He will be talking about a full Physically Plausible system including energy conservation, physical properties in materials, lights with area for solid angle sampling, importance sampling, ray-tracing, as well as plausible fur and volumes. This is directly related to Christos fxphd course this term.

Creating the Hulk and a digital New York for Marvel’s The Avengers
Jeff White – Visual Effects Supervisor, ILM

Jeff will discuss the challenges ILM had to overcome in creating a lifelike Hulk and a completely digital New York City for him to smash in the 2012 blockbuster film The Avengers. Jeff White joined Industrial Light & Magic in 2002 as a creature technical director. He is a graduate of Ithaca College’s Cinema and Photography program and has a Masters of Fine Arts from the Savannah College of Art and Design. Prior to graduate school, White worked as technical director for Laika Studios in Portland, Oregon.

Since arriving at ILM, he has worked as a Creature Technical Director on a variety of films including: Pirates of the Caribbean: Dead Man’s Chest, Lemony Snicket: A Series of Unfortunate Events, War of the Worlds, and Star Wars: Episode III “Revenge of the Sith”. Responsibilities on the shows included rigging and enveloping a wide variety of creatures running flesh, cloth and hair and muscle simulations. White served as Digital Production Supervisor on Transformers, and Associate Visual Effects Supervisor on the two subsequent Transformers films.

Trilobites & Tentacles: VFX for Prometheus
Martin Hill – VFX Supervisor, Weta Digital

This talk will focus on creating the VFX for Prometheus, and the many challenges that Weta overcame to deliver stunning CGI for this sci-fi movie.  Martin Hill has been involved in many productions at Weta Digital, having worked on King Kong, The Water Horse, Rise of the Planet of the Apes, The Adventures of Tintin, Prometheus, Abraham Lincoln: Vampire Hunter, and is currently working on The Hobbit.


For the full schedule of Pixar free talks click here


1 thought on “Pixar’s encouraging performance – by giving away the good stuff”

  1. Pingback: The State of Rendering – Part 1 | 次时代人像渲染技术XGCRT

Comments are closed.