Morphological Anti-Aliasing (MLAA) has been successfully applied in the game industry as a cheap and efficient method to alleviate the cost of Multi-Sampling Anti-Aliasing (MSAA) but also has been used in cel animation production for years. Talk MorphoLogical Anti-Aliasing in Production To set context for the discussion we’ll begin with a brief overview of our production pipeline.We’ll also discuss how we approach research at Pixar, including how we pick problems, what we look for when we consider the adoption of external research, and how we transfer technology to the rest of the studio.ģ:30 – 4:50 : Session 3 (chair: Cyrus Wilson) This talk will survey a number of current and recent projects underway in the Research Group at Pixar. Research Group Lead, Pixar Animation Studios
We detail specific software design choices, highlight successful intra- and inter-studio collaborations, and discuss our current development efforts towards scalable, maintainable tools for the ever increasing complexity involved in visual effects production.Ġ2:15-03:15 : Keynote Recent Research at Pixar This paper explains the evolution of our creature pipeline, focusing on look development, lighting and asset management. Paper Developing a Unified Pipeline with CharacterĪndrew Kaufman, John Haddon, Ivan Imanishi, Lucio Moser. Go”, a mid cost VFX film totaling about $10M for the entire VFX budget.
In this talk, we present our experiences in building an efficient, practical fur pipeline for generating 900+ shots containing two digital gorillas for the production of the movie “Mr. Talk Building Efficient Fur Pipeline for a Low Cost Production of Creature-based Feature Film We seek to raise the standard of quality for automatic and minimal cost 3D content in previz by automated generation of a believable 3D character performance from only an audio track and its transcription. Paper Towards Higher Quality Character Performance in Previz The set reconstruction process is guided interactively to efficiently produce satisfactory geometry, though the form and quality of input data may vary.ġ1:30-12:40 : Session 2 (chair: Doug Epps) We present a photogrammetry system which is flexible with respect to input data, and with respect to the specific needs of a shot. Mikhail Smirnov, Eugene Vendrovsky, Joe Mancewicz, Jonathan Meier, Cyrus Wilson. Paper An Interactive System for Set Reconstruction from Multiple Input Sources We accomplished that by utilizing deep images and casting curved rays through them to insure a smooth blend between viewing parameters of the stereo rigs in the user-defined transition area. We present a new stereoscopic compositing technique that combines volumetric output from several stereo camera rigs. Paper Stereo Compositing Accelerated by Quadtree Structures in Piecewise Linear and Curvilinear Spaces Image copyright – © LEIJI MATSUMOTO / CAPTAIN HARLOCK Film Partners. We show that our method reduces the amount of artifacts in the transfer and improves the efficiency of the animation pipeline. We propose a method to improve the quality of the facial blendshapes transfer by introducing contact and smoothness awareness to the original deformation transfer.
Paper Smooth Contact-Aware Facial Blendshapes Transfer
In this talk, Bill Polson will explain: What is subdivsion? What are the extensions and why are they useful? How are they drawn and evaluated on CPUs and GPUs? He’ll demonstrate OpenSubdiv in action, and provide an update on the project, the engineering roadmap, and industry adoption.ġ0:00-11:15: Session 1 (chair: Oliver James)
OpenSubdiv provides the definitive implementation of this specification - all the same code that Pixar uses - and provides it on a variety of platforms, including OpenGL, OpenCL, CUDA, Direct X, and others. The *Extended* Catmull-Clark specification adds fine control of these surfaces, including: concise encoding of edge sharpness (creases), local hierarchical edits, and explicit per-face texture control. DigiPro2013 proceedings are now in the ACM DLĠ9:00-09:50 : Invited Talk Pixar’s OpenSubdivĬatmull-Clark Subdivision provides a natural mathematical extension of B-Spline patches into arbitrary topology.