David Baraff, Andrew Witkin, Michael Kass
August 2003
Deficient cloth-to-cloth collision response is the most serious shortcoming of most cloth simulation systems. Past approaches to cloth-cloth collision have used history to decide wheter nearby cloth regions have interpenetrated. The biggest pitfall of history-based methods is that an error anywhere along the way can give rise to persistent tangles...This paper describes a history-free cloth collision response algorithm based on global intersection analysis of cloth meshes at each simulation step...
Available in the Proceedings of SIGGRAPH 2003
Dana Batali, Byron Bashforth, Chris Bernardi, Per H. Christensen, David Laur, Christophe Hery, Guido Quaroni, Erin Tomson, Thomas Jordan, Wayne Wooten
July 2003
Siggraph 2003 course notes.
Gary Bruins, Jon Reisch
May 2007
A particularly heavy effects sequence in the film Ratatouille depicts a colony of rats fleeing their home and crossing a long narrow river in several makeshift boats during a rainy afternoon. Nearly the entire sequence was photographed inches above the ground from a s perspective. Although this perspective provided a very interesting point of view of the world, it also brought an extra level of difficulty to an already challenging sequence. A raindrop creating a seemingly insignificant two-inch splash in the river could easily span one third the height of our final image. Described below is our approach to achieving believable water related effects from the point of view of a rat.
Available as Pixar Technical Memo #07-14
Gordon Cameron, Robert Russ and Adam Woodbury
May 2007
Contact is natural in the real world but often avoided in 3D animated features. Animators tend to make acting decisions that minimise or avoid contact within and between their characters and the world, and when apparent contact does occur, it can tend to feel both "floaty" and unrealistic. "Ratatouille" called for dynamic, tactile characters that would feel very much a part of their animated environment, squashing into and against each other, props and the world itself. The contact needed to be compelling and believable within that cartoon world, and with that in mind we aimed to develop and deploy technology that would allow animators to easily setup collision relationships, make acting decisions with contact response interactively, and alter the resultant shapes.
Available as Pixar Technical Memo #07-10
Jun Han Cho, Athena Xenakis, Stefan Gronsky, Apurva Shah
July 2007
The passion for cooking and food are the central theme of Pixar's recent film - Ratatouille. This complex and multi-faceted problem posed many challenges that were solved using diverse computer graphics and production techniques. In this course we will comprehensively cover all aspects including modeling, dressing, shading, lighting and effects.
The story called for working cooking stations and sloppy mess of a busy, functional kitchen. We will review some of the set concepts, visual framework and even dynamics simulation techniques that were used to create this illusion. We will illustrate with several examples including final plated dishes, mis-en-place setups and the Food Locker.
The challenge of shading food on Ratatouille was to work within the stylized look of the film and yet keep it recognizable and appealing to eat. We developed subtle illumination techniques that added up to a general approach we could use on a variety of objects. We will breakdown examples ranging from vegetables to plated dishes.
Lighting played a key role in making the food look appetizing, a task further complicated by different types of food such as bread, cheese, soup and wine that pushed the boundaries of standard surface based lighting. We will discuss our general approach to lighting food as well as specific challenges and solutions posed by the various dishes.
The film called for diverse cooking techniques ranging from chopping and peeling, to stirring and ladling. We will discuss Ratatouille's creative problems and the underlying challenge they represent as well as our solutions to them. We will also apply it to various case studies like chopping carrots, rolling dough and preparing potato-leek soup.
Although the course specifically discusses how we dealt with food in Ratatouille we want to emphasis that the basic approach and techniques can be used for other complex, multi-disciplinary visual challenges.
Available as Siggraph 2007 Course Notes, Course Number 30
Per H. Christensen, Julian Fong, David Laur, Dana Batali
September 2006
This paper describes how we extended Pixar's RenderMan renderer with ray tracing abilities. In order to ray trace highly complex scenes we use multiresolution geometry and texture caches, and use ray differentials to determine the appropriate resolution. With this method we are able to efficiently ray trace scenes with much more geometry and texture data than there is main memory. Movie-quality rendering of scenes of such complexity had only previously been possible with pure scanline rendering algorithms. Adding ray tracing to the renderer enables many additional effects such as accurate reflections, detailed shadows, and ambient occlusion.
The ray tracing functionality has been used in many recent movies,
including Pixar's latest movie 'Cars'. This paper also describes some
of the practical ray tracing issues from the production of 'Cars'.
Available in Proceedings of the IEEE Symposium on Interactive Ray Tracing
2006, pages 1-6. IEEE 2006
Per H. Christensen, Dana Batali
June 2004
We introduce a tiled 3D MIP map representation of global illumination data. The representation is an adaptive sparse octree with a "brick" at each octree node; each brick consists of 8 cubed voxels with sparse irradiance values. The representation is designed to enable efficient caching. Combined with photon tracing and recent advances in distribution ray tracing of very complex scenes, the result is a method for efficient and flexible computation of global illumination in very complex scenes. The method can handle scenes with many more textures, geometry, and photons than could fit in memory. We show an example of a CG movie scene that has been retrofitted with global illumination shading using our method.
Published as pp. 133-141 in the Proceedings of the Eurographics Symposium on Rendering 2004, Eurographics/ACM, June 2004.
Per H. Christensen, David Laur, Julian Fong, Wayne Wooten, Dana Batali
September 2003
When rendering only directly visible objects, ray tracing a few levels of specular reflection from large, low curvatures surfaces, and ray tracing shadows from point-like light sources, the accessed geometry is coherent and a geometry cache performs well. But in many other cases, the accessed geometry is incoherent and a standard geometry cache performs poorly: ray tracing of specular reflection from highly curved surfaces, tracing rays that are many reflection levels deep, and distribution ray tracing for wide glossy reflection, global illumination, wide soft shadows, and ambient occlusion. Fortunately, less geometric accuracy is necessary in the incoherent cases. This observation can be formalized by looking at the ray differentials for different typics of scattering: coherent rays have small differentials, while incoherent rays have large differentials. We utilize this observation to obtain efficient multiresolution caching of geometry and textures (including displacement maps) for classic and distribution ray tracing in complex scenes. We use an existing multiresolution caching scheme (originally developed for scanline rendering) for textures and displacement maps, and introduce a multiresolution geometry caching scheme for tessellated surfaces. The multiresolution geometry caching scheme makes it possible to efficiently render scenes that, if fully tessellated, would use 100 times more memory than the geometry cache size.
Published as pp. pp. 543-552 in Computer Graphics Forum (Eurographics 2003 Conference Proceedings), Blackwell Publishers, September 2003.
Per H. Christensen
July 2003
This survey gives an overview of the use of importance, an adjoint of light, in speeding up rendering. The importance of a light distribution indicates its contribution to the region of most interest --- typically the directly visible parts of a scene. Importance can therefore be used to concentrate global illumination and ray tracing calculations where they matter most for image accuracy, while reducing computations in areas of the scene that do not significantly influence the image. In this paper, we attempt to clarify the various uses of adjoints and importance in rendering by unifying them into a single framework. While doing so, we also generalize some theoretical results --- known from discrete representations --- to a continuous domain.
Available as IEEE Transactions on Visualization and Computer Graphics (TVCG), Volume 9, Number 3, pages 329-340. IEEE, July 2003.
Patrick Coleman, Eric Froemling
May 2007
In Disney/Pixar's Ratatouille, the creation of believable cooking environments with all their complexity has been an important element in presenting a rich world that helps draw the audience into the story. Part of that complexity arises in the preparation of food before cooking. To create complex animations of food in preparation, we designed a system that uses an animated cutting object, such as a knife, to procedurally model, simulate, deform, and prepare for shading various geometric food models as they are sliced, chopped, peeled, or otherwise broken apart. The motion of a knife (or other object) is analyzed relative to the food model to determine a sequence of cutting operations that will remodel the object as a collection of pieces. As each new piece is created, it is added to a physical simulation to generate believable response motion. We transfer surface shading parameterizations and scalar fields to resulting faces that correspond to surfaces on the original object, and we generate additional scalar fields to assist users in shading new internal surface faces. This approach to creating chopping effects entirely dependent on an animated knife allows animators to focus on character performance without needing to consider the complex modeling and motion associated with chopping.
Available as Pixar Technical Memo #07-13
Robert L. Cook, John Halstead, Maxwell Planck, David Ryu
May 2007
Many renderers perform poorly on scenes that contain a lot of detailed geometry. The load on the renderer can be alleviated by simplification techniques, which create less expensive representations of geometry that is small on the screen. Current simplification techniques for high-quality surface-based rendering tend to work best with element detail (i.e., detail due to the complexity of individual elements) but not as well with aggregate detail (i.e., detail due to the large number of elements). To address this latter type of detail, we introduce a stochastic technique related to some approaches used for point-based renderers. Scenes are rendered by randomly selecting a subset of the geometric elements and altering those elements statistically to preserve the overall appearance of the scene. The amount of simplification can depend on a number of factors, including screen size, motion blur, and depth of field.
Additional materials: [SiggraphSlides.pdf], [WithAndWithoutSimplification.mov]
Available in the Proceedings of Siggraph 2007.
Available as Pixar Technical Memo #06-05a
Other versions:
Robert L. Cook, Tony DeRose
August 2005
Noise functions are an essential building block for writing procedural shaders in 3D computer graphics. The original noise function introduced by Ken Perlin is still the most popular because it is simple and fast, and many spectacular images have been made with it. Nevertheless, it is prone to problems with aliasing and detail loss. In this paper we analyze these problems and show that they are particularly severe when 3D noise is used to texture a 2D surface. We use the theory of wavelets to create a new class of simple and fast noise functions that avoid these problems.
Additional materials: [RapLyrics.txt]
Available in the Proceedings of SIGGRAPH 2005
Robert L. Cook, Loren Carpenter, Edwin Catmull
July 1987
An architecture is presented for fast high-quality rendering of complex images. All objects are reduced to common world-space geometric entities called micropolygons, and all of the shading and visibility calculations operate on these micropolygons. Each type of calculation is performed in a coordinate system that is natural for that type of calculation. Micropolygons are created and textured in the local coordinate system of the object, with the result that texture filtering is simplified and improved. Visibility is calculated in screen space using stochastic point sampling with a z buffer. There are no clipping or perspective calculations. Geometric and texture locality are exploited to minimize paging and to support models that contain arbitrarily many primitives.
Robert L. Cook
January 1986
Ray tracing, ray casting, and other forms of point sampling are important techniques in computer graphics, but their usefulness has been undermined by aliasing artifacts. In this paper it is shown that these artifacts are not an inherent part of point sampling, but a consequence of using regularly spaced samples. If the samples occur at appropropriate nonuniformly spaced locations, frequencies above the Nyquist limit do not alias, but instead appear as noise of the correct average intensity. This noise is much less objectionable to our visual system than aliasing. In ray tracing, the rays can be stochastically distributed to perform Monte Carlo evaluation of integrals in the rendering equation. This is called distributed ray tracing and can be used to simulate motion blur, depth of field, penumbrae, gloss, and translucency.
Available in ACM Transactions on Graphics, Volume 6, Number 1, January 1996.
Robert L. Cook, Thomas Porter, Loren Carpenter
July 1984
Ray tracing is one of the most elegant techniques in computer graphics. Many phenomena that are difficult or impossible with other techniques are simple with ray tracing, including shadows, reflections, and refracted light. Ray directions, however, have been determined precisely, and this had limited the capabilities of ray tracing. By distributing the directions of the rays according to the analytic function they sample, ray tracing can incorporate fuzzy phenomena. This provies correct and easy solutions to some previously unsolved or partially solved problems, including motion blur, depth of field, penumbras, translucency, and fuzzy reflections. Motion blur and depth of field calculations can be integrated with the visible surface calculations, avoiding the problems found in previous methods.
Available in the Proceedings of SIGGRAPH 1984
Robert L. Cook
July 1984
Shading is an important part of computer imagery, but shaders have been based on fixed models to which all surfaces must conform. As computer imagery becomes more sophisticated, surfaces have more complex shading characteristics and thus require a less rigid shading model. This paper presents a flexible tree-structured shading model that can represent a wide range of shading characteristics. The model provides an easy means for specifying complex shading characteristics. It is also efficient because it can tailor the shading calculations to each type of surface.
Available in the Proceedings of SIGGRAPH 1984.
Robert L. Cook, Kenneth E. Torrance
January 1982
A new reflectance model for rendering computer sythesized images is presented. The model accounts for the relative brightness of different materials and light sources in the same scene. It describes the directional distribution of the reflected light and a color shift that occurs as the reflectance changes with incidence angle. A method for obtaining the spectral energy distribution of the light relfected from an object made of a specific real material is presented, and a procedure for accurately reproducing the color associated with the spectral energy distribution is discussed. The model is applied to the simulation of a metal and a plastic.
Published in Transactions on Graphics, Vol. 1, No. 1, January 1982.
Tony DeRose, Michael Kass, Tien Truong
August 1998
The creation of believable and endearing characters in computer graphics presents a number of technical challenges, including the modeling, animation and rendering of complex shapes such as heads, hands, and clothing. Traditionally, these shapes have been modeled with NURBS surfaces despite the severe topological restrictions that NURBS impose. In order to move beyond these restrictions, we have recently introduced subdivision surfaces into our production environment. Subdivision surfaces are not new, but their use in high-end CG production has been limited.
Here we describe a series of developments that were required in
order for subdivision surfaces to meet the demands of high-end
production. First, we devised a practical technique for
constructing provably smooth variable-radius fillets and blends.
Second, we developed methods for using subdivision surfaces in
clothing simulation including a new algorithm for efficient
collision detection. Third, we developed a method for
constructing smooth scalar fields on subdivision surfaces, thereby
enabling the use of a wider class of programmable shaders. These
developments, which were used extensively in our recently
completed short film Geri's game, have become a highly
valued feature of our production environment.
Available in the Proceedings of SIGGRAPH 1998.
Eric Froemling, Tolga Goktekin, Darwyn Peachey
May 2007
In Pixar's Ratatouille, a key story point involves a rat being swept through the sewers of Paris, plummeting down waterfalls and along steeply sloping tunnels, through a series of high-speed S- bends which cause the torrent of water to bank up sharply on each turn. Bringing the director's vision of this wild and perilous rapids sequence to the screen required us to use a wide variety of water effects techniques to give the appearance of rushing water, spray, foam and bubbles. One of the greatest challenges was to pull these diverse techniques together into a seamless sequence.
Available as Pixar Technical Memo #07-03
Tolga Goktekin, Jon Reisch, Darwyn Peachey, Apurva Shah
May 2007
Creating the digital effects for cooking in Ratatouille posed a number of unique challanges. First we had to adopt efficient methods for simulating a wide variety of material behaviours. Second we needed to direct our simulations in order to match the expressiveness of the character's animation, e.g. forming specific shapes while the character pounds a dough. Finally we had to apply shading to our simulated surfaces which underwent complex deformations and topological changes. In this sketch we will focus on materials ranging from elastoplastic solids to viscous liquids and illustrate with several shot examples from the film.
Available as Pixar Technical Memo #07-06
Alex Harvill
May 2007
An illustration of Gusteau comes to life and introduces a new facet of the film Ratatouille. To do this, a technique was needed to convert an animated 3D character into a 2D illustration. Existing renderman shaders based on normal and depth maps were difficult to control. Post processing per gprim id information robustly captures object outlines and has been previously extended with texture maps to feature some surface details. However, texture maps introduce artifacts that are quite objectionable during post processing. The id based approach is extended using scalar fields to produce well-behaved 2D edge/contour lines from 3D models. Specifically, mesh face ids and per-vertex line weights were rendered and processed to create lines and specify line thickness, respectively. This new method is fast, looks good, and is easy to control.
Available as Pixar Technical Memo #07-08
Geoffrey Irving, Craig Schroeder, Ronald Fedkiw
August 2007
We propose a numerical method for modeling highly deformable nonlinear incompressible solids that conserves the volume locally near each node in a finite element mesh. Our method works with arbitrary constitutive models, is applicable to both passive and active materials (e.g. muscles), and works with simple tetrahedra without the need for multiple quadrature points or stabilization techniques. Although simple linear tetrahedra typically suffer from locking when modeling incompressible materials, our method enforces incompressibility per node (in a one-ring), and we demonstrate that it is free from locking. We correct errors in volume without introducing oscillations by treating position and velocity in separate implicit solves. Finally, we propose a novel method for treating both object contact and self-contact as linear constraints during the incompressible solve, alleviating issues in enforcing multiple possibly conflicting constraints.
Available in the proceedings of SIGGRAPH 2007.
Geoffrey Irving, Eran Guendelman, Frank Losasso, Ronald Fedkiw
January 2006
We present a new method for the efficient simulation of large bodies of water, especially effective when three-dimensional surface effects are important. Similar to a traditional two-dimensional height field approach, most of the water volume is represented by tall cells which are assumed to have linear pressure profiles. In order to avoid the limitations typically associated with a height field approach, we simulation the entire top surface of the water volume with a state of the art, fully three-dimensional Navier-Stokes free surface solver. Our philosophy is to use the best available method near the interface (in the three-dimensional region) and to coarsen the mesh away from the interface for efficiency. We coarsen with tall, thin cells (as opposed to octrees or AMR), because they maintain good resolution horizontally allowing for accurate representation of bottom topography.
Additional materials: [movie.avi]
Available in the proceedings of SIGGRAPH 2006
Henrik Wann Jensen, Per H. Christensen
August 2007
Detailed descriptions of the ray-tracing and photon-mapping algorithms for rendering complex scenes with indirect illumination, caustics, participating media, and subsurface scattering. The emphasis is on the practical insight necessary to use and implement these algorithms in production of high-quality image in movies, games, architecture, etc.
Available as Siggraph 2007 course notes, Course Number 8
Pushkar Joshi, Mark Meyer, Tony DeRose, Brian Green, Tom Sanocki
May 2007
In this paper we consider the problem of creating and controlling volume deformations used to articulate characters for use in high-end applications such as computer generated feature films. We introduce a method we call harmonic coordinates that significantly improves upon existing volume deformation techniques. Our deformations are controlled using a topologically flexible structure, called a cage, that consists of a closed three dimensional mesh. The cage can optionally be augmented with additional interior vertices, edges, and faces to more precisely control the interior behavior of the deformation. We show that harmonic coordinates are generalized barycentric coordinates that can be extended to any dimension. Moreover, they are the first system of generalized barycentric coordinates that are non-negative even in strongly concave situations, and their magnitude falls off with distance as measured within the cage.
Additional materials: [HarmonicCoordinates.divx], [SiggraphSlides.pdf]
Available in the proceedings of Siggraph 2007.
Available as Pixar Technical Memo #06-02b
Other versions:
Michael Kass, Aaron Lefohn, John Owens
January 2006
Accurate computation of depth-of-field effects in computer graphics rendering is generally very time consuming, creating a problematic workflow for film authoring. The computation is particularly challenging because it depends on large-scale spatially-varying filtering that must accurately respect complex boundaries. Here we introduce an approximate depth-of-field computation that is good enough for film preview, yet can be computed interactively on a GPU. The method makes use of separable recursive filters to create efficient large-kernel convolutions. The individual recursive filters are derived from a minimum principle that produces spatially varying coefficients in the course of solving a tri-diagonal linear system. A straightforward GPU implementation of recursive filters would have poor performance, but using the well-established method of cyclic reduction, we are able to vectorize the computation and achieve interactive frame rates.
Additional materials: [movie.avi]
Available as Pixar Technical Memo #06-01
Michael Kass, John Anderson
January 2006
The Spacetime Constraints formulation attempts to marry the realism of physical simulation with the controllability of keyframe animation, but the resulting nonlinear optimization problems are generally extremely complicated and slow to solve. Here we explore the range of Spacetime Constraints problems that give rise to quadratic optimization functions solvable with linear systems of equations. We find that they generalize traditional splines to encompass oscillatory solutions. These problems can be solved at full frame rates, giving animators a keyframe animation tool with built in knowledge of a physical model. In addition to the splines themselves, we also introduce a new analysis method to extract oscillatory behavior from physical simulations in a way that can be connected naturally to the splines. It turns out that in order to have sufficient control of the frequency response of splines, we solve the Spacetime Constraints problems over the domain of complex numbers. As a consequence, our solutions have an imaginary part in addition to the real part. The imaginary part defines a phase angle that we show is very useful for controlling and generalizing oscillatory behavior whether extracted from simulation data or authored by hand.
Additional materials: [WigglySplines.mov]
Available as Pixar Technical Memo #06-06
Brian Knep, Craig Hayes, Rick Sayre, Tom Williams
June 1995
We present a system for animating an articulate figure using a physical skeleton, or armature, connected to a workstation. The skeleton is covered with sensors that monitor the orientations of the joints and send this information to the computer via custom-built hardware. The system is precise, fast, compact, and easy to use. It lets traditional stop-motion animators produce animation on a computer without requiring them to learn complex software. The working environment is very similar to the traditional environment but without the nuisances of lights, a camera, and delicate foam-latex skin. The resulting animation lacks the artifacts of stop-motion animation, the pops and jerkiness, and yet retains the intentional subtleties and hard stops that computer animation often lacks.
Appeared in the Proceedings of SIGCHI 1995
Sonoko Konishi, Michael Venturini
May 2007
It's difficult. In our society, rodents are usual portrayed in a negative manner. To overcome this, every department on Ratatouille strove to create appeal. In this sketch we will focus on how this was achieved through articulation.
Available as Pixar Technical Memo #07-12
Tom Lokovic, Eric Veach
August 2000
We introduce deep shadow maps, a technique that produces fast, high-quality shadows for primitives such as hair, fur, and smoke. Unlike traditional shadow maps, which store a single depth at each pixel, deep shadow maps store a representation of the fractional visibility through a pixel at all possible depths. Deep shadow maps have several advantages. First, they are prefiltered, which allows faster shadow lookups and much smaller memory footprints than regular shadow maps of similar quality. Second, they support shadows from partially transparent surfaces and volumetric objects such as fog. Third, they handle important cases of motion blur at no extra cost. The algorithm is simple to implement and can be added easily to existing renderers as an alternative to ordinary shadow maps.
Available in the Proceedings of SIGGRAPH 2000.
Mark Meyer, John Anderson
May 2007
Many applications in Computer Graphics contain computationally expensive calculations. These calculations are often performed at many points to produce a full solution, even though the subspace of reasonable solutions may be of a relatively low dimension. The calculation of facial articulation and rendering of scenes with global illumination are two example applications that require these sort of computations. In this paper, we present Key Point Subspace Acceleration and Soft Caching, a technique for accelerating these types of computations.
Key Point Subspace Acceleration (KPSA) is a statistical acceleration
scheme that uses examples to compute a statistical subspace and a set
of characteristic key points. The full calculation is then computed
only at these key points and these points are used to provide a
subspace based estimate of the entire calculation. The soft caching
process is an extension to the KPSA technique where the key points are
also used to provide a confidence estimate for the KPSA result. In
cases with high anticipated error the calculation will then ``fail
through'' to a full evaluation of all points (a cache miss), while
frames with low error can use the accelerated statistical evaluation
(a cache hit).
Additional materials: [SiggraphSlides.pdf], [softCaching.mov]
Available in the proceedings of Siggraph 2007
Available as Pixar Technical Memo #06-04b
Other versions:
Mark Meyer, John Anderson
January 2006
Global illumination provides important visual cues to an animation, however its computational expense limits its use in practice. In this paper, we present an easy to implement technique for accelerating the computation of indirect illumination for an animated sequence using stochastic ray tracing. We begin by computing a quick but noisy solution using a small number of sample rays at each sample location. The variation of these noisy solutions over time is then used to create a smooth basis. Finally, the noisy solutions are projected onto the smooth basis to produce the final solution. The resulting animation has greatly reduced spatial and temporal noise, and a computational cost roughly equivalent to the noisy, low sample computation.
Additional materials: [ShotRender.mov]
To appear in SIGGRAPH 2006
Available as Pixar Technical Memo #06-03
Darwyn Peachey
1990
Texture On Demand (TOD) is a technique for organizing large amounts of stored texture data in disk files and accessing it efficiently. Simply reading entire texture images into memory is not a good solution for real memory systems or for virtual memory systems. Texture data should be read from disk files only on demand. In the TOD technique, each texture image is stored as a sequence of fixed-size rectangular regions called tiles, rather than in the conventional raster scanline order. Tiles are an appropriate unit of texture data to read into memory on demand. As fixed-size units with good locality of reference in a variety of rendering schemes, tiles can be cached in main memory using the paging algorithms common in virtual memory systems. Good results have been obtained using an LRU tile replacement algorithm to select a tile to be deleted when main memory space is required.
Prefiltered textures are an important means of limiting bandwidth. TOD uses a set of prefiltered texture images called an r-set, a generalization of the texture pyramid (mip map). Texture filtering methods are reconsidered in terms of their performance in the TOD environment. Efficient filtering methods using the r-set are described.
The paper describes various implementations of TOD, including a virtual
memory implementation and a distributed implementation on a 16-processor
multicomputer.
Available as Pixar Technical Memo #217
Fabio Pellacini, Kiril Vidimce, Aaron Lefohn, Alex Mohr, Mark Leone, John Warren
August 2005
In computer cinematography, the process of lighting design involves placing and configuring lights to define the visual appearance of environments and to enhance story elements. This process is labor intensive and time consuming, primarily because lighting artists receive poor feedback from existing tools: interactive previews have very poor quality, while final-quality images often take hours to render.
This paper presents an interactive cinematic lighting system used in the production of computer-animated feature films containing environments of very high complexity, in which surface and light appearances are described using procedural RenderMan shaders. Our system provides lighting artists with high-quality previews at interactive framerates with only small approximations compared to the final rendered images. This is accomplished by combining numerical estimation of surface response, image-space caching, deferred shading, and the computational power of modern graphics hardware.
Our system has been successfully used in the production of two
feature-length animated films, dramatically accelerating
lighting tasks. In our experience interactivity fundamentally
changes an artist's workflow, improving both productivity and
artistic expressiveness.
Additional materials: [lpics_siggraph.mp4]
Available in the Proceedings of SIGGRAPH 2005
Lena Petrovic, Mark Henne, John Anderson
January 2005
Hair is one of the crucial elements in representing believable digital humans. It is one of the most challenging elements, too, due to the large number of hairs on a human head, their length, and their complex interactions. Hair appearance, in rendering and simulation, is dominated by collective properties, yet most of the current approaches model individual hairs. In this paper we build on the existing approaches to illumination and simulation by introducing a volumetric representation of hair which allows us to efficiently model collective properties of hair. We use this volumetric representation of hair to describe hair response to illumination, hair to hair collisions, and to subtly direct simulations of hair. Our method produces realistic results for different types of hair colors and styles and has been used in a production environment.
Additional materials: [clip1.mp4], [clip2.mp4], [clip3.mp4], [clip4.mp4], [clip5.mp4], [clip6.mp4]
Available as Pixar Technical Memo #06-08
Thomas Porter, Tom Duff
July 1984
Most computer graphics pictures have been computed all at once, so that the rendering program takes care of all computations relating to the overlap of objects. There are several applications, however, where elements must be rendered separately, relying on compositing techniques for the anti-aliased accumulation of the full image. This paper presents the case for four-channel pictures, demonstrating that a matte component can be computed similarly to the color channels. The paper discusses guidelines for the generation of elements and the arithmetic for their arbitrary compositing.
Available in the Proceedings of SIGGRAPH 1984.
Martin Reddy
August 2002
Siggraph 2002 course slides. See the abstracts in the additional materials below for more details.
Additional materials: [1-perception-abstract.txt], [2-perception.pdf], [3-terrain-abstract.txt], [4-terrain.pdf]
William T. Reeves, David H. Salesin, Robert L. Cook
July 1987
We present a solution to the aliasing problem for shadow algorithms that use depth maps. The solution is based on a new filtering technique called percentage closer filtering. In addition to antialiasing, the improved algorithm provides soft shadow boundaries that resemble penumbrae. We describe the new algorithm in detail, demonstrate the effects of its parameters, and analysze its performance.
Available in the Proceedings of SIGGRAPH 1987.
Andrew Riffel, Aaron Lefohn, Kiril Vidimce, Mark Leone, John Owens
August 2004
Real-time graphics hardware continues to offer improved resources for programmable vertex and fragment shaders. However, shader programmers continue to write shaders that require more resources than are available in the hardware. One way to virtualize the resources necessary to run complex shaders is to partition the shaders into multiple rendering passes. This problem, called the "Multi-Pass Partitioning Problem" (MPP), and a solution for the problem, Recursive Dominator Split (RDS), have been presented by Eric Chan et al. The O(n3) RDS algorithm and its heuristic-based O(n2) cousin, RDSh, are robust in that they can efficiently partition shaders for many architectures with varying resources. However, RDS's high runtime cost and inability to handle multiple outputs per pass make it less desirable for real-time use on today's latest graphics hardware. This paper redefines the MPP as a scheduling problem and uses scheduling algorithms that allow incremental resource estimation and pass computation in O(n log n) time. Our scheduling algorithm, Mio, is experimentally compared to RDS and shown to have better run-time scaling and produce comparable partitions for emerging hardware architectures.
Available in the Proceedings of Graphics Hardware 2004
David Ryu, Paul Kanyuk
May 2007
One of the major technical challenges in the animated film Ratatouille was creating a believable rat colony. In numbers as high as a thousand, these rats participated in highly complex and coordinated behaviors ranging from chaotic swarming to gourmet cooking. Often, the colony was the featured foreground element in the shot and thus subject to the strict motion standards of our hand- animated characters. We needed a crowds animation pipeline that had the flexibility to tackle a wide variety of complex behaviors, while also delivering motion that could pass muster alongside hero animation. We chose Massive Software's Massive Prime as our simulation engine. To ensure animation quality, our Massive pipeline preserved all the nuances of the input animation and allowed the animators to easily change the simulations, while still enabling TDs to use proceduralism to create complex behaviors. To this end we deployed a number of techniques including a system for baking our deformer based character rigs into point weights for Massive, motion retargeting tools for animation cycles, and procedural methods for swarming, leaping, carrying, and climbing, among others.
Available as Pixar Technical Memo #07-02
David Ryu
May 2007
Featuring plush rats, well-groomed humans, and a colony of rodents numbering a thousand strong, Ratatouille had shots where the original scene descriptions contained many hundreds of millions of hairs. To make these shots renderable, we developed many new technologies to optimize our RenderMan-based hair rendering pipeline, including caching to speed up runtime sculpting, a technique for optimizing bounding boxes for RiProcedurals, and a smooth level-of-detail system incorporating depth-of- eld and motion blur. These techniques resulted in an order of magnitude reduction in hair rendering time and memory compared to the already optimized pipeline used previously in Pixar films.
Available as Pixar Technical Memo #07-09
Apurva Shah, Justin Ritter, Chris King, Stefan Gronsky
May 2007
In Pixar's Ratatouille a lot of scenes take place inside the kitchen where reflective surfaces like counter tops, stoves, pots and pans abound. Furthermore, these surfaces were often burnished or covered with dents, scratches or other displacements, which meant that the reflections were soft and fuzzy. Physically accurate reflections are most often achieved by tracing reflected rays into the scene. When the ray encounters another object, computationally expensive lighting and shading calculations must be performed to determine the contribution of the reflecting point. Paradoxically, surfaces that have soft or fuzzy reflections are more expensive since they have a larger reflection cone angle requiring more rays to adequately sample the reflected scene. We present a technique that utilizes radiance caches to significantly speed up the reflection calculations and discuss some of the accuracy trade-offs inherent to this approach.
Available as Pixar Technical Memo #07-04
Chen Shen, Apurva Shah
May 2007
From pouring sauces to sudsy sink water to violent sewer rapids, realistic animation of fluids presented interesting challenges in Ratatouille. The various fluid effects were simulated either using a physically-based solver or directly with generic particle systems. Although the simulated particles move as a whole like a fluid, the number of particles was too small to give the appearance of a continuous surface if rendered directly. To address this, we developed a technique to efficiently extract temporally and spatially coherent surfaces from particles with parametrization that allows textural details to be later added in rendering.
Available as Pixar Technical Memo #07-05
Christine Waggoner, David Baraff
May 2007
He's fat. Enormously so. Essentially a clothed sphere. (A very deformable and complexly animated sphere, with legs, arms, and no neck.) For Ratatouille, creating dynamic costumes for every human character in the film required extensive development and innovation in Pixar's cloth pipeline, particularly for the character Gusteau. Modeling and shepherding high-quality simulated costumes through the production of an animated feature is a feat in itself, let alone for a character with such extreme proportions who dramatically squashes, stretches, floats and bursts through the air. We describe the development of several new modeling and simulation techniques required for Gusteau's costume in Ratatouille. These techniques were required to deal with conflicting goals of rendering versus simulation meshes, deliver improved simulate cloth behavior, resolve impossible collision situations, and achieve director-mandated shapes on portions of Gusteau's costume.
Available as Pixar Technical Memo #07-07