Technical Research Library
Open Source Projects
Technology News

Technical Research Library

2015

DWA Technical Memo 2015-348, 2015

Improved Deep Image Compositing Using Subpixel Masks

Jonathan Egstad, Mark Davis, Dylan Lacewell

(Slides from DigiPro 2015 presentation)

We present an improved method of producing and manipulating deep pixel data which retains important surface information calculated during the rendering algorithm for later use during compositing, allowing operations normally performed in the renderer to be deferred until compositing. These include pixelcoverage calculation, pixel filtering, hard-surface blending, and matte object handling. Current methodologies for representing and transmitting deep pixel data work well for combining volumetric and hard-surface renders but are not very successful at combining hard-surfaces. By retaining additional surface information a renderer’s final integration steps can be reconstructed later in compositing.

2014

Book

MULTITHREADING FOR VISUAL EFFECTS

Martin Watt, Ronald D. Henderson et. al.

In this book, developers from DreamWorks Animation and other companies provide practical advice on multithreading techniques and visual effects used in popular visual effects libraries, tools and proprietary animation systems. Concrete examples show how multithreading paradigms are applied to achieve high performance computation for visual effects in the challenging world of production environments, and how problems often encountered in visual effects are addressed. The book presents an overview of practical uses for the Intel TBB library as well as other multithreading constructs.

DWA Technical Memo 2014-320, 2014

Fast and Robust Continuous Collision Detection (fastCCD)

Suejung Huh, Young Ju Lee, Galen Gornowicz, Ronald D. Henderson

We present a fast and robust narrow-phase continuous collision detection (CCD) method. Our algorithm is based on the method originally proposed by [Provot 1997]. However, the traditional method has been reformulated analytically so as to obtain more robust collision detection results than other previous CCD methods available, while surpassing the performance of the traditional cubic solverbased methods. The novelties of our method are twofold. First, a new area inclusion test is introduced and analyzed to prove it does not miss a collision, proving our method is robust in terms of false negatives. Second, we provide a dimensional reduction technique to handle degenerate cases exhaustively, thereby enabling it to handle such challenging cases without incurring unnecessary computational overhead.

2013

DWA Technical Memo 2013-001, 2013

DEFORMATION AFFECTOR FRAME

Thanh Giang, Peter Farson, Paul DiLorenzo, Alex Powell

We introduce an improved method that aids in the refinement of smooth deformations for animation. Through the use of novel affector frames of subD affector surfaces, our method preserves desired hard edges around extremeties while accounting for volume loss and possible crunching artifacts which are usually present from SSD deformation techniques.

2012

Talk, Siggraph 2012

ISHAIR: IMPORTANCE SAMPLING FOR HAIR SCATTERING

Jiawei Ou, Feng Xie, Parashar Krishnamachari, Fabio Pellacini

This importance sampling method for the bidirectional scattering distribution function (bsdf) of hair provides an algorithm for hair scattering that is effective at reducing noise, simple to implement, and efficient to evaluate.

Talk, Siggraph 2012

EFFICIENT AND SEAMLESS VOLUMETRIC FRACTURING

Mihai Alden, Gustav Melich, Ken Museth

We present a novel framework for seamless volumetric fracturing that offers significant advantages over existing methods. Our approach is based on new memory-efficient data structure s and fast parallel algorithms.

This importance sampling method for the bidirectional scattering distribution function (bsdf) of hair provides an algorithm for hair scattering that is effective at reducing noise, s imple to implement, and efficient to evaluate.

Talk, DigiPro 2012

LIBEE: A MULTITHREADED DEPENDENCY GRAPH FOR CHARACTER ANIMATION

Martin Watt, Mark Hampton

This paper presents a new dependency graph evaluation engine called LibEE for the Dreamworks Animation next generation in-house animation tool. LibEE is designed to be very heavily m ultithreaded thus able to deliver extremely high performance to animators. The paper discusses the motivations, design choices and implementation of the graph engine and more.

Talk, Siggraph 2012

HERO QUALITY CROWDS IN MADAGASCAR 3: EUROPE'S MOST WANTED

Nathaniel Dirksen, Justin Fischer, Jung-Hyun Kim, Kevin Vassey, Rob Vogt

We discuss techniques used to achieve hero-quality deformations and animated performances for crowd characters in Madagascar 3: Europe's Most Wanted, while minimizing disk usage, net work bandwidth consumption, and workload for animators.

Talk, Siggraph 2012

POINT-BASED GLOBAL ILLUMINATION

Eric Tabellion

We introduce an improved Point-Based Global Illumination (PBGI) cut picking algorithm which uses directional importance mapping that drastically improves quality and render time when solving difficult illumination effects.

Talk, Siggraph 2012

AMORPHOUS: AN OPENGL SPARSE VOLUME RENDERER

Mark J. Matthews

This GPU volume renderer is optimized for sparse volumes and gives physically accurate results and interactive performance. It reduces the need for offline renders by tightening the loop between volumetric modeling, simulation, and visual feedback.

Talk, Siggraph 2012

CLOUD MODELING AND RENDERING FOR PUSS IN BOOTS

Brett Miller, Ken Museth, Devon Penney, Nafees Bin Zafar

These state of the art techniques describe methods for modeling clouds and efficiently rendering them to create very large cloudscape effects.

Talk, Siggraph 2012

MAGIC BEANSTALK RIDE IN PUSS IN BOOTS

Amaury Aubel, Nikita Pavlov

This paper presents the creative and technical challenges of developing the custom toolset used to grow the giant magic beanstalk in the film Puss in Boots.

Talk, Siggraph 2012

VORTEX OF AWESOMENESS

Can Yuksel, Domin Lee, Ronald D. Henderson

In this paper, we present an overview of techniques used to control simulated hero elements, run fast fluid simulations at high-resolutions (100-200 million voxels), and respond quickly to art direction using curve-based volume warping to create a massive storm cloud and tornado for the film Puss in Boots.

2010

DWA Technical Memo 2010-362, 2010

Deep Images and Cutouts

Janne Kontkanen

This document describes the basic compositing operations possible for deep image data format. We explain the idea of continuous alpha blending, and present formulas for computing cutout images that can be accurately composited by an ADD operation in the post-process compositing stage. The last section also talks about direct generation of cutout images from the renderer.

Copyright 2010,2011,2012,2013,2014,2015,2016 DreamWorks Animation LLC. All rights reserved.