Top conference SIGGRAPH 2024 announces best paper results: NVIDIA and CMU each account for 40%, with three domestic teams receiving honorary nominations

Today, ACM SIGGRAPH released the results. This year, 5 best papers and 12 honorable mentions were selected, and the Test of Time Award was awarded to 4 papers published in 2012 and 2013, continuing last year's tradition. Domestic institutions such as ShanghaiTech University, Huazhong University of Science and Technology, and the Chinese University of Hong Kong were on the list.

SIGGRAPH stands for Special Interest Group on Computer Graphics and is a branch of ACM.

Advertisement

As the top conference in the field of computer graphics and interactive technologies, SIGGRAPH is an important stage for computer graphics enthusiasts around the world to unleash their creativity.

The ideas, visions, and technologies presented at SIGGRAPH often go beyond our current imagination and will hopefully become benchmarks 5, 10, or even 15 years from now.

This year's SIGGRAPH will be held from July 28 to August 1 in Denver, Colorado, USA.

This is the first time the conference has returned to Colorado since its founding in 1974, returning to its original location, but the rapid development of technology is no longer the same.

Advertisement

SIGGRAPH 2024 continues the tradition of presenting Best Paper Awards and Honorable Mentions to papers selected for the prominence and cutting-edge nature of their research.

Best Paper

Paper 1: From Microfacets to Participating Media: A Unified Theory of Light Transport With Stochastic Geometry

Authors: Dario Seyb, Eugene d'Eon, Benedikt Bitterli, Wojciech Jarosz

Organization: Dartmouth College, NVIDIA

Paper address: pdf / stochastic-implicit-light-transport.pdf

In this work, the authors derive a theory of light transport on random implicit surfaces, a geometric model capable of expressing deterministic geometry, microsurfaces, participating media, and an exciting new continuum in between that includes aggregate appearances, nonclassical media, etc. The model intrinsically supports spatial correlations that are lacking in most existing random models.

Paper 2: Walkin' Robin: Walk on Stars With Robin Boundary Conditions

Authors: Bailey Miller, Rohan Sawhney, Keenan Crane, Ioannis Gkioulekas

Institution: CMU, NVIDIA

Paper address: labs / prl / publication / miller2024wost/

In this work, the authors develop a meshless Monte Carlo method for solving boundary value problems such as the Poisson equation with Dirichlet, Neumann, and Robin boundary conditions. Unlike traditional PDE solvers, the new method does not require a volume mesh or a global solver. The method is robust, parallelizable, scalable to complex geometries, and can be evaluated based on views.

Paper 3: Repulsive Shells

Authors: Josua Sassen, Henrik Schumacher, Martin Rumpf, Keenan Crane

Institutions: Ecole Normale Supérieure Paris-Saclay, University of Georgia, University of Bonn, CMU

Shape spaces are powerful tools for nonlinear interpolation, extrapolation, and averaging of geometric data, but previous shape spaces allow geometries to self-intersect in unphysical ways. In this work, the authors introduce a shape space where geometries naturally avoid intersections, and an adaptive collision potential that prevents collisions while refinement converges.

Paper 4: Lightning-Fast Method of Fundamental Solutions

Author: Jiong Chen, Florian Schäfer, Mathieu Desbrun

Institutions: French National Institute for Computer Science and Automation, Georgia Institute of Technology, École Polytechnique

Paper address: pubs / CSD24.pdf

In this work, we introduce a variational preconditioner based on inverse Cholesky factorization to improve the solution efficiency of dense systems discretized by boundary integral equations, thereby effectively solving the scalability problem commonly seen in boundary-based methods.

Paper 5: Robust Containment Queries Over Collections of Rational Parametric Curves via Generalized Winding Numbers

By Jacob Spainhour, David Gunderman, Kenneth Weiss

Institutions: University of Colorado at Boulder, Indiana University School of Medicine, Lawrence Livermore National Laboratory

Paper address: abs / 2403.17371

In this work, we extend the theory of generalized winding numbers to unstructured collections of rational parametric curves via a numerically stable algorithm, enabling robust and accurate containment classification of non-watertight and self-intersecting shapes at arbitrary locations.

Honorable Mentions

Paper 1: Solid Knitting

Authors: Yuichi Hirose, Mark Gillespie, Angelica M. Bonilla Fominaya, James McCann

Paper 2: PEA-PODs: Perceptual Evaluation of Algorithms for Power Optimization in XR Displays

Authors: Kenneth Chen, Thomas Wan, Nathan Matsuda, Ajit Ninan, Alexandre Chapiro, Qi Sun

Institution: New York University, Meta

Paper address: Che24 / che24.pdf
Project homepage: projects/sig24/ index.html

Displays are the main source of battery consumption in wireless XR devices. Energy-efficient rendering can reduce power consumption but at the expense of visual quality. In this work, the authors investigate popular VR display and rendering architectures through electronic measurement prototypes and large-scale perceptual studies. The milliwatt-to-quality model guides battery-aware rendering, display hardware design, and eye tracking decisions.

Paper 3: CLAY: A Controllable Large-scale Generative Model for Creating High-quality 3D Assets

Authors: Longwen Zhang, Ziyu Wang, Qixuan Zhang, Qiwei Qiu, Anqi Pang, Haoran Jiang, Wei Yang, Lan Xu, Jingyi Yu

Institutions: ShanghaiTech University, Yingmo Technology, Huazhong University of Science and Technology

Project homepage: CLAY-3D / OpenCLAY

In this work, the authors propose a new 3D native Diffusion Transformer generative large model CLAY, designed to easily transform imagination into 3D models. CLAY consists of a model with 1.5 billion parameters and excels in creating high-quality, realistic 3D assets, allowing experts and novices alike to reignite creativity and turn vibrant ideas into reality.

Paper 4: DressCode: Autoregressively Sewing and Generating Garments From Text Guidance

Authors: Kai He, Kaixin Yao, Qixuan Zhang, Jingyi Yu, Lingjie Liu, Lan Xu

Institutions: ShanghaiTech University, Yingmo Technology, University of Pennsylvania

Paper address: abs / 2401.16465
Project homepage: view/projectpage-dresscode

In this work, the authors present DressCode, a generative AI framework designed specifically for 3D clothing. It leverages the power of natural language, integrates SewingGPT for sewing pattern generation and a fine-tuned diffusion model to synthesize PBR textures, and demonstrates interactive and friendly applications for clothing generation, completion, and editing.

Paper 5: Bilateral Guided Radiance Field Processing

Author: Yuehao Wang, Chaoyi Wang, Bingchen Gong, Tianfan Xue

Institution: The Chinese University of Hong Kong

Paper address: abs / 2406.00448
Project homepage:https://bilarfpro.github.io/

In this work, for a set of multi-view images with photometric variations, the authors reconstruct high-quality radiance fields without “floating objects” by disentangling inconsistent camera processing between different views. In addition, the authors propose a radiance modification method that can upscale user-provided 2D modifications to 3D, achieving consistent and stunning rendering of the entire scene.

Paper 6: Fabric Tessellation: Realizing Freeform Surfaces by Smocking

Authors: Aviv Segall, Jing Ren, Amir Vaxman, Olga Sorkine-Hornung

Institutions: ETH Zurich, University of Edinburgh

Paper address: projects / 3dsmocking / 3dsmocking_paper.pdf

In this work, the authors propose a new method to realize free-form surfaces with flat fabrics, by using a technique called “stitching” to stitch together points on the fabric to generate curvature. Given a target surface and a desired stitching pattern, the corresponding two-dimensional stitching pattern can be output, and the fabric can be made by stitching together the specified points. The produced textile approximates the target shape and exhibits pleasing wrinkles.

Paper 7: Capacitive ToUh Sensing on General 3D Surfaces

Authors: Gianpaolo Palma, Narges Pourjafarian, Jürgen Steimle, Paolo Cignoni

Paper 8: SMERF: Streamable Memory Efficient Radiance Fields for Real-time Large-scene Exploration

Authors: Daniel Duckworth, Peter Hedman, Christian Reiser, Peter Zhizhin, Jean-François Thibert, Mario Lučić, Richard Szeliski, Jonathan T. Barron

Paper 9: Spin-It Faster: Quadrics Solve All Topology Optimization Problems That Depend Only on Mass Moments

Authors: Christian Hafner, Mickaël Ly, Chris Wojtan

Paper 10: Ray Tracing Harmonic Functions

By: Mark Gillespie, Denise Yang, Mario Botsch, Keenan Crane

Institution: CMU, Technical University Dortmund

Paper address: Research / harnack-tracing / HarnackTracing.pdf

In this work, the authors introduce a method similar to sphere tracing, Harnack tracing, for a completely different class of surfaces encoded by harmonic functions, opening up new possibilities for visual computing. The method takes larger steps than simple ray marching, avoids the numerical problems common to general root finding methods, and like sphere tracing, only requires a point-by-point evaluation of the function at each step.

Paper 11: Seamless Parametrization in Penner Coordinates

By Ryan Capouellez, Denis Zorin

Paper 12: Theory of Human Tetrachromatic Color Experience and Printing

Authors: Jessica Lee, Nicholas Jennings, Varun Srivastava, Ren Ng

Test of Time Award

This year marks the second year of the ACM SIGGRAPH Test-of-Time Award, which covers papers presented at the SIGGRAPH conference from 2012 to 2014.

There are four winning papers in total, which have had a profound impact on computer graphics and interactive technology in the past decade.

Paper 1: 3D Shape Regression for Real-Time Facial Animation (2013)

Author: Chen Cao, Yanlin Weng, Stephen Lin, Kun Zhou

Institution: Zhejiang University, Microsoft Research Asia

Reason for the award: In this work, the authors introduce a groundbreaking method for real-time, accurate 3D face tracking and performance capture using a monocular RGB camera, paving a path for creating realistic facial animation on mobile devices.

Paper address: 2013 / vface.pdf

Paper 2: Femto-Photography: Capturing and Visualizing the Propagation of Light (2013)

Authors: Andreas Velten, Di Wu, Adrián Jarabo, Belén Masiá, Christopher Barsi, Chinmaya Joshi, Everett Lawson, Moungi Bawendi, Diego Gutierrez, Ramesh Raskar

Institution: MIT, University of Zaragoza

Citation: In this work, the authors introduce a system for capturing the transport of light. It enables slow-motion videos of well-known phenomena such as reflection and scattering, creating unprecedented visualizations of core concepts in computer graphics.

Paper address: bitstream / handle / 1721.1/82039 / Bawendi_Femto-photography.pdf?sequence=1&isAllowed=y

Paper 3: RobuInside-Outside Segmentation Using Generalized Winding Numbers (2013)

Authors: Alec Jacobson, Ladislav Kavan, Olga Sorkine-Hornung

Institution: ETH Zurich

Citation: In this work, the authors defined a robust operator for identifying the interior volume of a 3D mesh. It has become a key element of the standard toolbox for manipulating meshes, such as Boolean operations between meshes and tetrahedralization algorithms.

Paper address: /jacobson13robust/ jacobson13robust.html

Paper 4: How do Humans Sketch Objects? (2012)

Author: Mathias Eitz, James Hays, Marc Alexa

Institution: Technical University of Berlin, Brown University

Citation: In this work, the authors introduce a first-of-its-kind database of 20,000 object sketches and demonstrate how to use it. The database has been instrumental in advancing research in a wide range of areas including sketch generation, cleaning, recognition, and retrieval.

Paper address: eitz / pdf / 2012_siggraph_classifysketch.pdf

References:

  • https://blog.siggraph.org/2024/06/siggraph-2024-technical-papers-awards-best-papers-honorable-mentions-and-test-of-time.

Advertising Statement: The external jump links contained in the article (including but not limited to hyperlinks, QR codes, passwords, etc.) are used to convey more information and save selection time. The results are for reference only. All articles in Gamingdeputy include this statement.

Advertisement