CHICAGO—SIGGRAPH 2021 has announced 191 selected research projects from 22 countries as part of both its Technical and Art Papers programs this summer. The state-of-the-art research projects cover the many subfields of computer graphics and interactive techniques.
The 48th annual conference’s live events will run 9–13 August, with on-demand talks from researchers available 2 August–29 October.
For its 2021 selections, the Technical Papers program reviewed more than 440 submissions, with additional projects pulled from peer-reviewed journal ACM Transactions on Graphics (TOG). Members of the review committee chose each project through a double-blind process. Two major themes of the incredibly strong research being presented are applications of artificial intelligence and a focus on fundamentals.
“It has been an honor to oversee this year’s programming,” shared SIGGRAPH 2021 Technical Papers Chair Sylvain Paris, of Adobe Research. “After a trying year, it’s a thrill to see that the amazing work of researchers remains strong and to showcase some truly groundbreaking applications of computer science.”
Alongside research from Princeton University, Adobe, Google, MIT, NVIDIA, Yale University, and Autodesk, highlights from the 2021 Technical Papers program include:
Mixture of Volumetric Primitives for Efficient Neural Rendering
Authors: Stephen Lombardi, Tomas Simon, Gabriel Schwartz, Michael Zollhoefer, Yaser Sheikh, and Jason Saragih, Facebook Reality Labs
Mixture of Volumetric Primitives (MVP) represents dynamic 3D scenes using a collection of moving volumetric primitives that can be learned from multi-view video data and rendered in real time.
SP-GAN: Sphere-guided 3D Shape Generation and Manipulation
Authors: Ruihui Li, Xianzhi Li, Ka-Hei Hui, and Chi-Wing Fu, The Chinese University of Hong Kong
SP-GAN is a new unsupervised generative model, adopting a "sphere prior" to enable high-quality 3D shape (point clouds) synthesis and structure-aware shape manipulation through an implicitly embedded dense correspondence.
StyleCariGAN: Caricature Generation via StyleGAN Feature Map Modulation
Authors: Wonjong Jang, Gwangjin Ju, Yucheol Jung, and Seungyong Lee, POSTECH; Jiaolong Yang and Xin Tong, Microsoft Research Asia
This paper presents a caricature generation framework based on StyleGAN, which automatically creates a realistic and detailed caricature from an input photo with optional controls on shape exaggeration and color stylization.
SWAGAN: A Style-based Wavelet-driven Generative Model
Authors: Rinon Gal, Tel Aviv University and WAY2VAT; Dana Cohen Hochberg, Amit Bermano, and Daniel Cohen-Or, Tel Aviv University
This paper presents SWAGAN, a style- and wavelet-based generative adversarial network designed to side-step the spectral bias of neural networks and alleviate the recently identified high-frequency deficiency in image synthesis models.
Solid-fluid Interaction With Surface-tension-dominant Contact
Authors: Liangwang Ruan and Baoquan Chen, Peking University and Beijing Film Academy; Jinyuan Liu and Bo Zhu, Dartmouth College; Shinjiro Sueda, Texas A&M University; and, Bin Wang, Beijing Film Academy
The paper proposes a novel, three-way coupling framework to simulate the surface-tension-dominant contact between rigid and fluid, which uses a Lagrangian surface membrane to handle the interactions between solids and fluid.
For even more highlights, check out the recently released Technical Papers Preview on YouTube: https://youtu.be/Ros7ZXqLbFg.
Out of more than 90 submissions, the 2021 Art Papers jury has selected 14 projects for this summer’s conference — 10 long papers and four short papers. The chosen works will be featured in the first-ever SIGGRAPH special issue of the Proceedings of the ACM in Computer Graphics and Interactive Techniques, or PACMCGIT, and encompass the full scope of digital media arts from varying perspectives. Highlights include:
BEST IN SHOW
Bad Mother / Good Mother: The Poetics and Politics of the Sounds of Invisible Labor
Author: Annina Rüst, Florida Atlantic University
“Bad Mother / Good Mother” is an audiovisual performance about motherhood and the workplace. It explores societal perceptions about ideal and non-ideal motherhood through sound and projected visuals. The paper contributes to an existing matrilineal tradition within audiovisual performance and electronic sound.
“Can the Subaltern Speak?”: Feminism in Robotic Mask Design
Author: Behnaz Farahi, California State University, Long Beach
Inspired by Iranian historical masks used to protect women from the gaze of colonial patriarchal oppressor, this project suggests strategies of resistance for women against contemporary patriarchal oppression. In the work, two masks begin to communicate with each other, blinking their eyelashes in rapid succession using AI-generated Morse code.
“The artists and research our jury has selected for 2021 deserve this platform to exhibit their thought-provoking, important work. It’s exciting to see art and science push the boundaries of creative expression, and the projects you’ll discover employ everything from machine learning to indigenous media arts,” said SIGGRAPH 2021 Art Papers Chair Daria Tsoupikova, of University of Illinois at Chicago.
All research during the 2021 virtual conference will be more integrated than ever, with multiple sessions — including the popular Technical Papers Fast Forward event — featuring projects from both programs. Access to SIGGRAPH 2021 Technical and Art Papers programming is available at varying registration levels.
Image at top: “Solid-fluid Interaction With Surface-tension-dominant Contact” © 2021 Peking University, Beijing Film Academy, Dartmouth College, and Texas A&M University