Graphics4Science: Computer Graphics for Scientific Impacts DOI: XXXXXXX.XXXXXXX Conference: Special Interest Group on Computer Graphics and Interactive Techniques Conference; August 10-14, 2025; Vancouver, BC, Canada CCS: Computing methodologies Computer graphics

Peter Yichen Chen, Minghao Guo, Hanspeter Pfister, Ming Lin, William Freeman, Qixing Huang, Han-Wei Shen, Wojciech Matusik · 2025

Read Graphics4Science: Computer Graphics for Scientific Impacts DOI: XXXXXXX.XXXXXXX Conference: Special Interest Group on Computer Graphics and Interactive Techniques Conference; August 10-14, 2025; Vancouver, BC, Canada CCS: Computing methodologies Computer graphics by Peter Yichen Chen, Minghao Guo, Hanspeter Pfister, Ming Lin, William Freeman, Qixing Huang, Han-Wei Shen, Wojciech Matusik (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-15786.txt.

Introduction

Peter Chen and co-authors present a course on how computer graphics serves science. Their 2025 course runs at a major graphics meeting in Vancouver. Graphics is famed for film and games, yet it has long aided medical imaging and simulation.

Core graphics tools, such as geometric reasoning and physical modeling, act as built-in priors for science problems. Those priors help most where data are scarce. The course traces past wins, live uses, and open questions in the shared space. It also works to bridge word gaps between the two fields. Newcomers and experts alike are invited to take on high-impact science tasks.

Catalog record

CREATORPeter Yichen Chen, Minghao Guo, Hanspeter Pfister, Ming Lin, William Freeman, Qixing Huang, Han-Wei Shen, Wojciech Matusik
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE11 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-15786

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Graphics4Science: Computer Graphics for Scientific Impacts DOI: XXXXXXX.XXXXXXX Conference: Special Interest Group on Computer Graphics and Interactive Techniques Conference; August 10-14, 2025; Vancouver, BC, Canada CCS: Computing methodologies Computer graphics Peter Yichen Chen, Minghao Guo, Hanspeter Pfister, Ming Lin, William Freeman, Qixing Huang, Han-Wei Shen, Wojciech Matusik Abstract . Computer graphics, often associated with films, games, and visual effects, has long been a powerful tool for addressing scientific challenges—from its origins in 3D visualization for medical imaging to its role in modern computational modeling and simulation. This course explores the deep and evolving relationship between computer graphics and science, highlighting past achievements, ongoing contributions, and open questions that remain. We show how core methods, such as geometric reasoning and physical modeling, provide inductive biases that help address challenges in both fields, especially in data-scarce settings. To that end, we aim to reframe graphics as a modeling language for science by bridging vocabulary gaps between the two communities. Designed for both newcomers and experts, Graphics4Science invites the graphics community to engage with science, tackle high-impact problems where graphics expertise can make a difference, and contribute to the future of scientific discovery. Additional details are available on the course website: https://graphics4science.github.io . Figure 1. Computer graphics have evolved from a tool for visualization into a driving force behind scientific discovery, shaping advancements in biology, physics, and beyond. This course explores how graphics techniques have supported interdisciplinary research and opened up new possibilities across scientific domains. Teaser Figure: Graphics4Science 1. Introduction Various computer graphics subfields for solving scientific problems. Figure 2. Graphics4Science. Various computer graphics subfields play important roles in solving scientific problems. From left to right: Visualization enables the exploration of nanoscale astroglial cells ( Mohammed et al. 2017 ) ; Imaging helps unveil the first-ever image of a black hole ( Akiyama et al. 2019 ) ; Geometry provides representations for biomechanics ( Xia et al. 2025 ) ; Simulation allows doctors to interact with virtual human organs ( Yang and Lin 2015 ) ; and Design empowers the manufacturing of artificial hearts ( Buchner et al. 2023 ) . Computer graphics has a strong tradition in model-based modeling. It relies on partial differential equations, rendering equations, and geometric representations to represent and simulate physical phenomena. These structures embed physical priors that serve as powerful inductive biases, supporting generalization and interpretability in computational models. While graphics is often associated with entertainment—such as movies, games, and virtual environments—its core methods are broadly applicable to scientific problems. From medical imaging to structural biology, graphics techniques have made substantial, if sometimes underrecognized, contributions. Both graphics and science increasingly face shared challenges: limited real-world data, high costs of acquisition, and the need for accurate simulations. In the context of machine learning, addressing these challenges requires incorporating strong inductive biases—an area where graphics’ long-standing focus on efficient, physically grounded modeling is especially valuable. This paper introduces Graphics4Science , an initiative aimed at foregrounding both the historical and emerging role of computer graphics in scientific discovery. Our goals are twofold: to revisit past intersections between graphics and science, and to explore new opportunities where graphics can contribute to solving pressing scientific problems. A key barrier to deeper collaboration is the mismatch in language across disciplines. Researchers often describe similar concepts in field-specific terms—for instance, “collision” in physics versus “clashing” in protein folding. Yet these problems frequently share common computational structures: geometry, constraints, dynamics, and nonphysical states. Without a shared vocabulary, such parallels can be overlooked, and opportunities for collaboration missed. Graphics4Science seeks to bridge this gap. By translating scientific questions into forms familiar to the graphics community, identifying benchmarkable tasks, and showcasing examples of cross-disciplinary impact, the initiative aims to expand the scope and relevance of graphics research within science. 2. Course Format 1. Introduction (10 mins) – Peter Yichen Chen, Minghao Guo 2. Graphics for Science: Techniques & Applications (2 h 5 min) Each talk: 20 min presentation + 5 min Q&A • Visualization for Science – Hanspeter Pfister • Simulation for Science – Ming Lin • Imaging for Science – Bill Freeman • Geometry for Science – Qixing Huang • Design for Science – Wojciech Matusik 3. Panel Discussion with NSF (45 min) Panelists: Han-Wei Shen and selected session speakers 3. Description of Course This course aims to highlight how computer graphics techniques can extend beyond traditional entertainment applications and play a growing role in scientific research. By highlighting how graphics techniques contribute to areas such as data visualization, imaging, geometric modeling, simulation, and computational design (see Fig. 2 ), the course encourages researchers and practitioners to explore new interdisciplinary opportunities. The goal is to encourage a broader view of graphics as a tool that can contribute to addressing real-world scientific questions – from accelerating materials discovery to improving medical imaging and advancing physics-based simulations. Through this, we hope to inspire the community to bridge the gap between graphics and scientific advancements. Rather than centering on specific technical tools, this course provides a broad vision of how graphics can drive scientific discovery. A key focus is identifying open scientific questions that the graphics community can help solve. Each session will highlight key challenges, emerging trends, and impactful applications across different scientific domains. Attendees will gain a deeper understanding of not just how graphics techniques are applied to science, but why they matter and where they could lead in the future. The course will conclude with a panel discussion, encouraging cross-disciplinary dialogue and outlining future research directions, funding opportunities, and collaborative possibilities between graphics and sciences. Visualization turns unwieldy, multi-modal datasets into actionable insight. A landmark example is Foldit ( Cooper et al. 2010 ) , whose intuitive 3D interface enabled thousands of citizen scientists to manipulate protein structures and solve problems that had stumped automated algorithms. Building on such successes, this session surveys techniques that enable feature extraction and dimensionality reduction, especially for biological and medical data. We outline interactive, scalable systems that reveal hidden patterns and propose rigorous evaluation methods, showing how thoughtful visual analytics can shorten the path from data to discovery. Imaging techniques that fuse physics-based forward models with data-driven priors now let scientists extract meaning from data once thought irrecoverable, from imaging the central supermassive black hole and tracking galactic dynamics ( Akiyama et al. 2019 ) , to magnifying intracellular motion in live cells and inferring missing observations in remote-sensing streams. This session surveys super-resolution and data-synthesis methods, comparing trade-offs in resolution, acquisition speed, and interpretability. Case studies across astronomy, microscopy, and environmental monitoring illustrate how filling in the data with informed priors accelerates discovery and opens new questions for graphics research. Geometry lies at the heart of many scientific problems, providing the foundation for both representing and reasoning about physical structures. Scientific domains rely on a wide range of geometric representations—from molecular conformations and crystal lattices to protein surfaces and porous materials. In this module, we explore how geometric modeling techniques, including meshes, implicit surfaces, and neural signed distance fields (SDFs), are used to capture and manipulate complex structures across scales. We cover both synthesis (e.g., generative models for molecular structures) and analysis (e.g., structural inference from sparse data), highlighting how modern graphics techniques enable more flexible, accurate, and

Read the complete text