arXivSub Start free trial

SIGGRAPH 2025 Papers — Page 4

ACM SIGGRAPH (Transactions on Graphics) · 306 papers

ViSA: Physics-based Virtual Stunt Actors for Ballistic Stunts

Minseok Kim, Jungdam Won

GenerationReinforcement LearningPhysics Related

🎯 What it does: Developed an interactive animation system called ViSA for generating ballistic stunt actions commonly seen in movies and television series.

VR-Doh: Hands-on 3D Modeling in Virtual Reality

Zhaofeng Luo (Carnegie Mellon University), Minchen Li (Carnegie Mellon University)

Computational EfficiencyRobotic IntelligenceGaussian SplattingSimultaneous Localization and MappingOptical FlowPoint CloudMesh

🎯 What it does: Developed VR-Doh, a system that realizes real-time physical simulation and 3D model editing through hand interaction in virtual reality.

What is HDR? Perceptual Impact of Luminance and Contrast in Immersive Displays

Kenneth Chen, Alexandre Chapiro

Video

🎯 What it does: Study the impact of peak brightness and contrast of HDR display in virtual reality environments, build a high-precision haploscope prototype, collect VR-related HDR video datasets, implement custom tone mapping, and collect data through subjective preference experiments to fit the model.

When Gaussian Meets Surfel: Ultra-fast High-fidelity Radiance Field Rendering

Keyang Ye (Zhejiang University), Kun Zhou (Zhejiang University)

GenerationData SynthesisComputational EfficiencyNeural Radiance FieldGaussian SplattingOptical FlowImagePoint Cloud

🎯 What it does: Propose Gaussian-enhanced Surfels (GESs), which combine 2D opaque Surfels with 3D Gaussians through a two-stage rendering process to achieve fast and popping-free view synthesis.

WishGI: Lightweight Static Global Illumination Baking via Spherical Harmonics Fitting

Junke Zhu (University of Science and Technology of China), Zhangjin Huang (University of Science and Technology of China)

OptimizationComputational EfficiencyMesh

🎯 What it does: A lightweight static global illumination baking method called WishGI for low-end platforms is proposed. It uses spherical harmonic functions to approximate and reverse probe distribution, achieving vertex-level lighting reconstruction and single ProbeMap storage.

xADA: Controllable and Expressive Audio-Driven Animation

Sarah Taylor, Iain Matthews

GenerationData SynthesisRecurrent Neural NetworkTransformerDiffusion modelAuto EncoderGenerative Adversarial NetworkVideoAudio

🎯 What it does: Developed a generative model called xADA that directly generates facial, tongue, and head animations from speech audio. It supports automatic generation and allows users to override detected emotions and blink timings, while also mapping to an animation controller compatible with MetaHuman.