Presentation

<
VfxDB: A Visual Effects Volume Dataset and Benchmark for VDB-Native Generative Modeling
SessionMultimodal Learning
DescriptionVfxDB introduces a pioneering 1-million-sample dataset tailored for VDB-native generation of sparse volumetric VFX sequences, alongside a reproducible benchmark suite. Validated by a scalable diffusion baseline featuring our Atomic-Continuous prior, this comprehensive pipeline establishes essential infrastructure for future 3D volume generative modeling.
Authors
Junwei Shu
East China Normal University
Hantang Liu
KunByte
Dawei Miao
Independent Researcher
Wenzheng Song
East China Normal University
Mingyang Yuan
East China Normal University
Wenjie Liu
East China Normal University
Changgu Chen
East China Normal University
Yang Li
East China Normal University
Changbo Wang
East China Normal University
Event Type
Technical Paper
TimeThursday, 23 July 20264:35pm - 4:45pm PDT
LocationRoom 403 A
Digital Library PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Thursday, 23 July 20263:45pm - 5:15pm PDTRoom 403 A
Keywords
Artificial Intelligence/Machine Learning
Registration Categories
Full Conference Supporter
Full Conference
Next PresentationNext PresentationMultimodal Learning - Interactive Discussion
Similar Presentations
Generative 3D Gaussians with Learned Density Control
CasLayout: Cascaded 3D Layout Diffusion for Indoor Scene Synthesis with Implicit Relation Modeling
PatternGSL: A Structured Specification Language for Template-Free and Simulation-Ready 3D Garments