Presentation

<
EasyVFX: Frequency-Driven Decoupling for Resource-Efficient VFX Generation
SessionFilm/VFX Production
DescriptionEasyVFX is a resource-efficient framework for high-fidelity VFX synthesis. It leverages frequency-domain decomposition to disentangle spatial detail and motion dynamics, reducing learning complexity. A frequency-aware MoE enables efficient training, while a test-time adaptation strategy refines unseen effects. The approach achieves realistic, consistent results with minimal data and computation.
Authors
Yue Ma
Hong Kong University of Science and Technology
Xu Ye
Hong Kong University of Science and Technology
Qinghe Wang
Dalian University of Technology
Yucheng Wang
Hong Kong University of Science and Technology
Hongyu Liu
Hong Kong University of Science and Technology
Yinhan Zhang
Hong Kong University of Science and Technology
Xinyu Wang
Tsinghua University
Yuanpeng Chen
Independent
Shanhui Mo
Independent
Paul Liang
Massachusetts Institute of Technology (MIT)
Fangneng Zhan
Massachusetts Institute of Technology (MIT)
Qifeng Chen
Hong Kong University of Science and Technology
Event Type
Technical Paper
TimeWednesday, 22 July 20262:20pm - 2:30pm PDT
LocationRoom 408 B
Digital Library PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Wednesday, 22 July 20262:00pm - 3:30pm PDTRoom 408 B
Keywords
Generative AI
Registration Categories
Full Conference Supporter
Full Conference
Next PresentationNext PresentationLongE2V: Long-Horizon Event-based Video Reconstruction, Prediction, and Frame Interpolation with Video Diffusion Models
Similar Presentations
Colorful-Noise: Training-Free Low-Frequency Noise Manipulation for Color-Based Conditional Image Generation
ViewWeaver: Geometry-Grounded Generative Rendering for 3D-Aware Image Customization
BFS: Back-to-Front Layered Image Synthesis via Knowledge Transfer