Presentation

<
OmniRoam: World Wandering via Long-Horizon Panoramic Video Generation
SessionVideo Generation
DescriptionThis work introduces a controllable panoramic video generation framework designed for long-horizon scene wandering. It utilizes a global-to-local preview-and-refine pipeline to synthesize high-fidelity, spatially and temporally coherent panoramic videos from a single input panoramic image or video.
Authors
Yuheng Liu
University of California Irvine
Xin Lin
University of California San Diego
Xinke Li
City University of Hong Kong
Baihan Yang
University of California San Diego
Chen Wang
University of Pennsylvania
Kalyan Sunkavalli
Adobe Research
Yannick Hold-Geoffroy
Adobe Research
Hao Tan
Adobe Research
Kai Zhang
Adobe Research
Xiaohui Xie
University of California Irvine
Zifan Shi
Adobe Research
Yiwei Hu
Adobe Research
Event Type
Technical Paper
TimeMonday, 20 July 20262:10pm - 2:20pm PDT
LocationRoom 408 B
Digital Library PDF
Session Time & Location
Sunday, 19 July 20266:00pm - 8:45pm PDTHall K
Monday, 20 July 20262:00pm - 3:30pm PDTRoom 408 B
Keywords
Artificial Intelligence/Machine Learning
Generative AI
Registration Categories
Full Conference Supporter
Full Conference
Next PresentationNext PresentationActCam: Zero-Shot Joint Camera and 3D Motion Control for Video Generation
Similar Presentations
LongE2V: Long-Horizon Event-based Video Reconstruction, Prediction, and Frame Interpolation with Video Diffusion Models
MTPano: Multi-Task Panoramic Scene Understanding via Label-Free Integration of Dense Prediction Priors
HL-OutPaint: Coarse-to-Fine Video Outpainting for High-Resolution Long-Range Videos