View-Dependent Video Textures for 360° Video
summary
A system implements gated clips that hold playback at a gate time until a filmmaker-defined gaze condition (looking at an ROI) is met, using view-dependent video textures that duplicate timelines per discretized view and compute perceptual costs to find seamless backward loops. The view×frame grid becomes a graph and the algorithm uses graph cut to replace forward edges with low-cost backward arcs. Preprocessing for a 7 s 30 fps clip takes ~3.3 hours, graph cut ~0.5 s, and 9 of 11 participants preferred gated clips.
AI-written summary · about
chapters (5)
related talks
18mGaze-Driven Video Delivery: Science Fiction or Viable Link to the Metaverse?
23mRewind capture format on IG Stories
21mDynamic video in the browser
03mJust realized we had a “Time Machine” after Demuxed’18 day 1
13mVideoCompositor: a Case Study of Live Video Composing
37mFrom View to Pixel
14mReactive Video: Adaptive Video Playback Based on User Motion for Supporting Physical Activity
04mSub-frame Latency
