1 / 41

Video Textures

Video Textures. Arno Schödl Richard Szeliski David Salesin Irfan Essa Microsoft Research, Georgia Tech. Still Photos. Video Clips. Video Textures. Problem Statement. video clip. video texture. Approach. How do we find good transitions?. Finding Good Transitions.

hanzila
Download Presentation

Video Textures

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Video Textures Arno Schödl Richard Szeliski David Salesin Irfan Essa Microsoft Research, Georgia Tech

  2. Still Photos

  3. Video Clips

  4. Video Textures

  5. Problem Statement video clip video texture

  6. Approach How do we find good transitions?

  7. Finding Good Transitions Compute L2 distance Di, j between all frames vs. frame i frame j Similar frames make good transitions

  8. Markov Chain Representation Similar frames make good transitions

  9. Transition Costs Transition from i to j if successor of i is similar to j Cost function: Cij = Di+1, j

  10. Transition Probabilities Probability for transition Pij inversely related to cost: high  low 

  11. Preserving Dynamics

  12. Preserving Dynamics

  13. Preserving Dynamics Cost for transition ij Cij = wkDi+k+1, j+k

  14. Preserving Dynamics – Effect Cost for transition ij Cij = wkDi+k+1, j+k

  15. Deadends No good transition at the end of sequence

  16. Future Cost • Propagate future transition costs backward • Iteratively compute new cost Fij = Cij +  minkFjk

  17. Future Cost • Propagate future transition costs backward • Iteratively compute new cost Fij = Cij +  minkFjk

  18. Future Cost • Propagate future transition costs backward • Iteratively compute new cost Fij = Cij +  minkFjk

  19. Future Cost • Propagate future transition costs backward • Iteratively compute new cost Fij = Cij +  minkFjk

  20. Future Cost • Propagate future transition costs backward • Iteratively compute new cost Fij = Cij +  minkFjk • Q-learning

  21. Future Cost – Effect

  22. Finding Good Loops • Alternative to random transitions • Precompute a good set of loops up front (using dynamic programming)

  23. Visual Discontinuities • Problem: Visible “Jumps”

  24. Crossfading • Solution: Crossfade from one sequence to the other.

  25. Morphing • Interpolation task:

  26. Morphing • Interpolation task: • Compute correspondencebetween pixels of all frames

  27. Morphing • Interpolation task: • Compute correspondence between pixels of all frames • Interpolate pixel position andcolor in morphed frame • based on [Shum 2000]

  28. Results – Crossfading / Morphing

  29. Results – Crossfading / Morphing Jump Cut Crossfade Morph

  30. Crossfading

  31. Frequent Jump & Crossfading

  32. Video Portrait Useful for web pages

  33. Video Portrait – 3D Combine with IBR techniques

  34. Region-based Analysis • Divide video up into regions • Generate a video texture for each region

  35. Automatic Region Analysis

  36. User-controlled Video Textures User selects target frame range slow variable fast

  37. Time Warping Lengthen / shorten video without affecting speed shorter original longer

  38. Summary • Video clips  video textures • define Markov process • preserve dynamics • avoid dead-ends • disguise visual discontinuities

  39. Summary • Extensions • regions • external constraints • video-based animation

  40. Discussion • Some things are relatively easy

  41. Discussion • Some are hard

More Related