~/talk

MoSculp: Interactive Visualization of Shape and Time

on YouTube 20m
Talkat UIST 2018 · 2018-10-14..10-17· [data-visualization, computer-vision]

speakers

name:Xiuming Zhang
links:none on file
name:Tali Dekel
links:none on file
name:Tianfan Xue
links:none on file
name:Andrew Owens
links:none on file
name:Qiurui He
links:none on file
name:Jiajun Wu
links:none on file
name:Stefanie Mueller
links:none on file
name:William T. Freeman
links:none on file

resources

video:https://youtu.be/mQIBdQ4WggE
slides:none on file

summary

MoSculp computes a 3D “motion sculpture” from a single RGB video by detecting 2D keypoints with CMU OpenPose, fixing flips via an HMM, then jointly estimating per-frame 3D body shape and pose with reprojection, pose priors and temporal smoothness. It composites a hybrid depth-aware rendering refined by optical flow, supports material/lighting controls and 3D-printable export, handles moving cameras, and scores better than alternatives (about 75% of study participants prefer its motion clarity); it fails on tightly localized, highly periodic motion.

AI-written summary · about

chapters (14)

  1. 0:00Introduction
  2. 0:25Motion Sculpture
  3. 1:42Outline
  4. 2:20The Problem
  5. 3:25User Interface
  6. 5:11User Studies
  7. 6:08Approach Overview
  8. 8:17Smooth Post Evolution
  9. 9:49Rendering
  10. 11:09Hybrid Approach
  11. 12:05Results
  12. 13:13Jumping Sequence
  13. 15:21Conclusion
  14. 16:35Questions

related talks

~/search
type to search 6,249 conferences, 138,987 speakers and 185,172 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.