Skip to content
ComputationalRoboticsPublic

About

Website for XM4D project.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

8 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

XM4D project page

Single-page site for XM4D: Non-Rigid ConveX Structure from Motion with Embedded Deformation Graphs for 4D Reconstruction. Plain HTML, CSS, and JS; no build step. The layout is adapted from the BEACON page, which is based on the gated-memory-policy template (CC BY-SA 4.0).

Preview

python3 -m http.server 8080   # then open http://localhost:8080

Files

index.html        all page content
style.css         all styling; colors are tokens under :root (--accent = xmblue, --accent-2 = xmorange)
main.js           BibTeX copy, gallery tabs, lazy video playback, method-step figure swap
assets/
  favicon.svg
  group-logo.png  Computational Robotics Group logo
  img/            teaser.jpg, view-graph.jpg (rendered from the paper figures)
    teaser/       crops of the teaser (from the paper's teaser.pdf) for the hero walkthrough
    runs/         posters of the Example runs clips
  video/          the walkthrough's clips (horsejump-high); each plays over its panel at its stop (data-stop)
    teaser-tracks.mp4  Input: every input frame with its tracks, over the front frame of the 2D-tracks stack
    teaser-depth.mp4   Input: every frame's monocular depth, over the front frame of the depth stack
    teaser-lift.mp4    Lift: frame 33 lifted along its camera rays to its depth, over the Preprocess schematic
    teaser-solve.mp4   Solve: over panel (b), the XM solver's iterates on this video (from a replay of its solve
                       that saved every iterate: the cameras, and every frame's lifted depth placed by its camera
                       and depth scale), then the dynamic tracks through time (cameras along the path, tracks growing)
    teaser-render.mp4  Render: the trained model rendered at every frame from panel (c)'s camera and crop,
                       with its tracked keypoints
    runs/         the Example runs gallery below the teaser: per run, one input camera's render, 3D tracks, and
                  graph cells, cut from its showcase video (visualize_showcase_grid.py in the research code);
                  posters in img/runs/

Filling in the skeleton

  • Links: the Paper buttons (top bar and hero) point to #; Code points to XM4D-release.
  • Venue, BibTeX: see the TODO comments in index.html.
  • Hero video: uncomment the hero-video block and drop the files under assets/video/.
  • Method figures: each .step has a data-img that the sticky stage shows while the step is active.
  • Result videos: replace a <div class="tile-ph"> with <video muted loop playsinline preload="none" poster="….jpg" data-src="….mp4"></video>; main.js loads it lazily and plays it only while it is on screen.

About

Website for XM4D project.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages