Hi, I'm Jacob Junyi Chen 👋

Talk is cheap. Show me the code. 🧑‍💻

• Image/Video Generation:

  • VINO: unified visual generator / omni-model (understanding & generation) / world model
  • iMontage: many to many image generation
  • Infinite-Forcing: long video generation
  • Tiny-Sora2: audio-video generation

• World Models, 3D/4D Generation:

  • DeepVerse: interactive 4d video generation
  • Aether: unified 4d reconstruction and generation
  • GVGEN: text/image-to-3DGS in 7 seconds
  • GST: unified novel view synthesis and camera pose estimation
  • MeshCraft: text/image-to-mesh in 3 seconds (800 face level)
  • CONE: text-to-nerf

• 3D/4D Reconstruction:

  • Pi3: feed-forward 3d/4d reconstruction w/o reference view
  • WinT3R: feed-forward 3d/4d reconstruction in a streaming manner
  • GigaGS: city-level 3d surface reconstruction
  • CoSurfGS: large-scale 3d surface reconstruction with distributed learning
  • OmniWorld: multi-modal 4d dataset