Ant Research · 2026-08-20 · notable
4DAnyone — turn one handheld video of a person into a 4D model
4DAnyone rebuilds a moving person in 4D from a single ordinary video. The Ant Research system generates multiview-consistent video first, then lifts it into 4D Gaussian Splatting. Code and checkpoints are public.
4DAnyone reconstructs a moving person in 4D from one casual monocular video, with code and checkpoints released.
What is it?
4DAnyone takes a casual single-camera video of a person and rebuilds them as a 4D Gaussian Splatting scene you can view from any angle over time. The work comes from Ant Research with Zhejiang University and HKUST, and was accepted to SIGGRAPH Asia 2026. Code, checkpoints and example videos are already public.
How does it work?
The pipeline generates multiview-consistent video of the subject, then lifts those views into 4D Gaussian Splatting. Two ideas keep generation stable as the number of target views grows: Reference Context Packing compresses the reference views into a fixed-length mixed-resolution context, and Target Context Routing rotates which views are denoised together so information crosses groups. Depth-buffered skeleton renderings supply 3D-aware pose conditioning, and training adds MVGameHuman, a game-engine dataset the team built.
Why does it matter?
Capturing a person in 4D normally needs a camera rig or a studio. Working from a phone-style clip puts free-viewpoint replay, avatar creation and 3D video editing in reach of people with no capture hardware. The released repo and checkpoints let researchers reproduce the result instead of only reading about it.
Who is it for?
graphics and computer vision researchers
Try it
https://github.com/ant-research/4DAnyone