K2

K² · Artificial intelligence

Virtual try-on videos where you can spin the camera all the way around

The authors define a new research task called Camera-controllable Video Virtual Try-on (CaM-VVT), which extends video virtual try-on to allow arbitrary, unconstrained camera movements.

HS
HY

Hao Sun, Hao Yan, Mengting Chen et al.

10 authors · cs.CV

arXiv preprintArtificial intelligenceJun 2026 · ~65s read

The 30-second scan

Today's clothing try-on videos have a hidden handcuff: they can only show you the exact angles the original camera filmed.

  1. They identify that existing Video Virtual Try-on paradigms are fundamentally constrained by a passive dependency on source camera trajectories and cannot support omnidirectional viewpoint exploration.
  2. They present TryOnCrafter, described as the first unified DiT-based framework specifically architected for the CaM-VVT task.
  3. They introduce a Renderable 4D Try-on Proxy that explicitly decouples the human subject from the environment.
360-degree orbital viewing as a supported downstream application
Read the originalPDF