Skip to content

SIFU

Rebuild 3D clothed avatars from a side view image

Visit official site

What sifu is

The method reconstructs high quality 3D clothed virtual humans from side view images. A side view implicit function supports geometry, while 3D consistent texture optimization handles texture quality and editing.

What you can do with it

  • Reconstruct a high quality 3D virtual human from a single side view image
  • Simulate complex garments with better geometric accuracy
  • Feed the reconstructed model into 3D printing or animation work

What to watch out for

  • Input is a side view image rather than a full multi view capture
  • Texture editing relies on a text to image diffusion model

Pros & cons

✓ What we like

  • Complex poses and loose clothing are handled
  • Virtual scene creation and in the wild scene reconstruction are listed

! What to watch out for

  • Input is a side view image rather than a full multi view capture
  • Texture editing relies on a text to image diffusion model

FAQ

What input does it use?

The material describes reconstructing from a single side view image.

What outputs are listed?

3D printing support, texture editing, animation and virtual scene creation are listed.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review