Skip to content

Phidias

Reference-enhanced 3D generation with diffusion

Visit official site

What Phidias is

The model is described as a diffusion-based generative model for reference-enhanced 3D generation with a unified framework for three input types.

What you can do with it

  • Generate 3D assets from image, text or 3D
  • Guide generation with 3D references
  • Adjust geometry interactively

Application scenarios

  • A designer turns concept art into a 3D prototype
  • A game developer builds complex scenes
  • A researcher creates a 3D experiment setting

Pros & cons

✓ What we like

  • Dynamic reference routing reduces mismatch
  • Interactive generation adjusts geometry

! What to watch out for

  • Output depends on the 3D reference
  • Generation relies on diffusion techniques

FAQ

Which inputs are supported?

The feature list names image, text and 3D conditions.

What are the key components?

Meta-ControlNet, dynamic reference routing and self-reference augmentation.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review