Skip to content

DiffBIR

Blind image restoration with generative diffusion priors

Visit official site

What DiffBIR is

The model is described as a blind image restoration model based on generative diffusion priors. The model works in two stages that remove image degradation and refine image detail, and the material states its use is free.

What you can do with it

  • Remove degradation from images
  • Refine fine detail in images
  • Adjust parameters to balance fidelity and quality
  • Apply the model to general image restoration
  • Apply the model to face image restoration

Model notes

  • The two-stage process is built on generative diffusion priors
  • The material describes high-quality image restoration results
  • Flexible parameter settings allow a fidelity and quality trade-off
  • Image quality optimization is listed among the uses

Pros & cons

✓ What we like

  • A two-stage process covers degradation removal and detail refinement
  • The material states the model is free to use

! What to watch out for

  • Fidelity and quality are balanced through parameter settings
  • The description centres on blind image restoration

FAQ

How many stages does it use?

The material describes a two-stage process that removes degradation and refines detail.

Is it free to use?

The material states that using the model is free.

Last reviewed: 2026-11-15

More AI office assistant tools

View all →

How we review