Skip to content

DiffusionLight

Estimate lighting from a single input image

Visit official site

What DiffusionLight is

The technique is described as estimating lighting from a single image by drawing a mirrored ball and unwrapping it into a panoramic lighting map.

What you can do with it

  • Draw a chrome ball from one photo
  • Unwrap the ball into a panorama map
  • Apply the estimate to indoor or outdoor scenes
  • Combine exposures for HDR output

Who it is for

  • Indoor and outdoor environment modelling and rendering
  • Compositing adjustments after image editing
  • Portrait photo enhancement

Pros & cons

✓ What we like

  • Indoor, outdoor and portrait scenes are named
  • Robustness is described against traditional methods

! What to watch out for

  • The technique starts from a single input image
  • HDR output relies on LoRA multi-exposure training

FAQ

Which model draws the ball?

The material names a trained Stable Diffusion XL model.

How is HDR produced?

Through LoRA based multi-exposure training on an LDR model.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review