LightLab: Controlling Light Sources in Images with Diffusion Models
Nadav Magar, Amir Hertz, Eric Tabellion, Yael Pritch, Alex Rav-Acha, Ariel Shamir, Yedid Hoshen
2025-05-15
Summary
This paper talks about LightLab, a new AI technique that lets you change and control the lighting in photos with much more accuracy than before, using a special kind of model called diffusion.
What's the problem?
The problem is that changing the way light looks in images—like making it brighter, dimmer, or moving the light source around—has always been really hard for computers to do realistically, especially if you want to control exactly how the lighting looks.
What's the solution?
The researchers trained their model using pairs of real and computer-generated images, so the AI could learn how lighting changes in different situations. With this training, LightLab can now adjust both the main light sources and the overall brightness in a photo, giving users precise control over how the final image looks.
Why it matters?
This matters because it opens up new creative possibilities for photographers, artists, and anyone editing images, making it easier to get the perfect lighting without needing to retake photos or use complicated editing tools.
Abstract
A diffusion-based method fine-tuned on real and synthetic image pairs provides precise control over light sources and ambient illumination in images, offering more effective relighting than existing methods.