The authors developed LOCI using a Regional LoRA layer that replaces the scalar LoRA coefficient with a per-token scaling vector aligned to the spatial layout.
LOCI enables independent continuous editing of multiple subjects in a single image
Each subject can be edited at an independently chosen strength without cross-region interference.
Big Tech
Sai Vidyaranya Nuthalapati · Shlok Kumar Mishra
Meta AI
Research Digest··2 min read
The authors present LOCI, a framework that combines spatial localization with per-region continuous strength control for editing multiple subjects in one image.
The Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.