ali-vilab/AnyDoor
Diffusion-based model for zero-shot object-level image customization that transfers objects from reference images into target scenes.

Velocity · 7d
+4.0
★ / day
Trend
→steady
star history
This repository contains the official implementation of AnyDoor, a diffusion-based generative model for zero-shot object-level image customization. The model enables placing objects from a reference image into target scenes while preserving object-specific details and characteristics. It supports image composition, object transfer, and various customization tasks through a single trained model without fine-tuning.