Updated 2 months ago
Path: .
LayerDiffusion is an advanced text-guided image editing model that excels in performing multiple edits on single images, like background replacement and attribute changes, ensuring seamless integration and consistency. It leverages a text-to-image model with a novel layered control optimization and diffusion training to allow detailed subject modifications while preserving their identity. This repository contains the model weights
model nlp computer vision huggingface transformers image generation text-to-image generation dvc git arxiv