kongzhecn/OMG
A diffusion model framework for generating images with multiple personalized concepts while handling occlusion, published at ECCV 2024.
Not currently ranked — collecting fresh signals.
star history
OMG (Occlusion-friendly Multi-concept Generation) is a research framework enabling high-quality image generation with multiple characters and styles simultaneously. It builds on diffusion models and supports integration with LoRA adapters for character and style customization, as well as InstantID for multi-identity face generation from single reference images. The framework handles complex scenes where characters may occlude each other.
Frequently asked
- What is kongzhecn/OMG?
- A diffusion model framework for generating images with multiple personalized concepts while handling occlusion, published at ECCV 2024.
- Is OMG open source?
- Yes — kongzhecn/OMG is an open-source project tracked on heatdrop.
- What language is OMG written in?
- kongzhecn/OMG is primarily written in Python.
- How popular is OMG?
- kongzhecn/OMG has 701 stars on GitHub.
- Where can I find OMG?
- kongzhecn/OMG is on GitHub at https://github.com/kongzhecn/OMG.