ziqihuangg/ReVersion
A diffusion-based method for inverting visual relations from example images into reusable prompts for image generation.
Not currently ranked — collecting fresh signals.
star history
ReVersion proposes the task of Relation Inversion—extracting relational patterns from example image pairs and applying them to generate new images. The method uses a diffusion model backbone with specialized relation modeling to learn reusable relation prompts. It enables generalization from few examples to novel scenes while preserving the extracted relationship pattern.
Frequently asked
- What is ziqihuangg/ReVersion?
- A diffusion-based method for inverting visual relations from example images into reusable prompts for image generation.
- Is ReVersion open source?
- Yes — ziqihuangg/ReVersion is an open-source project tracked on heatdrop.
- What language is ReVersion written in?
- ziqihuangg/ReVersion is primarily written in Python.
- How popular is ReVersion?
- ziqihuangg/ReVersion has 503 stars on GitHub.
- Where can I find ReVersion?
- ziqihuangg/ReVersion is on GitHub at https://github.com/ziqihuangg/ReVersion.