adobe-research/custom-diffusion
A method for fine-tuning text-to-image diffusion models with few images to add new concepts or styles.
Not currently ranked — collecting fresh signals.
star history
Custom Diffusion enables few-shot fine-tuning of Stable Diffusion and other text-to-image diffusion models given 4-20 images of a new concept. The method optimizes only key and value projection matrices in cross-attention layers, enabling fast training (~6 minutes on 2 A100 GPUs) and lightweight per-concept storage (75MB). The approach supports single and multi-concept customization including object-style combinations and new category additions.
Frequently asked
- What is adobe-research/custom-diffusion?
- A method for fine-tuning text-to-image diffusion models with few images to add new concepts or styles.
- Is custom-diffusion open source?
- Yes — adobe-research/custom-diffusion is an open-source project tracked on heatdrop.
- What language is custom-diffusion written in?
- adobe-research/custom-diffusion is primarily written in Python.
- How popular is custom-diffusion?
- adobe-research/custom-diffusion has 2k stars on GitHub.
- Where can I find custom-diffusion?
- adobe-research/custom-diffusion is on GitHub at https://github.com/adobe-research/custom-diffusion.