Awesome Diffusion Models

The list provides a navigable map of a rapidly evolving research field, organized by topic (unconditional generation, text-to-image, image editing, video generation, 3D generation, audio synthesis) with links to both the original papers and open-source implementations.

The collection spans foundational papers (DDPM, DDIM, Score Matching, Flow Matching) alongside applied work on efficient sampling (LCM, SDXL-Turbo, Consistency Models), conditioning mechanisms (ControlNet, IP-Adapter, T2I-Adapter), fine-tuning approaches (DreamBooth, LoRA, Textual Inversion), and multimodal diffusion models.

Each entry includes the paper title, venue, year, links to the arXiv preprint and official code, and brief categorization tags making it possible to filter for specific capability areas.

As a GitHub awesome-list, the repository accepts community contributions via pull requests, with maintainers reviewing submissions for relevance and formatting consistency.

It has become one of the primary navigation tools for ML researchers entering the diffusion model space, serving as a living bibliography that complements static textbooks and review papers.

The breadth of coverage from early 2020 DDPM papers through the latest video generation research makes it useful both for historical perspective and tracking the current state of the art.