#101 · Primary category: Image Generation

PosterCraft

category-generative-vision generative-ai image-generation poster-generation python research

[ICLR'26] Rethinking High-Quality Aesthetic Poster Generation in a Unified Framework

Project last updated:01/27/26

GitHub Stars

545

Forks

35

Contributors

2

License

Other

Why we included this project

PosterCraft generates finished posters from a text prompt, and it focuses on the two things that usually sink text-to-image tools: typography that stays legible and a layout that holds together. It builds on FLUX.1 dev with a multi-stage training recipe and a reinforcement-learning stage that tries to balance text accuracy against overall visual harmony, so the result looks like a designed poster rather than an illustration with text pasted on. The repo ships model weights, inference scripts, and a Gradio demo, which makes it easy to judge whether the approach meets your quality bar for marketing collateral, event graphics, or automated design work. There's also a Poster100K dataset, and a community member has integrated PosterCraft into ComfyUI, so you can try it inside an existing generation workflow. One caveat: the model is non-commercial.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category