paperbanana

Why we included this project

Drawing the architecture diagrams and flow charts that go into a paper is often the slowest part of writing it up. PaperBanana automates that step: you give it a description of your methodology and a figure caption, and a multi-agent pipeline turns that into a draft illustration, with a vision-language model doing the drawing. It is a community-driven implementation of Google Research's PaperBanana, extended with additions such as slide generation, and ships as a Python package with a CLI and a Colab notebook, so you can try it without setting up much infrastructure. The output is meant to look like the figures in conference papers, not marketing graphics. If you write papers or put together technical decks, it is a quick way to iterate on figure drafts instead of starting from a blank canvas.