PaperBanana supports five main illustration types: Methodology Diagrams (neural network architectures, algorithm flowcharts, system pipelines), Statistical Plots (bar charts, line graphs, scatter plots with accurate data), Aesthetic Enhancement (polishing rough sketches into publication-quality graphics), Educational Infographics (visual explanations for lectures and tutorials), and Aesthetic Refinement (improving existing diagrams' visual quality).

PaperBanana uses a multi-agent workflow with five specialized agents. The Retriever finds relevant reference examples, the Planner translates your content into detailed descriptions, the Stylist ensures adherence to academic aesthetic standards, the Visualizer renders the images, and the Critic inspects and provides feedback for iterative refinement.

PaperBanana works with text descriptions of your research content. You can provide methodology descriptions, data for statistical plots, or descriptions of concepts you want to visualize. The more detailed your input, the better the results. You can also upload reference images for style guidance.

Yes. If the first result isn't perfect, you can adjust your prompt and regenerate. You can also use PaperBanana's Aesthetic Refinement feature to polish specific visual elements like colors, fonts, or layout while keeping your original structure.

Yes, all illustrations generated by PaperBanana are yours to use in research papers, presentations, posters, and other academic materials. The output is optimized to meet the aesthetic standards of top-tier venues like NeurIPS, ICML, and ICLR.

PaperBanana outputs high-resolution images suitable for publication. For statistical plots, you can also download the generated Python code to further customize the visualization in your preferred environment.