What

PaperBanana is an academic illustration generator for researchers who need methodology diagrams, AI research figures, statistical plots, and cleaned-up paper visuals. It is positioned as a research-focused alternative to generic AI image tools, using paper text, references, sketches, and existing figures to reduce manual redrawing and improve figure clarity near submission time.

Features

Text-to-figure generation: Converts method sections, notes, or reference text into structured methodology diagrams and system overviews, which helps users avoid starting from a blank canvas.
Sketch and reference-based generation: Accepts whiteboard sketches, screenshots, draft diagrams, and other reference images to guide figure creation or cleanup.
AI figure editing: Lets users revise existing visuals with text instructions to improve composition, labels, spacing, and overall readability.
Statistical plot and panel support: Supports benchmark charts, ablation figures, comparison panels, and poster-style multi-panel layouts for research communication.
Iteration and export controls: Offers settings such as iteration depth, aspect ratio, and upfront credit estimates, with stated refunds for unused credits if a figure converges early.

Helpful Tips

Use actual paper text or method notes rather than short promotional-style prompts; the page explicitly indicates this improves stage order, label accuracy, and figure logic.
Start with a rough sketch or reference image when layout matters, since this can help preserve structure and reduce revision cycles.
Treat free drafts as concept previews; the site states paid models are better suited for cleaner labels and more publication-ready outputs.
Validate technical faithfulness before submission, especially for plots and architecture figures, because the page emphasizes iterative refinement but does not prove full automatic correctness in every case.

OpenClaw Skills

PaperBanana could likely fit into OpenClaw workflows for research ops, paper production, and lab knowledge management, although the page does not confirm any native integration. Likely use cases include an agent that turns a draft method section into figure briefs, a review workflow that checks figure-caption consistency, or a publication assistant that routes sketches, references, and revision notes into a repeatable figure-generation process. Combined with OpenClaw, this could reduce manual diagram work for researchers, technical writers, and R&D teams by making figure creation a more structured step in the research publishing pipeline.