Research map / Large language models / Agents, code and math
PaperBanana: Automating Academic Illustration for AI Scientists
Dawei Zhu, Rui Meng, Yale Song, Xiyu Wei, Sujian Li, Tomas Pfister, Jinsung Yoon
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use PaperBanana agentic workflow instead of vanilla image generation | Nano-Banana-Pro · PaperBananaBench (diagrams) | Overall win score: 43.2 -> 60.2 | Improved |
| add Critic refinement (3 iterations) (ablation) | PaperBanana · PaperBananaBench (diagrams) | Overall win score: 45.6 -> 60.2 | Improved |
| add Critic refinement (3 iterations) (ablation) | PaperBanana · PaperBananaBench (diagrams) | Faithfulness: 30.7 -> 45.8 | Improved |
| increase Critic iterations from 1 to 3 (ablation) | PaperBanana · PaperBananaBench (diagrams) | Overall win score: 51.8 -> 60.2 | Improved |
| add Stylist agent (ablation) | PaperBanana (no Critic) · PaperBananaBench (diagrams) | Conciseness: 61.7 -> 79.2 | Improved |
| add Stylist agent (ablation) | PaperBanana (no Critic) · PaperBananaBench (diagrams) | Faithfulness: 39.2 -> 30.7 | Hurt |
| use semantic retriever instead of random retriever (ablation) | PaperBanana (no Stylist, no Critic) · PaperBananaBench (diagrams) | Overall win score: 48.3 -> 49.2 | No clear effect |
| add reference examples (retriever) to planner (ablation) | PaperBanana (no Stylist, no Critic) · PaperBananaBench (diagrams) | Overall win score: 44.2 -> 49.2 | Improved |
| add 10 few-shot examples to prompt | Nano-Banana-Pro · PaperBananaBench (diagrams) | Overall win score: 43.2 -> 41.8 | No clear effect |
| swap Nano-Banana-Pro for GPT-Image-1.5 as image generator | PaperBanana · PaperBananaBench (diagrams) | Overall win score: 60.2 -> 19.0 | Hurt |
| use image generation instead of code for Visualizer | PaperBanana · Statistical plots | Faithfulness and Conciseness | Mixed |
| apply PaperBanana instead of vanilla Gemini-3-Pro | Gemini-3-Pro · Statistical plots (ChartMimic-derived, 240 cases) | Overall: +4.1% | Improved |
Later papers that compared with this method
1 later accepted paper reports PaperBanana as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Automatic Method Illustration Generation for AI Scientific Papers via Drawing Middleware Creation, Evolution, and Orchestration (ECCV 2026 · better on 4 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers (NeurIPS 2025 · similarity 0.87)
- P2P: Automated Paper-to-Poster Generation and Fine-Grained Benchmark (ICLR 2026 · similarity 0.86)
- PaperBench: Evaluating AI’s Ability to Replicate AI Research (ICML 2025 · similarity 0.84)
- Presenting a Paper is an Art: Self-Improvement Aesthetic Agents for Academic Presentations (ICLR 2026 · similarity 0.84)
- AAAR-1.0: Assessing AI’s Potential to Assist Research (ICML 2025 · similarity 0.84)
- Agent-Native Research Artifacts (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
