AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

AutoFigure: Generating and Refining Publication-Ready Scientific Illustrations

Minjun Zhu, Zhen Lin, Yixuan Weng, Panzhong Lu, Qiujie Xie, Yifan Wei, Sifan Liu, QiYao Sun, Yue Zhang

ICLR 2026 · Image generation › Text-to-image diffusion · approach: Layout-to-image control · 19 citations (OpenAlex)

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 changedWhereMeasuredEffect
use AutoFigure instead of best code baselinescientific illustration generation · FigureBench PaperOverall score: 6.35 -> 7.03Improved
use AutoFigure instead of SVG-Codescientific illustration generation · FigureBench BlogWin-Rate vs reference: 45.0% -> 75.0%Improved
use AutoFigure instead of Gemini-SVGscientific illustration generation · FigureBench TextbookWin-Rate vs reference: 76.9% -> 97.5%Improved
add critique-and-refine iterations (0 to 5) (ablation)AutoFigure · FigureBenchOverall score: 6.28 -> 7.14Improved
add final rendering stage after the symbolic layout (ablation)AutoFigure with GPT-5 · FigureBenchOverall score: 6.38 -> 7.480Improved
use SVG intermediate format instead of PPT (ablation)AutoFigure · FigureBenchOverall score: 6.12 -> 8.98Improved
use HTML intermediate format instead of PPT (ablation)AutoFigure · FigureBenchOverall score: 6.12 -> 8.85Improved
use Qwen3-VL-235B backbone instead of GLM-4.5V (ablation)AutoFigure · FigureBenchOverall score: 5.99 -> 7.08Improved
use GPT-5 backbone instead of Qwen3-VL-235B (ablation)AutoFigure · FigureBenchOverall score: 7.08 -> 7.48Improved
use Qwen3-VL-235B backbone instead of Gemini-2.5-Pro (ablation)AutoFigure · FigureBenchOverall score: 6.99 -> 7.08Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy