AutoFigure-Edit is a tool that converts scientific paper method sections into fully editable SVG figures with an embedded editor for refinement.
The tool addresses the challenge of manually creating scientific illustrations by automating the generation of editable vector graphics from text descriptions. It works through a multi-stage pipeline: first generating an initial raster image from method text, then using segmentation and reconstruction techniques to convert that image into editable SVG format, and finally allowing users to refine the result in an integrated editor. This approach bridges the gap between natural language descriptions and publication-ready vector illustrations that researchers can customize.
Researchers writing scientific papers who need to generate figures from method descriptions are the primary audience. The tool suits projects where method sections already contain detailed procedural information that can be translated into visual form. The project explicitly positions itself as a successor to an earlier version called AutoFigure, building on that foundation with the addition of SVG editability and refinement capabilities.
Development activity shows active maintenance with recent releases introducing practical workflow improvements. The project added support for importing user-supplied raster figures to skip initial generation steps. Official OpenAI model integration was implemented, including support for specific image and text models. Support for OpenAI-compatible API gateways was added alongside a built-in route for a specific compatible service. The codebase was expanded to support bilingual configuration and user interface workflows in both English and Chinese.
