paper-illustration-image2

To generate publication-quality academic figures, this Skill uses an iterative multi-step planning and layout optimization workflow to produce paper-ready architecture diagrams and method illustrations with correct palette, typography, and arrow semantics. It strictly scores the visual output to target a 9/10 rating before finalizing the image.

Yes, you need a local Codex app-server MCP bridge (codex-image2) to render native academic illustrations. The workflow requires this local native raster rendering bridge for explicit preflight, bounded render, and strict visual scoring, actively avoiding non-native fallbacks to ensure publication quality.

Yes, you can integrate generated paper illustrations directly into LaTeX. The workflow automatically packages the final accepted image as figure_final.png and emits a latex_include.tex file, allowing direct inclusion of the academic figure into your LaTeX document pipeline.

The best way to ensure correct layout and label semantics in architecture diagrams is using a planner and reviewer workflow that iteratively optimizes the figure layout. This process applies strict style verification to validate arrow and label semantics before promoting the image to a finalized paper-ready artifact.

Finalizing method illustration pipelines produces three main artifacts: figure_final.png containing the best accepted image, latex_include.tex for document integration, and review_log.json alongside verify.json to document the strict visual scoring and style verification results.

A key limitation of native image rendering for academic figures is the strict dependency on the local Codex app-server MCP bridge. The workflow avoids non-native fallbacks, meaning the bounded render and preflight checks will fail if the local codex-image2 environment is not properly configured.