VILA-Lab

VILA-Lab / FigMirror

Public

Plot your data in any paper's style.

claude-code codex data-visualization llm-agents matplotlib
14
1
89% credibility
Found May 23, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

FigMirror is a research tool that copies the visual style of any chart from a published paper and applies it to your own data. You paste a reference image and your data, and an AI assistant iteratively draws and refines a figure until it matches the paper's look — then hands you both the finished image and the clean code behind it. It runs locally as a web app or as a plugin inside AI coding assistants like Claude Code or Codex.

How It Works

1
🎨 Copy a paper's visual style onto your data

You upload any chart from a research paper and paste your own data — FigMirror learns the visual style and recreates it with your numbers.

2
📦 Install in one click

You run a simple one-line installer and everything sets itself up automatically — no technical setup required.

3
🖼️ Drop in your reference image and data

You paste a screenshot from any paper and either type or paste your data table directly in the browser.

4
⚡ Watch your figure come to life

An AI assistant draws and reviews the figure repeatedly, each time getting closer to matching the paper's exact style — you watch the iterations appear live.

5
💬 Ask for fine-tuning in plain language

If something isn't quite right, you type a request like 'make the legend smaller' or 'use bolder colors' and the assistant updates the figure instantly.

🎉 Download your camera-ready figure and clean code

You get a polished, publication-quality image plus the full editable code so you can tweak it however you want in the future.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 14 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is FigMirror?

FigMirror is a Python tool that takes any paper figure as a style reference and recreates it with your own data, outputting editable matplotlib code plus a PDF ready for publication. It works through an agentic loop where a "Drawer" generates candidates and a "Reviewer" compares them against the reference, iterating until the output matches the source style. You can run it as a Claude Code or Codex skill, or spin up a local web UI where you upload a reference image, paste your data, and watch iterations appear in real time.

Why is it gaining traction?

The hook is solving the "my plots look different from the paper" problem that plagues anyone trying to match a publication's aesthetic without manually reverse-engineering matplotlib styling. Instead of tweaking rcParams for hours, you point at the figure you want and let the agents work it out. The project ships a 139-figure gallery covering 25 chart types so you don't need to hunt for a reference, and the grounded measurement approach means the reviewer actually checks pixel-level fidelity rather than just trusting the agent's self-assessment.

Who should use this?

Researchers writing papers who need their plots to visually match a specific publication style will get the most value. Data scientists creating figures for presentations where consistency with a reference design matters. Python developers who want to generate publication-quality plots without deep matplotlib expertise.

Verdict

The concept is solid and the implementation is more mature than the 14 stars suggest, with a working web UI and support for both major AI coding agents. At 0.8999999761581421% credibility, it's early but functional -- the main risks are that the skill installation process assumes familiarity with Claude Code/Codex, and the 3D figure support is still maturing. Worth trying if you frequently generate figures that need to match a paper's visual language.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.