viznoir

English | 한국어 | 中文 | 日本語 | Deutsch | Français | Español | Português

<p align="center"> <img src="https://raw.githubusercontent.com/kimimgo/viznoir/main/www/public/showcase/cavity_story.webp" alt="viznoir showcase" width="720" /> <br> <strong>VTK is all you need.</strong><br> Cinema-quality science visualization for AI agents. </p>

<p align="center"> <b><a href="#quick-start">Quickstart</a></b> · <b><a href="https://kimimgo.github.io/viznoir/docs">Docs</a></b> · <b><a href="https://github.com/kimimgo/viznoir">GitHub</a></b> </p>

<p align="center"> <a href="https://github.com/kimimgo/viznoir/actions/workflows/ci.yml"><img src="https://github.com/kimimgo/viznoir/actions/workflows/ci.yml/badge.svg" alt="CI"></a> <a href="https://pypi.org/project/mcp-server-viznoir/"><img src="https://img.shields.io/pypi/v/mcp-server-viznoir" alt="PyPI"></a> <a href="https://github.com/kimimgo/viznoir/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-MIT-blue.svg" alt="License: MIT"></a> <a href="https://github.com/kimimgo/viznoir/stargazers"><img src="https://img.shields.io/github/stars/kimimgo/viznoir?style=flat" alt="Stars"></a> </p>

---

What is viznoir?

An MCP server that gives AI agents full access to VTK's rendering pipeline — no ParaView GUI, no Jupyter notebooks, no display server. Your agent reads simulation data, applies physics filters, renders cinema-quality images, and exports animations. All headless.

---

How it works

<table> <tr> <td align="center" width="33%"> <h3>01</h3> <b>Point to your data</b><br><br> <code>inspect_data("cavity.foam")</code> </td> <td align="center" width="33%"> <h3>02</h3> <b>Ask in natural language</b><br><br> <em>"Render pressure with cinematic lighting"</em> </td> <td align="center" width="33%"> <h3>03</h3> <b>Get cinema-quality output</b><br><br> PNG · MP4 · glTF · LaTeX </td> </tr> </table>

---

Works with

<table> <tr> <td align="center" width="16%">_{<b>Claude Code</b>}</td> <td align="center" width="16%">_{<b>Cursor</b>}</td> <td align="center" width="16%">_{<b>Windsurf</b>}</td> <td align="center" width="16%">_{<b>Gemini CLI</b>}</td> <td align="center" width="16%">_<b>Codex</b></td> <td align="center" width="16%">_{<b>Any MCP Client</b>}</td> </tr> </table>

<p align="center"><em>If it speaks MCP, it renders.</em></p>

---

Right for you if

• ✅ You run CFD/FEA simulations and want automated post-processing
• ✅ You want cinema-quality renders without learning ParaView
• ✅ You need headless visualization in CI/CD pipelines
• ✅ You want one prompt to go from raw data to publication figures
• ✅ You process 50+ file formats (OpenFOAM, CGNS, Exodus, STL, ...)

---

Features

<table> <tr> <td align="center" width="33%"> <h3>🎬 Cinema Render</h3> 3-point lighting, SSAO, FXAA, PBR materials. Publication-ready in one call. </td> <td align="center" width="33%"> <h3>🔬 Physics Analysis</h3> Vortex detection, stagnation points, gradient stats, Reynolds number. </td> <td align="center" width="33%"> <h3>📊 Data Extraction</h3> Line plots, surface integrals, time-series probes, statistical summaries. </td> </tr> <tr> <td align="center" width="33%"> <h3>🎞️ Animation</h3> 7 physics presets, 17 easing functions, scene transitions, video export. </td> <td align="center" width="33%"> <h3>🧩 50+ Formats</h3> OpenFOAM, VTK, CGNS, Exodus, STL, glTF, NetCDF, PLOT3D, and more. </td> <td align="center" width="33%"> <h3>🤖 Agent Harness</h3> <code>auto_postprocess</code> meta-tool with MCP sampling for full autonomy. </td> </tr> <tr> <td align="center" width="33%"> <h3>📐 Adaptive Resolution</h3> analyze 480p, preview 720p, publish 1080p. Context-aware quality scaling. </td> <td align="center" width="33%"> <h3>🔄 Pipeline DSL</h3> Compose multi-step filter chains into a single executable pipeline. </td> <td align="center" width="33%"> <h3>🖥️ Headless GPU</h3> EGL/OSMesa rendering, Docker support, no display server needed. </td> </tr> </table>

---

Without viznoir vs. With viznoir

<table> <tr> <th width="50%">Without viznoir</th> <th width="50%">With viznoir</th> </tr> <tr> <td>❌ Open ParaView GUI, click through menus, export manually</td> <td>✅ One prompt, headless, cinema-quality, automated</td> </tr> <tr> <td>❌ Write 200-line VTK Python scripts for each visualization</td> <td>✅ Natural language — the agent writes the pipeline</td> </tr> <tr> <td>❌ No rendering in CI/CD — need a display server</td> <td>✅ EGL/OSMesa headless — runs anywhere, including Docker</td> </tr> <tr> <td>❌ Manual camera placement, lighting, colormap tuning</td> <td>✅ PCA auto-camera, 3-point lighting, adaptive resolution</td> </tr> </table>

---

What viznoir is NOT

Not a simulation solver	It visualizes results, it does not run CFD/FEA solvers
Not ParaView	No GUI — pure headless API designed for AI agents
Not a Jupyter widget	MCP server, not an interactive notebook extension
Not a mesh generator	It reads meshes, it does not create them

---

Quick Start

pip install mcp-server-viznoir

Add to your MCP client config:

{
  "mcpServers": {
    "viznoir": {
      "command": "mcp-server-viznoir"
    }
  }
}

Then ask your AI agent:

"Open cavity.foam, render the pressure field with cinematic lighting, then create a physics decomposition story."

---

Numbers

22 MCP tools · 12 resources · 4 prompts · 1505+ tests 97% coverage · 50+ file formats · 7 animation presets · 17 easing functions

---

Documentation

Homepage — kimimgo.github.io/viznoir

Developer docs — kimimgo.github.io/viznoir/docs — full tool reference, domain gallery, architecture guide

---

Contributing

Contributions are welcome. Please open an issue first to discuss what you would like to change.

pip install -e ".[dev]"
pytest --cov=viznoir -q
ruff check src/ tests/

---

License

MIT

---

Star History

<p align="center"> <a href="https://star-history.com/#kimimgo/viznoir&Date"> <img src="https://api.star-history.com/svg?repos=kimimgo/viznoir&type=Date" alt="Star History Chart" width="600" /> </a> </p>

---

<p align="center"> <em>Open source under MIT. Built for engineers who'd rather prompt than click.</em> </p>