viznoir

viznoir showcase
VTK is all you need.
Cinema-quality science visualization for AI agents.

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

01

Point to your data

inspect_data("cavity.foam")

02

Ask in natural language

"Render pressure with cinematic lighting"

03

Get cinema-quality output

PNG · MP4 · glTF · LaTeX

Works with

_{Claude Code}

_Cursor

_Windsurf

_{Gemini CLI}

_Codex

_{Any MCP Client}

If it speaks MCP, it renders.

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

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

Without viznoir vs. With viznoir

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

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

Open source under MIT. Built for engineers who'd rather prompt than click.

viznoir

viznoir

What is viznoir?

How it works

01

02

03

Works with

Right for you if

Features

🎬 Cinema Render

🔬 Physics Analysis

📊 Data Extraction

🎞️ Animation

🧩 50+ Formats

🤖 Agent Harness

📐 Adaptive Resolution

🔄 Pipeline DSL

🖥️ Headless GPU

Without viznoir vs. With viznoir

What viznoir is NOT

Quick Start

Numbers

Documentation

Contributing

License

Star History

Reviews