Skip to content
MCP server

Thicken

Turn photos of a book, card or badge into a real-size 3D model plus JSON metadata

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted
Runs
Locally
Version
0.4.1
Listing updated
today

Similar MCP servers

About Thicken

Thicken is an open-source MCP server for design and media. It runs locally on your machine and is started by your AI client.

Each run writes a GLB model sized in millimetres and a JSON sidecar that records what the object is. The sidecar is built from codes on the object, such as an ISBN barcode or a grader's QR code, public catalogs, and optionally vision models that read printed text. For a graded card slab it can capture the grader, cert number, grade, signer and set; for a book, the catalog record with title, authors, publishers and subjects.

What it knows about each kind of object lives in a profile. Bundled profiles cover books, single cards, graded slabs, boxes, badges and puffy 3D lettering, and you can add your own. The README says it is not photogrammetry: it builds geometry from known sizes and maps your photos onto it.

It is a Node 22.18+ command-line tool and library that also runs as an MCP server over stdio. Uncropped or angled photos and video need an OpenAI-compatible vision endpoint.

Install

Run in a terminal

claude mcp add thicken --env THICKEN_VISION_KEY='<THICKEN_VISION_KEY>' --env GOOGLE_BOOKS_API_KEY='<GOOGLE_BOOKS_API_KEY>' -- npx -y thicken

Add to claude_desktop_config.json

{
  "mcpServers": {
    "thicken": {
      "command": "npx",
      "args": [
        "-y",
        "thicken"
      ],
      "env": {
        "THICKEN_VISION_KEY": "<THICKEN_VISION_KEY>",
        "GOOGLE_BOOKS_API_KEY": "<GOOGLE_BOOKS_API_KEY>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "thicken": {
      "command": "npx",
      "args": [
        "-y",
        "thicken"
      ],
      "env": {
        "THICKEN_VISION_KEY": "<THICKEN_VISION_KEY>",
        "GOOGLE_BOOKS_API_KEY": "<GOOGLE_BOOKS_API_KEY>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "thicken": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "thicken"
      ],
      "env": {
        "THICKEN_VISION_KEY": "<THICKEN_VISION_KEY>",
        "GOOGLE_BOOKS_API_KEY": "<GOOGLE_BOOKS_API_KEY>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "thicken": {
      "command": "npx",
      "args": [
        "-y",
        "thicken"
      ],
      "env": {
        "THICKEN_VISION_KEY": "<THICKEN_VISION_KEY>",
        "GOOGLE_BOOKS_API_KEY": "<GOOGLE_BOOKS_API_KEY>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "thicken": {
      "command": "npx",
      "args": [
        "-y",
        "thicken"
      ],
      "env": {
        "THICKEN_VISION_KEY": "<THICKEN_VISION_KEY>",
        "GOOGLE_BOOKS_API_KEY": "<GOOGLE_BOOKS_API_KEY>"
      }
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Configuration

  • THICKEN_VISION_URLAn OpenAI-compatible vision endpoint (optional): photos needn't be cropped
  • THICKEN_VISION_KEYThe vision endpoint's key
  • THICKEN_VISION_MODELSThe endpoint's model names, comma-separated, primary first
  • GOOGLE_BOOKS_API_KEYYour own Google Books quota, for book sizes
  • THICKEN_CONTRIBUTEContribute the photos you make models from to the corner model thicken uses: yes or no; ask, or unset, leaves it to thicken's own question (PRIVACY.md#contributing-photos)

What it does

Real-size GLB output
Models are built in millimetres from known object sizes.
Metadata sidecar
JSON with identification, catalog data and how each fact was found.
Profiles per object
Bundled profiles for book, card, slab, box, badge and lettering.
Code-based lookup
Reads ISBN barcodes and grader QR codes; ISBN-only mode needs no photo.
Optional vision models
Any OpenAI-compatible endpoint reads labels and printed text.
Batch mode
Makes a model per photo, PDF, video or folder.

Good for

  • Build a 3D shelf of your books from cover photos
  • Catalogue graded trading cards with cert numbers and grades
  • Make extruded 3D versions of badge artwork
  • Generate metadata for a collection from a photo folder
  • Let an agent make a model of an object on your desk

Pricing

Open source

Thicken FAQ

How do I run it as an MCP server?

Use npx -y thicken mcp on stdio, or install the Claude Code plugin from the bikemeardsley/thicken marketplace.

Do I need a vision model?

Only for uncropped photos, angled shots and video, or to read labels. Cropped straight-on photos work as they are.

What are the requirements?

Node 22.18 or later.

Can I add other kinds of object?

Yes. Each profile is one file in src/profiles and the README invites contributions.

Is Thicken free?

Yes. Thicken is open source, so you can use and self-host it for free.

How do I add Thicken to Claude Code?

Run this command in your terminal: claude mcp add thicken --env THICKEN_VISION_KEY='<THICKEN_VISION_KEY>' --env GOOGLE_BOOKS_API_KEY='<GOOGLE_BOOKS_API_KEY>' -- npx -y thicken

What are the best alternatives to Thicken?

The closest alternatives to Thicken are HyperFrames by HeyGen, next-ai-draw-io, mcp-for-blender, ffmpeg-skill and cursor-talk-to-figma-mcp.

Where this comes from

  • Description: written from the project's README.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers