SceneCraft

Transforms text into storyboard plans and optional vertical video via an MCP-compatible workflow.

python

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "snippetwizard-scenecraft_mcp": {
      "command": "python",
      "args": [
        "-m",
        "scenecraft_mcp.mcp_server"
      ],
      "env": {
        "LLM_PROVIDER": "mock",
        "OLLAMA_MODEL": "llama3",
        "OPENAI_MODEL": "gpt-4.1-mini",
        "OPENAI_API_KEY": "sk-...",
        "OLLAMA_BASE_URL": "http://127.0.0.1:11434",
        "LOW_LATENCY_MODE": "true"
      }
    }
  }
}

SceneCraft MCP turns plain text into a visual plan (storyboard) and optionally an assembled vertical video. It provides a lightweight, extensible workflow you can drive from local code or an MCP-compatible agent, enabling fast script-to-screen prototyping and automation-ready outputs.

How to use

You can use this MCP server by connecting an MCP-compatible client and invoking the create_storyboard tool. You supply a script or scene text, and the system returns a structured storyboard with scene and shot data. You can then generate visuals frame-by-frame or assemble a draft video if you opt into the full pipeline.

Typical usage patterns include: creating a storyboard from a single scene text, then optionally producing per-shot frames with a frame generator and stitching those into a vertical video. The tool interface is designed to be orchestrated by LLMs or other automation clients so you can automate your planning and execution flow.

How to install

Prerequisites: you need Python and a working networked environment. Optional components for frames and video require additional dependencies.

Install core dependencies and the MCP server package.

pip install -r requirements.txt

If you want optional frame generation and video assembly, install those dependencies as well.

pip install moviepy diffusers
# Then install PyTorch for your system, CPU-only example:
# torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu

Start the MCP server component that exposes the create_storyboard tool.

pip install mcp
python -m scenecraft_mcp.mcp_server

Configuration

Use the configuration options to select your LLM provider and performance settings. You can run offline for demos or connect to a remote LLM service for production usage.

Key environment variables you will encounter include provider selection and model settings.

Example environment setup (conceptual):

# LLM Provider (openai / ollama / mock)
LLM_PROVIDER=mock

# OpenAI
OPENAI_API_KEY=sk-your-key
OPENAI_MODEL=gpt-4.1-mini

# Ollama
OLLAMA_MODEL=llama3
OLLAMA_BASE_URL=http://127.0.0.1:11434

# Performance
LOW_LATENCY_MODE=true

Using the MCP server (start and connect)

You can connect an MCP-compatible client to the local server to call tools like create_storyboard. The server exposes a tool that accepts a script and returns a project_id and related counts.

Start the server, then connect your client using the MCP protocol. The following command starts the server in your environment.

pip install mcp
python -m scenecraft_mcp.mcp_server

Examples and outputs

Storyboard JSON is saved under your project directory as a file you can inspect or reuse for rendering.
Optional frames and video outputs appear in designated folders when you enable the full pipeline.

Available tools

create_storyboard

MCP tool that accepts a script and optional title and returns a structured project with counts for scenes and shots, enabling end-to-end storyboard generation via MCP clients.

Built by

VeilStrat

AI signals for GTM teams