Apppublic

Agents-MCP-Hackathon/ai-image-video-mcp-server

sourceHugging Facemitupdated 1y agoView on Hugging Face

1likes

App README

I made this for myself to do bulk image editing (Background removal, upscale, edit) and products marketing/demo video generation agent. Only works locally because we can't pass images from Claude MCP Client to MCP server without passing a local path. If you are a jury please contact me @shdkhasan on X.com so i can give you the API keys to test locally. API keys below are required:

MODALLABSKEY=xxxx MODALLABSENDPOINT="https://xxxx.modal.run" GEMINI_API=XXXX

📺 Demo Video

👉 Watch how it works (Loom)

🔧 Tools Available

1. Remove Background

—Model: BiRefNet v2 (hosted on Modal Labs)
—Input: Image with background
—Output: Transparent PNG

2. Upscale

—Tile Upscaler: Highly accurate enhancer using tiled upscaling (hosted on Modal Labs)
—Clarity Upscaler: General quality enhancer (calls external Gradio Space API)

3. Image-to-Video

—Model: Google Veo 2
—Input: Image + Prompt
—Output: Cinematic video clip (5–8 sec)
—⚠️ Note: Image must be visually coherent; typically used after background editing

4. Edit Image with Gemini

—Model: Gemini 2.0 Flash Preview Image Generation
—Purpose: Add backgrounds/scenes to background-removed subjects
—✅ Important: Prompt must specify to not alter the subject, only modify the environment.

🧑‍💻 How to Use With Claude Desktop (MCP)

To use this space as an MCP server:

1.Download [Claude Desktop](https://claude.ai)
2.In Claude's MCP config, add this server and filesystem:

json

{
  "mcpServers": {
    "gradio": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://127.0.0.1:7860/gradio_api/mcp/sse"
      ]
    },
    "filesystem": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "C:\\Users\\YOUR_USERNAME\\Desktop\\claude-accessible-folder"
      ]
    }
  }
}

🗂️ Replace YOUR_USERNAME with your actual Windows username. Make sure the folder claude-accessible-folder exists on your Desktop. Claude will use it to share image/video files with the tools.