Blender MCP Tutorial Center

Create Stunning Minecraft-Style 3D Scenes

Create stunning Minecraft-style 3D scenes in just 10 minutes using natural language prompts, even with no prior Blender experience.

Step 1: Install MCPrep Plugin

First, you need to install the MCPrep plugin to access Minecraft assets in Blender:

1. Download the MCPrep plugin from theduckcow.com
2. In Blender, go to Edit > Preferences > Add-ons > Install
3. Select the downloaded .zip file and activate the plugin
4. Enable the plugin by checking the box next to "MCPrep"

Step 2: Set Up Blender MCP

Make sure Blender MCP is properly set up and connected to your AI model of choice:

1. Follow the steps in the "Getting Started" section to install and configure Blender MCP
2. Start the MCP server
3. Connect Blender to the MCP server

Step 3: Creating Minecraft Terrain

Use natural language prompts to create Minecraft-style terrain:

Example prompts:
"Create a Minecraft-style terrain with hills and a valley"
"Add a river flowing through the terrain"
"Create a forest with Minecraft-style trees"

Blender MCP will generate Python code to create procedural terrain using Perlin noise and other techniques to achieve that distinctive Minecraft look.

Step 4: Building Minecraft Structures

Create buildings and structures with these example prompts:

Example prompts:
"Build a small Minecraft house with wood and stone"
"Create a castle with towers and a moat"
"Add a mine entrance with torches"

Step 5: Adding Characters and Details

Enhance your scene with characters and details:

Example prompts:
"Add a Minecraft character like Steve near the house"
"Place some farm animals around the area"
"Add some furniture inside the house"

Combining Blender MCP with Server-Sequential-Thinking

Enhance your 3D modeling workflow by combining Blender MCP with other MCP tools like server-sequential-thinking. This powerful combination allows for complex modeling tasks with step-by-step planning and execution.

Example Workflow: Creating a High-Rise Building

Here's an example of how to use server-sequential-thinking with Blender MCP to create a complex high-rise building model:

Prompt: "I want to use Blender to create a high-rise building. Please use server-sequential-thinking tool to help me generate all the operation steps, then use blender-mcp to execute the generation. Remember to monitor each modification, and if there's anything unreasonable, make timely adjustments. Please also add a larger pavilion next to the existing one."

How It Works:

1. Planning Phase (server-sequential-thinking): The sequential thinking server breaks down the complex task into logical steps:
• Step 1: Create the base structure of the high-rise building
• Step 2: Add floors and windows to the building
• Step 3: Design the roof structure
• Step 4: Create the existing pavilion
• Step 5: Add a larger pavilion next to the existing one
• Step 6: Add environmental details and finishing touches
2. Execution Phase (blender-mcp): Blender MCP executes each step with appropriate Python code
3. Monitoring Phase: The system checks each modification for quality and makes adjustments as needed

Tool Comparison

Setting Up the Integration

1. Install server-sequential-thinking from GitHub
2. Configure it with your preferred LLM
3. Connect it to Blender MCP via API
4. Start using step-by-step planning for complex 3D modeling tasks

VSCode with Roo Code Configuration

VSCode integration with Roo Code allows you to interact with Blender MCP using an in-editor AI assistant:

1. Install the Roo Code extension in VSCode
2. Configure your API key (OpenRouter, Claude, or other supported providers)
3. Connect to the Blender MCP server
4. Use the chat interface to send natural language instructions to Blender

This setup allows you to send natural language instructions directly to Blender from VSCode, leveraging powerful language processing capabilities.

DeepSeek Integration

Blender MCP supports not only Claude but also other large language models like DeepSeek R1:

1. Get access to DeepSeek R1 through OpenRouter.ai
2. Select the DeepSeek model in your configuration
3. Leverage DeepSeek's powerful capabilities for 3D modeling

Free Usage Options

You can use Blender MCP for free with these methods:

Option 1: Using OpenRouter's Free Models

1. Visit OpenRouter.ai and create an account
2. Get your API key (new accounts come with some free credits)
3. Configure your API key and OpenRouter's API base URL in config.py
4. Choose free models such as:
   • google/gemini-2.0-flash-thinking-exp:free
   • deepseek/deepseek-r1-distill-llama-70b:free
   • qwen/qwen2.5-vl-72b-instruct:free

Option 2: Running Models Locally with Ollama

1. Install Ollama: Download and install Ollama from the official Ollama website (supports Windows, MacOS, and Linux)
2. Run the Ollama service (default port is localhost:11434)
3. Pull a model: Run ollama pull <model-name> in the command line to download your desired model, e.g., ollama pull llama3 or ollama pull qwen2.5:7b
4. Verify the model: Ensure the model was successfully downloaded by running ollama list
5. Configure Roo Code: Install the Roo Code extension in Visual Studio Code, open Roo Code settings, select "Ollama" as the API provider, and ensure the Base URL is set to http://localhost:11434 (Ollama's default address)
6. Enter your downloaded model name (e.g., qwen2.5:7b or whatever model you pulled)
7. Test the connection: Type a simple task in the Roo Code chat interface (e.g., "write a Python code") to check if it responds correctly