Summary This repository curates practical, real-world workflows integrating Blender and Seedance for AI-driven filmmaking. It targets creators, researchers, and power users seeking to leverage 3D previs, camera control, and agent-assisted techniques for advanced video generation. The collection offers source-backed examples and tutorials, enabling users to explore and implement sophisticated AI video production pipelines.

How It Works The project bridges 3D scene creation and AI video synthesis by combining Blender's capabilities (previs, camera blocking, asset manipulation) with Seedance's generative power. Workflows leverage Blender MCP, AI agents (Codex, Claude), and diverse reference inputs like FBX, Mixamo, and even physical objects to guide Seedance. This approach prioritizes concrete, reproducible examples derived from public creator posts, offering practical control over camera movement, character actions, and scene elements for AI video generation.

Quick Start & Requirements

• Installation:
  1. Set up Blender MCP following official guides (specific URL not provided, refer to EvoLink documentation).
  2. Install EvoLink skills: npm i evolink-seedance and npm i evolink-topaz-video-upscale.
  3. Export API key: export EVOLINK_API_KEY="<your-evolink-api-key>".
• Prerequisites: Blender, Node.js (for npm), EvoLink API key.
• Resources: Comprehensive setup and execution guidance is available via the EvoLink Blender-to-video cookbook.

Highlighted Details

• Agentic Integration: Demonstrates AI agent (Codex, Claude) workflows for automated Blender scene creation, camera work, and reference video generation.
• Multi-Tool Pipelines: Showcases complex pipelines integrating tools like Hermes, Krea, ComfyUI, Blender MCP, and Seedance for comprehensive creative output.
• Diverse Reference Inputs: Utilizes a wide array of reference sources, including Blender previs, FBX, Mixamo animations, Midjourney start frames, and physical toy models.
• Style and World Variation: Explores preserving motion from Blender references while varying Seedance's render style (e.g., anime) or generating multiple distinct worlds from a single motion input.

Maintenance & Community This repository is actively maintained through community contributions, with creators encouraged to submit new use cases via issues or pull requests. A correction policy is in place for attribution accuracy.

Licensing & Compatibility The project is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0), which permits commercial use provided attribution is given.

Limitations & Caveats Specific workflows highlight limitations, such as challenges with precise foot motion control, the need for iteration on complex cloth physics, and the necessity of prompt reinforcement alongside reference videos for optimal Seedance generation.