PoPo addresses the challenge of generating expressive 3D character animations for MMD models from natural language. It targets animators and developers seeking an intuitive way to create poses, leveraging LLMs and a novel MMD-specific pose language (MPL) to produce anatomically correct and consistent results. The primary benefit is simplifying complex animation workflows through semantic AI interpretation.

How It Works

PoPo employs a fine-tuned LLM (GPT-4o-mini) to translate natural language prompts into MMD Pose Language (MPL) code. MPL is a structured, semantic description language designed for MMD, offering advantages over raw rotation quaternions by enabling better training convergence, consistent outputs, and anatomically correct poses through built-in constraints. This approach allows the AI to learn the "grammar" of human movement, resulting in debuggable and reliable pose generation. The process flows from text input to MPL code, which is then rendered in real-time using the Reze Engine.

Quick Start & Requirements

A live demo is available at popo.love. The project utilizes Next.js, shadcn/ui, and TypeScript for the frontend, with Reze Engine for 3D rendering. GPT-4o-mini is the AI model. Specific installation instructions or hardware prerequisites (like GPU requirements) are not detailed in the provided text, suggesting the demo is the primary way to experience the project.

Highlighted Details

• Generates MMD poses from natural language prompts like "wave right hand with big laugh, inviting me for dinner."
• Outputs semantic MPL code, which is human-readable and modifiable.
• Features real-time rendering for instant pose creation and smooth bone animations.
• Designed specifically for MMD anime characters, incorporating proper bone constraints and physics.

Maintenance & Community

No specific details regarding contributors, sponsorships, community channels (e.g., Discord, Slack), or a public roadmap were found in the provided README content.

Licensing & Compatibility

The project is licensed under the GPL-3.0 License. This is a strong copyleft license, meaning derivative works must also be licensed under GPL-3.0, which may impose restrictions on commercial use or integration into closed-source projects.

Limitations & Caveats

The provided README content does not explicitly state any limitations, alpha status, known bugs, or unsupported platforms.