Acknowledgment
This project was created during GenAI Genesis 2025 (largest Canadian AI hackathon), where we won Best AI for Creativity & Generative Arts.
- Archie Shou — Three.js + 3D model integration (Shap‑E)
- Jessica Yunke Yi — idea + pitch
- Allen Lian — frontend + integration testing
Inspiration
We wanted to scale LEGO-based social-competence therapy into something more accessible and more engaging—without turning neurodivergence into “deficiency.” LEGO therapy assigns social roles and encourages interactive play, which can increase intrinsic motivation and reduce stress during sessions.
What it does
The generator takes an input image (e.g., a child’s drawing) and outputs a 3D, layer-by-layer LEGO build with role-playing instructions for a guide, supplier, and builder.
How we built it
- Front-end: React, Tailwind CSS, Three.js
- Back-end: Python (Flask API), Gemini API, Shap‑E, Modal
Challenges
- Brick allocation and placement (sorting + structural constraints)
- Colorization after voxelization
- Stitching together the pipeline (frontend ↔ backend ↔ model workflows)
Accomplishments
- Brick-placement algorithm leveraging border detection and structural integrity principles
- End-to-end integration across interfaces, APIs, and generated outputs
- Won Best AI for Creativity & Generative Arts at GenAI Genesis 2025