I’ve always been fascinated by how Large Language Models "understand" our world. To us, a poem is a sequence of words. To an AI, it's a complex map of coordinates in a massive, 1,536-dimensional space. I wanted to bring this invisible geometry to life. That’s why I built Poetry3D. The Concept: Every word has a position based on its meaning. In this project, I take those high-dimensional embeddings and "flatten" them into a 3D structure—much like a shadow of a tree. Meaningful connections: Words with similar vibes stay close, forming branches. Unique Fingerprints: No two poems look the same. A haiku creates a delicate shrub; an epic poem might grow into a sprawling oak. Export your art: You can download your unique tree as a PNG or an OBJ file for 3D printing or further digital art. Why I built this: I wanted to bridge the gap between cold math and human emotion. It’s an experiment in "Semantic Art"—seeing the structure of our thoughts through the eyes of an AI. I’d love to hear your feedback! What does your favorite poem look like? Which features should I add next? (I’m already thinking about sentiment-based coloring! 🎨) Can’t wait to see the forest of poems you all create today!

Poetry3D

Transforms poems into 3D semantic trees.

Transforms poems into 3D semantic trees.