Inspiration
I struggled with completing YouTube origami tutorials! Sometimes the folds were unclear and it was hard to figure out where I went wrong.
What it does
Takes in a YouTube link, works with a physics simulator to generate the folds + render the folds. Then a live AI coach can walk you through building it and correct the mistakes by comparing your folds vs the simulation.
How we built it
Using Python, and extending some existing open source libraries on origami simulations.
Challenges we ran into
Simulating paper. Ran into memory exceeded errors when the paper was folded due to the number of collisions between paper particles. Simulating paper is surprisingly difficult. It's also difficult to track the paper folds when gemini is checking given the limited depth / noticeable shadows of paper with only one camera.
Accomplishments that we're proud of
Getting it working on some examples.
What we learned
How to work with a physics simulator, basics of building a harness.
What's next for Ori
Hopefully expanding beyond origami and lego!
Built With
- 3js
- gemini
- javascript
- mujoco
- python

Log in or sign up for Devpost to join the conversation.