YouTube Demo: https://youtu.be/8JtZbi48wHM
What if you could describe something, like a chair or a cup, and have it physically created?
Think Copilot, but for the IRL.
Twirl, or Text-to-World IRL, is an AI agent for turning thoughts into reality via 3D printing.
Inspired by the theme of Perspectives, we set out to bridge the gap between imagination and reality. Twirl empowers anyone to create real-world objects without requiring technical expertise. By reimagining how we design and build, Twirl expands perspectives on creation, making the complex world of 3D modelling accessible for all.
Give any prompt, like "rocking chair with three back supports" and attach an image if you'd like.
Twirl will extract and infer features (ex. dimensions, object spatial relationships) and create a parameterized 3D CAD model instantly.
It'll allow you to continuously iterate on model generation until you have the perfect creation, making modifications with more prompts or images.
You can even adjust sliders and fields that change model variables like dimensions, colour, translations, and rotations. The entire generated CAD model is parameterized, allowing full customization within the web UI.
We render the 3D model live on the UI as it is created, allowing you to examine every pixel instantly.
Twirl doesn't just stop at visualization. It turns perspectives into physical reality. Models can be exported for 3D printing, making it possible to create functional objects directly from ideas. During the hackathon, we tested this end-to-end pipeline by designing and printing a chair model generated entirely by Twirl, with 0 human assistance.
With Twirl, we redefine the creative process, allowing everyone to see their ideas from a new perspective—one where imagination is the only limit.
