This tool was made by Edouard Berard as part of a project conducted at the Glasgow School of Art, BA InteractionDesign.
It uses P5.js, ML5.js and the PoseNet machine learning model, the code was taken from The Coding Train by Daniel Schiffman.
This project explores the limit of computer vision to generate a font from your posture. The stiffer you are the sharper the font will be.
For best results please stand about 2 meters from your webcam, from head to toe.

The variable font is open-source and can be downloaded here.