What is it about?

Our work presents an AI-based solution to predict sewing patterns of garments from their 3D shape. We want to enable the processing of garments as worn by humans and captured by 3D sensors: with all the design variations, wrinkles, stretching of the fabric, and scanning artifacts. With our NeuralTailor, we can accurately recognize garments' original sewing patterns from those complex inputs and even do so for garment types that our AI has never seen at training time!

Featured Image

Why is it important?

Reconstructing sewing patterns of garments enables easy usage of those garments in novel conditions -- e.g. dressing new characters and applying new animations as well as new material properties or appearance -- starting from just a static deformed garment! Obtaining this control ability in a simple and scalable manner is a critical problem in SocialVR and Virtual Try-On. Such a technology would allow people to bring their favorite physical garment pieces with them to the metaverse, representing their true selves in virtual social interactions, or experimenting and sharing their garments with others =)

Read the Original

This page is a summary of: NeuralTailor, ACM Transactions on Graphics, July 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3528223.3530179.
You can read the full text:




The following have contributed to this page