Event description
Abstract: The realization of animated 3D characters has always been expensive regarding resources and (human) time. Automatic methods for building high-quality shapes with the desired characteristics for the animation are complex to design. In the last decade, acquiring 3D models from the real world has become within everyone's reach.
This talk presents a complete pipeline to combine realistic geometries acquired from the real world and the desired properties to create an animation-ready model using a modeling-by-matching technique. The presentation will explore the advantages of using data-driven techniques in 3D clothes modeling and the challenges in the Neural Network design for a 3D world, expecially in the segmentation and classification tasks. We will face the issues of building the datasets and the difficulties in choosing the right representation for massive trainings.
With such techniques, we obtain a fully automatic pipeline to provide an animated 3D setup starting from 3D scans, providing new approaches for modeling and animation.