2021/08/19 by Garvita Tiwari, Nikolaos Sarafianos, Tiwari, Garvita +5 · 1 citation
Computer Science · Engineering · #3D Shape Modeling and Analysis #Advanced Vision and Imaging #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Human Motion and Animation
paper · pdf · doi:10.48550/arxiv.2108.08807
openalex publication_date 2021/08/19 · openalex created_date 2022/09/12 · openalex updated_date 2026/07/28
We present Neural Generalized Implicit Functions(Neural-GIF), to animate\npeople in clothing as a function of the body pose. Given a sequence of scans of\na subject in various poses, we learn to animate the character for new poses.\nExisting methods have relied on template-based representations of the human\nbody (or clothing). However such models usually have fixed and limited\nresolutions, require difficult data pre-processing steps and cannot be used\nwith complex clothing. We draw inspiration from template-based methods, which\nfactorize motion into articulation and non-rigid deformation, but generalize\nthis concept for implicit shape learning to obtain a more flexible model. We\nlearn to map every point in the space to a canonical space, where a learned\ndeformation field is applied to model non-rigid effects, before evaluating the\nsigned distance field. Our formulation allows the learning of complex and\nnon-rigid deformations of clothing and soft tissue, without computing a\ntemplate registration as it is common with current approaches. Neural-GIF can\nbe trained on raw 3D scans and reconstructs detailed complex surface geometry\nand deformations. Moreover, the model can generalize to new poses. We evaluate\nour method on a variety of characters from different public datasets in diverse\nclothing styles and show significant improvements over baseline methods,\nquantitatively and qualitatively. We also extend our model to multiple shape\nsetting. To stimulate further research, we will make the model, code and data\npublicly available at: https://virtualhumans.mpi-inf.mpg.de/neuralgif/\n