← ArchivePaper2023

D-Cloth: Skinning-based Cloth Dynamic Prediction with a Three-stage Network

Yudi Li, Min Tang, Yun Yang, Ruofeng Tong, Bailin An, Shuangcai Yang, Yao Li, Qilong Kou

CGFAcademic9 citesCFX ML Deformation Skinning

Decomposes cloth deformation into static skinning, coarse dynamic, and wrinkle components predicted by three sequential network stages.

Open the paper ↗Trace in the lineage graph

Abstract

We propose a three‐stage network that utilizes a skinning‐based model to accurately predict dynamic cloth deformation. Our approach decomposes cloth deformation into three distinct components: static, coarse dynamic, and wrinkle dynamic components. To capture these components, we train our three‐stage network accordingly. In the first stage, the static component is predicted by constructing a static skinning model that incorporates learned joint increments and skinning weight increments. Then, in the second stage, the coarse dynamic component is added to the static skinning model by incorporating serialized skeleton information. Finally, in the third stage, the mesh sequence stage refines the prediction by incorporating the wrinkle dynamic component using serialized mesh information. We have implemented our network and used it in a Unity game scene, enabling real‐time prediction of cloth dynamics. Our implementation achieves impressive prediction speeds of approximately 3.65ms using an NVIDIA GeForce RTX 3090 GPU and 9.66ms on an Intel i7‐7700 CPU. Compared to SOTA methods, our network excels in accurately capturing fine dynamic cloth deformations.

How to read this

Category: Method: skinning-based learned cloth dynamics prediction
Contributions: Decomposes cloth deformation into static, coarse dynamic, and wrinkle dynamic components predicted by a three-stage network
Builds a static skinning model with learned joint and skinning-weight increments, then adds coarse dynamics from serialized skeleton information
Refines wrinkle detail from serialized mesh information and runs in real time, demonstrated in a Unity scene on both GPU and CPU
Context: Sits in the neural cloth deformation line of work, building on Neural Cloth Simulation (Bertiche et al. 2022) with an explicit skinning-based decomposition.Builds on: Neural Cloth Simulation
Correctness: Reports real-time speeds and finer detail than prior methods, but the staged decomposition assumes deformation separates cleanly into static, coarse, and wrinkle parts, and the demonstrated speeds are tied to the cited hardware.
Clarity: Reasonably clear given the explicit three-stage structure; a first pass conveys the pipeline, a second pass is needed for the per-stage inputs and increments.
How to read it: Focus on what each of the three stages consumes and predicts; a second pass pays off for the learned skinning increments and how serialized skeleton vs mesh data drive coarse vs wrinkle detail.

Builds on

Neural Cloth Simulation 2022

Built upon by

Nothing yet.

Related work

NeuroSkinning: Automatic Skin Binding for Production Characters with Deep Graph Networks 2019 / SIGGRAPH
A Statistical Model of Human Pose and Body Shape 2009 / CGF
NiLBS: Neural Inverse Linear Blend Skinning 2020 / arXiv
Learning Skeletal Articulations with Neural Blend Shapes 2021 / SIGGRAPH

Keywords

cloth dynamic prediction skinning-based model three-stage network linear blend skinning skinning weights graph sequence gru gru wrinkle dynamics skeleton sequence mesh sequence dynamic deformation graph convolutional network penetration handling arcsim skeleton-based characters loose-fitting garments real-time prediction

This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →