3DSTNet: Neural 3D Shape Style Transfer

Abstract

In this work, we propose a 3D style transfer framework, 3DSTNet, to transfer shape or geometric properties from style to content 3D objects. We analyze the effects of multiple model hyperparameters on 3D style transfer. To evaluate the proposed 3D style transfer framework, we conduct a user study with 3D designers. Our evaluation results demonstrate that our approach effectively generates new designs and the generated designs aid in designers’ creativity.