SurfGen: Adversarial 3D Shape Synthesis with Explicit Surface Discriminators

TL;DR

SurfGen introduces an adversarial 3D shape synthesis method that directly models surface geometry using a spherical CNN, resulting in high-fidelity, diverse 3D shapes with explicit surface representations.

Contribution

It presents a novel framework that applies adversarial training directly on 3D surface representations via a differentiable spherical projection layer and spherical CNNs.

Findings

Produces high-fidelity 3D shapes with diverse topology

Outperforms existing voxel and point-cloud based models

Effective on large-scale shape datasets

Abstract

Recent advances in deep generative models have led to immense progress in 3D shape synthesis. While existing models are able to synthesize shapes represented as voxels, point-clouds, or implicit functions, these methods only indirectly enforce the plausibility of the final 3D shape surface. Here we present a 3D shape synthesis framework (SurfGen) that directly applies adversarial training to the object surface. Our approach uses a differentiable spherical projection layer to capture and represent the explicit zero isosurface of an implicit 3D generator as functions defined on the unit sphere. By processing the spherical representation of 3D object surfaces with a spherical CNN in an adversarial setting, our generator can better learn the statistics of natural shape surfaces. We evaluate our model on large-scale shape datasets, and demonstrate that the end-to-end trained model is capable of generating high fidelity 3D shapes with diverse topology.