Passing Multi-Channel Material Textures to a 3-Channel Loss

TL;DR

This paper introduces a method to compute neural textural losses for multi-material channels by passing random triplets to a 3-channel loss, enabling high-quality texture generation for physically based rendering.

Contribution

It presents a novel approach to extend 3-channel neural losses to multi-channel textures using random triplets, overcoming pretrained model limitations.

Findings

Effective multi-channel texture generation demonstrated

High-quality textures achieved for various material channels

Method compatible with existing neural loss frameworks

Abstract

Our objective is to compute a textural loss that can be used to train texture generators with multiple material channels typically used for physically based rendering such as albedo, normal, roughness, metalness, ambient occlusion, etc. Neural textural losses often build on top of the feature spaces of pretrained convolutional neural networks. Unfortunately, these pretrained models are only available for 3-channel RGB data and hence limit neural textural losses to this format. To overcome this limitation, we show that passing random triplets to a 3-channel loss provides a multi-channel loss that can be used to generate high-quality material textures.