Controlled GAN-Based Creature Synthesis via a Challenging Game Art Dataset -- Addressing the Noise-Latent Trade-Off

TL;DR

This paper improves StyleGAN2 for card art synthesis by suppressing coarse noise, enhancing control, and achieving better image quality, facilitating artistic creation in digital media.

Contribution

It introduces a modified StyleGAN2 that suppresses coarse-scale noise, reducing unwanted variations and improving control in art synthesis from a novel Yu-Gi-Oh dataset.

Findings

Superior FID scores achieved

Enhanced local style exploration with noise changes

Markedly improved identity control

Abstract

The state-of-the-art StyleGAN2 network supports powerful methods to create and edit art, including generating random images, finding images "like" some query, and modifying content or style. Further, recent advancements enable training with small datasets. We apply these methods to synthesize card art, by training on a novel Yu-Gi-Oh dataset. While noise inputs to StyleGAN2 are essential for good synthesis, we find that coarse-scale noise interferes with latent variables on this dataset because both control long-scale image effects. We observe over-aggressive variation in art with changes in noise and weak content control via latent variable edits. Here, we demonstrate that training a modified StyleGAN2, where coarse-scale noise is suppressed, removes these unwanted effects. We obtain a superior FID; changes in noise result in local exploration of style; and identity control is markedly improved. These results and analysis lead towards a GAN-assisted art synthesis tool for digital artists of all skill levels, which can be used in film, games, or any creative industry for artistic ideation.