On Digital Art Generation Using Generative Adversarial Networks

Document Type

Conference Proceeding

Source of Publication

International Conference on Electrical, Computer, and Energy Technologies, ICECET 2024

Publication Date

1-1-2024

Abstract

Art has cultivated human imagination across cultures for centuries. The peak of human creativity is fully expressed by artists through artworks. With technological advancement, researchers have pondered over the possibility of teaching computers to create arts. Although earlier computer arts were uniform and lacked creativity, there has been a recent shift in computer arts with the introduction of generative adversarial networks (GANs). GANs have generated photo-realistic human faces and other images that are virtually indistinguishable from real images. Although GANs have been used to generate various art forms, including cartoons and NFTs, their application to abstract and landscape or nature arts have been limited. This paper evaluates and compares two arts generation models through an end-user survey. It explores the possibility of GANs in generating abstract and landscape arts by using two different GAN architectures. Results from a case study indicate that the GAN generated artworks are more interesting to the volunteers than those generated by real artists. The GAN generated artworks are also on part with the real artworks in terms of overall quality.

ISBN

[9798350395914]

Publisher

IEEE

Disciplines

Computer Sciences

Keywords

computer arts, deep learning, digital art generation, GAN

Scopus ID

85207398122

Indexed in Scopus

yes

Open Access

no

Link to Full Text

Share

COinS