Analysis of Texture Synthesis Algorithms in Computer Graphics
Vladimir V. Gurenko, Vyacheslav Sergeevich Eliseev, Maxim Usmanov, Maria S. Selezneva · 2022 Conference of Russian Young Researchers in Electrical and Electronic Engineering (ElConRus) · 2022
The use of computer graphics methods is associated with the ongoing problem of texture synthesis. There are two groups of approaches to this problem. The first one is the creating textures by a digital artist using such software tools as Substance Painter. The second group includes automatic methods based on specialized algorithms. Some examples include synthesis using non-parametric sampling, as well as machine learning algorithms based on PSGAN. This paper provides a comparative analysis based on quality and performance criteria of texture synthesis algorithms. It is worth to mention that the difficulty of assessing the quality of synthesis is caused by the lack of a formal quantitative metric. The authors made conclusions about the preferred fields for application of the analyzed algorithms. The results of this survey can be used to optimize the production activities of digital art studios in terms of the time it takes to make textures.