Meta Style: The Visual Maze of AIGC Art

oleh: Gao Jie, Yin Jun

Format: Article
Diterbitkan: EDP Sciences 2024-01-01

Deskripsi

This article aims to explore whether the visual style in AIGC art is constrained by certain factors that hinder the emergence of new styles and to analyze possible paths beyond this limitation. Methodologically, the article first deconstructs and subverts established visual rules in AIGC art from a post-structuralist perspective. Subsequently, it analyzes the limitations that AIGC art currently faces in innovation, namely, excessive reliance on imitating existing styles, from a perspective of technological philosophy. Finally, it proposes breakthroughs in style innovation for AIGC art through diversified training data, human-machine collaboration, and assimilating multicultural resources. The conclusion suggests that there is potential to surpass the 'visual maze' in AIGC art, but it requires progress in various aspects such as data, models, human-machine interaction to achieve this breakthrough. This will not only promote the development of AIGC art itself but also enrich the aesthetic experience of humanity.