Research on Generating Naked-Eye 3D Display Content Based on AIGC

Abstract

The lack of three-dimensional (3D) display content has posed significant obstacles to the widespread adoption and development of the industry. However, with the advancement of artificial intelligence (AI), Artificial-Intelligence-Generated Content (AIGC) is being increasingly applied to efficient 3D modeling workflows. his paper leverages mainstream AIGC 3D modeling techniques to efficiently create naked-eye 3D display content, focusing on exploring methods for producing such content using AIGC. Unlike traditional approaches, which often rely on extensive datasets or real-world data points, this research addresses the need for flexible and scalable methods to generate 3D subject-and-scene models with fewer data requirements, improving efficiency and lowering the entry barriers for content creation. First, this paper reviews the background and key techniques of AIGC in 3D model creation. Then, it presents several mainstream AIGC methods for generating subject-based 3D models and scene-based 3D models, analyzing and summarizing the strengths and limitations of each through a comparative evaluation of the modeling results. Additionally, this paper examines two workflows: text-to-3D and text-to-2D-image-to-3D, highlighting how different workflows impact 3D content generation. Finally, experiments converting 3D models into 3D display content validate the effectiveness of these methods, aiming to provide valuable insights for the production of naked-eye 3D display content.