Text-Driven Complex 3D Shape Generation using GAN for Information Systems

In the domain of creative content generation, our review focuses on the text-driven generation of complex 3D shapes via Generative Adversarial Networks (GANs). The primary objective is to empower users to effortlessly convey their 3D object ideas through textual description, which can then be automatically translated into ready to load models. This review paper explores the evolving landscape of creative content generation, with a specific focus on the text-driven synthesis of intricate 3D shapes using Generative Adversarial Networks (GANs). The primary objective is to empower users to effortlessly articulate their 3D concepts through textual descriptions, subsequently translated into ready-to-use 3D models. Key features addressed include text-to-3D translation, model flexibility across diverse categories, and the synthesis of highly detailed 3D shapes. This paper includes insights on established techniques such as 3D GANs and 3D Diffusion models. Evaluation metrics not only include the accuracy and coherence between generated shapes and user descriptions but also the rendered resolution, measured through Frechet Inception Distance (FID) and Inception Score (IS). These methods can be used for businbess applications where GAN and 3D shape Generation has been involved.