Shap-Explorer: Introducing Manipulable Text-to-3-D Generation Into 3-D Art Creation

Advisors: Daragh Byrne
Abstract: Generative Artificial Intelligence (GAI) technologies are increasingly employed by artists for their capability to rapidly generate intricate designs. While text-to-image (T2I) applications have been explored in art creation with constructive outcomes, the integration of text-to-3-D (T23D) methodologies remains underexplored, primarily due to the knowledge gaps between users and tool-makers. Developers aim to refine models for speed and accuracy, yet artists struggle to intuitively guide the generation process. To bridge this gap, this research introduces Shap-Explorer, a tool that streamlines the use of T23D, enabling users to incrementally control the generation and iteratively modify the output models. Through a series of user studies, this research examines the affordance of T23D and the impact of enhanced interactions on the generative system. With Shap-Explorer, the research provides insights into the manipulation capabilities of generative AI tools in the design workflow, offering a step forward in the interactive creation of 3-D art.