An investigation of AI-based 3D model Generators on ship hull design at the conceptual design stage

Objectives The Ship hull design process severely relied on the experience and skills of designers, parent ship data, and foreign commercial CAD software. This study explores the application of AI-based 3D model Generators in ship conceptual design to address these limitations. Methods First, by inputting different requests from prompt textures and sketch drafts, the generation effects of four mainstream large models—Hunyuan 3D, Meshy, Rodin, and Tripo were analyzed. Second, a surface quality evaluation method based on slope detection was adopted to quantify the smoothness of the generated hull samples. Finally, the surface quality of the samples was improved using Laplacian and Taubin algorithms, enabling resistance calculations. Results By analyzing the resistance coefficients of the three candidate ship hull samples at Fr=0.20~0.30, the ship C2 with the lowest resistance originated from Rodin, which preliminary verified the feasibility of using AI-based 3D model Generators for ship hull design. Conclusions AI-based 3D model Generators have potential value in ship hull concept design, hydrodynamic performance analysis like resistance, and even AI-based hull optimization design frameworks.