BOgen: Generating Part-Level 3D Designs Based on User Intention Inference through Bayesian Optimization and Variational Autoencoder

Advancements in generative artificial intelligence (AI) have introduced various AI models capable of producing impressive visual design outputs. However, when it comes to AI models in the design process, prioritizing outputs that align with designers' needs over mere visual craftsmanship becomes even more crucial. Furthermore, designers often intricately combine parts of various designs to create novel designs. The ability to generate designs that align with the designers' intentions at the part level is pivotal for assisting designers. Hence, we introduced BOgen, which empowers designers to proactively generate and explore part-level designs through Bayesian optimization and variational autoencoders, thereby enhancing their overall user experience. We assessed BOgen's performance using a study involving 30 designers. The results revealed that, compared to the baseline, BOgen fulfilled the designer requirements for part recommendations and design exploration space guidance. BOgen assists designers in navigation and development, offering valuable design suggestions and fosters proactive design exploration and creation.