FROM SLICES TO SPACES Design ideation on architectural models through AI-generated image sequences

The paper presents a novel methodology for applying AI-driven style transfer to complex 3D architectural models. By converting 3D models into 2D image sequences, the process integrates sequential slicing, training, video-guided diffusion and reconstruction to transform existing 3D models based on text, image, or video prompts into new stylised forms. This enables architects to explore diverse design concepts, focusing on spatial composition, visual appearance and tectonics through high-resolution outputs that capture both exterior and interior spatial relations. The results demonstrates the setups potential in enhancing early-stage design ideation through AI, by both outperforming existing video diffusion platform while also facilitating a fast exploration of different outcomes - capabilities which were validated in a design course. The study highlights an approach for utilising advanced 2D image-based AI models to generate intricate and meaningful 3D architectural transformations.