The video is based on the idea that any image or object can be interpreted only within the previously obtained visual experience. Technically, this was implemented using the CGAN neural network, trained on a photos of the facades of houses and buildings. The movement and repositioning of objects serves as a starting point for the generation of various textures and patterns, interwoven between each other in search of a new geometry. The constant change in the building reflects the impossibility of choosing one specific design, and such inconstancy makes it possible to go beyond the limits of statics into a dynamic architecture.