Artificial intelligence can now turn a simple architectural sketch into a photorealistic visualization, create 3D models and write algorithms with which to develop individual elements of buildings. Arch. Radul Shishkov spoke about the new possibilities of the technology during the conference “Creation for the Future - 100 Years of the Birth of Prof. Dr. Eng. Ganyu Shtilyanov - the Visionary with a Cause“ in a presentation dedicated to the application of AI in architectural design.
According to him, artificial intelligence finds a place in both sides of architecture - the creative and the technical. In the first, it can generate images and designs, and in the second - create code and algorithms for solving specific design tasks.
One of the most developed applications is already the creation of architectural visualizations. The process is significantly faster, as a simple sketch can be transformed into a realistic image without changing the geometry and position of the camera. AI can also help in presenting architectural ideas through diagrams and graphs.
The technology is also trying to create plans and layouts of rooms, but there are still limitations here. “It can create relatively reliable layouts, but certainly far from perfect“, Shishkov pointed out.
The next step is significantly more interesting for practical design - the image created with AI can be converted into a 3D model, which can then be refined with algorithms to precise geometry. Thus, artificial intelligence is gradually connecting with the parametric design tools already used by architects.
New technologies allow AI to work with professional architectural software. Through the Model Context Protocol (MCP), artificial intelligence can use the functions of other programs. This opens up possibilities for application in the creation of concepts, technical developments, documentation, visualizations and project coordination.
A particular strength of modern AI models, according to Shishkov, is writing code. Artificial intelligence can create an algorithm that sets the geometry of a given object. Thus, instead of the architect receiving only one ready-made 3D model, he also has code whose parameters can be changed and checked. In this way, different variants of facades, layouts and other parts of the building can be developed.
The Faculty of Architecture of the University of Applied Sciences and Arts of Bulgaria is already conducting experiments with students and doctoral students, including with the 3D modeling program Rhino. The goal is to connect different AI technologies in a single process - from the initial image to the three-dimensional model and its subsequent refinement.
According to Shishkov's presentation, four technologies - image generation, MCP protocols, procedural and algorithmic modeling - are gradually starting to work together. It is their combination that has the potential to change the way architects create and develop their projects.
Detailed statistics on average property prices in Bulgaria by city and neighborhood can be found at imot.bg