Architectural visualization has always been about storytelling. Helping see what doesn’t exist yet. Traditionally, this meant hand-drawn renderings that took days or weeks to complete. Then came 3D modeling and rendering software, which made images more precise but still demanded countless hours of modeling, material tweaking, and Photoshop post-processing. More recently, real-time rendering engines from gaming brought unprecedented speed and interactivity to BIM models.
Now AI image generation is here and it's getting real. I took a rendering of a proposed building and asked ChatGPT’s image tool and Nano Banana to place it into a site photo where an existing building is slated for demolition.
First round. Upload two images and ask AI to combine.
Results:
Scale is off. The number of stories isn’t correct.
Second round. Same prompt with a street view that matches the rendering and scale the images:
Surprisingly good. Is this the future of architectural visualization? No more rendering or photoshop. Just screen grabs and prompts.