Nano Banana for Architecture
Explore Google's Gemini 2.5 Flash Image (Nano Banana) AI model and learn how to apply it to real architectural workflows: modifying renders, changing weather and materials, generating 3D model views, and converting Rhino exports into photorealistic images.
Dusan Cvetkovic walks through Google AI Studio and the Nano Banana model (official name: Gemini 2.5 Flash Image Preview), focusing on what actually works for architecture rather than generic image generation. The course is built around real back-and-forth with the model, showing honest failure cases alongside successful workflows.
Covered use cases include modifying exterior render lighting and time of day, changing weather conditions from a single reference image, swapping floor textures and interior plants, adding people and vegetation to scenes, converting Rhino viewport captures into photorealistic renders, generating multiple orthogonal views for a Hunyuan 3D pipeline, and converting rendered buildings into a physical architectural model aesthetic.
- How to navigate Google AI Studio, select Gemini 2.5 Flash Image Preview, and understand the temperature slider
- How to use the branching workflow to checkpoint successful prompts and iterate in multiple directions
- How to attach blank aspect-ratio reference images as a workaround to control output dimensions
- How to modify exterior render lighting, time of day, and weather conditions through iterative prompting
- How to replace architectural materials in interior renders using follow-up prompts
- How to add and remove scene elements using annotated reference images with colored markers
- How to capture a Rhino viewport at high resolution and prompt the AI to apply realistic materials and environment
- How to generate orthographic views of a building from a single perspective image for input to a 3D generation pipeline
