Image generation and editing with GenAI
This workflow leverages OpenAI's vision model (GPT Image-1) to assist an interior design with the creation and editing of photorealistic visualizations for room restyling.
The workflow beings by specifying a target room (e.g., private office) and a list of essential furniture items that must remain unchanged. A base image of the room is generated to reflect its current state.
Then, using parameterized prompts, the vision model produces three stylistically distinct reinterpretations (e.g., minimalist, industrial, bohemian) by editing only decorative elements, accessories, or layout—preserving the original furniture.
Each resulting image is saved with a style-specific filename (e.g., minimalist.png) and all outputs are displayed side by side in a table for easy comparison.
Note: Generating and editing images is significantly more expensive than text generation. Please refer to OpenAI Docs for more details on how the costs are calculated.