What is Text to render (prompt to image)?
Generating an architectural scene from a written description alone, with no input image. Used for concept exploration before a site photo or model exists.
Updated 4 November 2026
Text to render, also called text to image or prompt to image, generates a picture from a written description with no input image. You describe the building, setting, materials, light and camera, and the model invents a scene that matches.
Why it matters
It is the fastest way to explore ideas before any drawing, photo or model exists. A designer can try ten directions for a hotel lobby in minutes, or a client can show what they have in mind. Because nothing is anchored to a real design, the images are inspiration, not a representation of a specific project.
Example
A prompt such as "narrow timber townhouse, deep window reveals, evening lights on, street view, 35 mm" produces a concept image. If it works, the designer sketches or models that direction and then uses an image-based tool to render the real geometry.
Common mistake
Treating the output as a design. The model has no site, regulations or structure in mind, so the building may be impossible to build. Another mistake is a prompt with no camera or lighting information, which gives flat, generic results.
Practical tips
Write prompts in the order a photographer would think: subject, setting, materials, light, lens. Generate several variations and pick by composition first, details second. Save the prompts that work so a series of images stays consistent. When a direction is approved, move to an image-based tool with a sketch or model so the final render reflects a real design.
In Archome AI
The Imagine AI tool in Archome AI generates scenes from text. Free accounts get 3 free renders, no card required, and a render takes about 30 seconds.