Stockertim Dictionary · Artificial intelligence
Text-to-image model
Also called text-to-image, text-to-image generator, image generator
In short
An AI model that generates a new image from a written description, starting from random noise rather than from an existing picture.
- Beginner
- Technology
- Checked
In detail
A text-to-image model takes a written prompt, such as a description of a scene, and produces a picture that tries to match it. Most current models are diffusion models: they start from random noise and refine it into an image guided by the meaning of the words. The model has learned links between words and visual features from a very large collection of captioned images.
The words in the prompt carry most of the control. A useful image prompt usually covers the subject, setting, composition, lighting, style, medium and aspect ratio. Many tools also accept a negative prompt listing what to leave out, and some accept a reference image. This differs from image-to-image work, where an existing picture is the starting point and the prompt describes how to change it.
Text-to-image models are fast and can produce many options in minutes, which makes them useful for ideation. Their limits matter too. They can misread complex instructions, mix up which object gets which colour, invent garbled text and reflect stereotypes from their training data. Ownership and permitted use of the output depend on the tool's licence terms and on the law where the work is used.
In design work, people use these models for mood boards, concept sketches, storyboards and placeholder imagery, then refine, edit or redraw the chosen direction.
Worked example
A designer at Harbour Foods needs a picture for a summer menu.
- Weak prompt: "food on a table".
- Improved prompt: "overhead photo of a grilled fish salad on a pale wooden table, soft morning light, plenty of empty space at the top for a headline, square format".
- The improved prompt names the subject, view, lighting, mood and space for text, so fewer rounds are needed.
The second prompt gives a usable concept image in two rounds instead of six.
In practice
Teams use text-to-image models for early concepts, internal mock-ups and campaign exploration. Agencies record which images were generated, check the tool's licence terms, avoid prompts that name living artists and treat final assets with the same rights and brand checks as any stock image.
Watch out
- Text-to-image means generating from words alone; image-to-image starts from an existing picture.
- A realistic result is not a photograph of a real event and should not be presented as one.
Learn it properly
Official sources
- Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile · National Institute of Standards and Technology
- Copyright and artificial intelligence · US Copyright Office
Facts last checked on 3 November 2026.
Education, not advice: the Stockertim Dictionary explains what terms mean. It never recommends buying or selling anything.