AI Generation

Text to Image Models, Explained

Also called t2i, txt2img, text2img, what is text to image

Text to image is the generation task where a model turns a written prompt into a still picture with no reference image. A text to image model samples from what it learned rather than retrieving anything, so the same words with a different seed produce a different but equally valid image.

What the model is doing

Your prompt is encoded into a numeric representation of meaning. The model then starts from a field of random noise and repeatedly removes a little of it, at each step nudging the picture toward something that would plausibly be described by that representation. After a few dozen steps the noise has resolved into an image. Some newer systems predict image tokens in sequence instead of denoising, but the practical consequence is identical: you are drawing a sample from a distribution, not looking anything up.

That is why identical words can give completely different pictures, and why "the model got it wrong" is often better read as "the model gave you one of many valid answers". Rerunning is not a workaround, it is how the tool is meant to be used.

Where your control actually sits

  • The prompt. Order matters. Earlier tokens carry more weight, so lead with what you are least willing to lose.
  • The seed. The one lever that makes a result repeatable. Note the seed of anything you might want to revisit.
  • Aspect ratio. Not a crop. A 16:9 request and a 2:3 request from the same prompt produce different compositions, because the model was trained on differently framed pictures.
  • Reference images. The moment you add one you have left this task and moved into image to image.
  • Negative prompts. Useful for recurring artifacts, useless as a substitute for control the model does not have.

How to compare text to image models

Six axes, and almost no model leads on all of them:

AxisWhat to test with
Prompt adherenceA deliberately over-specified prompt, then count what survived
Text renderingOne short quoted string on a sign or label
AnatomyHands doing something, not hands at rest
Style rangeThe same subject as photo, ink drawing, and 3D render
Aspect and resolutionYour actual delivery ratio, not a square
Cost and latencyHow many attempts you can afford per idea

Cost belongs on that list because output variance is high by design. A model you can run eight times often beats a better model you can run twice.

A prompt structure that holds up

[subject] [doing what] in [environment], [lighting], [lens and framing], [medium or style]

Keep it to one subject. With two subjects, attribute binding starts failing: ask for a woman in a red coat and a man in a blue coat and you will regularly get the colors swapped, because the model has no reliable mechanism tying each attribute to each person. If you need two specific subjects, make them separately and combine with an editing pass.

Three habits that waste tokens: stacking quality adjectives (ultra detailed, 8k, masterpiece) which the model mostly ignores now; giving contradictory framing (a wide establishing shot that is also a tight portrait); and phrasing exclusions positively (writing "no hands" tends to summon hands, because the words are present either way).

The failure modes worth recognizing

Garbled lettering, miscounting anything above three, attribute bleed between subjects, unwanted symmetry, and style collapse toward the model's house look are all normal and none of them are fixed by writing more. Each has a specific answer: composite real type, split the subjects, change the seed, or pick a model whose default aesthetic is closer to what you want.

The prompt for this

A starting point that reliably produces the effect. Adjust the subject and setting; keep the technical clauses.

A lone lighthouse keeper on a wet stone jetty at dawn, low overcast light, 35mm, muted palette, shot from behind at eye level

Try Text-to-Image (T2I) yourself

Open the generator with a starting point already filled in.

Frequently asked questions

What is text to image, in one sentence?
It is a task where you describe a picture and a model produces it, with nothing but your words as input. Everything visible in the result was decided by the model, which is why t2i is the most open-ended and least repeatable of the image tasks.
What is the difference between t2i and i2i?
T2I has only your prompt to work from, so composition, subject, and style are all invented. I2I starts from an image you supply and modifies it, so layout and color are largely inherited. Use t2i to find an idea and i2i to change one you already have.
Is txt2img the same as t2i?
Yes. T2I, txt2img, and text2img all name the same task. The short forms come from open-model communities and appear in file names and node graphs; the spelled-out version is what product documentation uses.
Why does text inside my image come out garbled?
Most models treat letters as texture rather than symbols, so they reproduce the look of writing without spelling it. Adherence improves if you keep the string very short, put it in quotes, and ask for one text element only. For anything that must be exact, composite real type over the image afterward.
How do I get the same character in two different images?
You do not, with prompting alone. Fixing the seed only repeats a whole picture, not a subject inside a new one. Generate the character once, then use that image as a reference in an image to image or editing pass, which is what reference-image inputs exist for.
Why do two runs of the same prompt look different?
Because generation starts from random noise, and the seed selects which noise. A fixed seed plus a fixed prompt plus fixed settings is repeatable. Change any one of them, including the aspect ratio, and you get a new composition rather than a variation.

Related terms