Why generate the character alone before any scene?

Because you are building a reference, not an illustration. A plain single figure carries forward far more reliably than a character embedded in a busy scene.

How many defining details should I list?

Three to five countable ones. Enough to be recognisable, few enough that you will actually repeat all of them in every later prompt.

Can I just say "same character as before"?

It carries some context but drifts. OpenAI's guidance is to restate critical constraints, and in practice restating the countable details is what holds a series together.

Pass this image as the reference to the continue-the-story template, which describes a new scene while repeating the same appearance constraints.

GPT Image 2.5 prompt — a character reference sheet you can carry across scenes

Generate the character on a plain background first. That image becomes the reference for every page after it.

A watercolour storybook character in a green hooded tunic with a small wooden bow, against a plain forest background

Settings the Design a character you can reuse on every page prompt ran at
Mode	Model	Quality	Size	Ratio	Credits
Text to Image	Think	Standard	1K	2:3	6 cr

Published by OpenAI as an example for this model — not generated on this site. Model: gpt-image-2.5-sunburst. Source: OpenAI The published example ran at 1024x1536, a legal size we do not currently offer; the frame here is the same shape at our nearest size.

A watercolour storybook hero, introduced against a plain forest background with nothing else competing for attention. This is step one of a two-step workflow: you are not making an illustration here, you are making a reference. The plain background, the single figure and the explicit list of defining details all exist so the *next* prompt has something unambiguous to point at.

In full

The prompt, nothing cut

Create a children's book illustration introducing a main character.

Character:
A young, storybook-style hero inspired by a little forest outlaw,
wearing a simple green hooded tunic, soft brown boots, and a small belt pouch.
The character has a kind expression, gentle eyes, and a brave but warm demeanor.
Carries a small wooden bow used only for helping, never harming.

Theme:
The character protects and rescues small forest animals like squirrels, birds, and rabbits.

Style:
Children's book illustration, hand-painted watercolor look,
soft outlines, warm earthy colors, whimsical and friendly.
Proportions suitable for picture books (slightly oversized head, expressive face).

Constraints:
- Original character (no copyrighted characters)
- No text
- No watermarks
- Plain forest background to clearly showcase the character

829 characters

Run it here

This exact prompt, at these exact settings

The mode and all five settings are locked to this template, so there is nothing to line up by hand. Press Run this prompt above and the panel opens filled in — prompt, mode, quality, size and frame — and your first image needs no account. If your plan sits below what this ran at, the line under the panel says so.

This example was published at Think · Standard · 1K. A free run is Fast · Standard · 1K, so expect less fine detail and softer small text. The composition is the same; the finish is not.

3 levers

What is safe to change

Most libraries publish this list and stop, which is why so many copied prompts come back worse than the original: the swappable nouns are the safe half. The three sections under it are the other half — the clauses that are doing the work, the ones that break it if you touch them, and why it is written in this order.

The plain background constraint
"Plain forest background to clearly showcase the character" is not an aesthetic choice. A busy first image makes a worse reference, because the model then has to work out which parts of it were the character.
The countable details
Green hooded tunic, brown boots, belt pouch, wooden bow. Four specific, checkable items. These are what you repeat verbatim in every later prompt — vague traits like "friendly-looking" cannot be repeated usefully.
The proportion note
"Slightly oversized head, expressive face" fixes the drawing convention. Without it the register drifts between picture-book and young-adult across a series.

Three ways to break it

Each one is a real failure with a reason attached. A rule without a reason is not usable.

01Putting the character in a scene on the first pass
You lose the reference. A character mid-action in a detailed environment is much harder to carry forward than the same character standing plainly, because there is more ambiguity about what is the character and what is the world.
02Naming a copyrighted character as shorthand
The prompt says "inspired by a little forest outlaw" rather than naming one, deliberately. Naming the character is both a policy problem and a quality problem — you inherit a specific design you then cannot legally use.
03Describing the character with feelings instead of features
"Brave and kind" cannot be checked against the next image. "Green hooded tunic and a small belt pouch" can. Consistency is enforced by repeating checkable nouns, so the reference has to be made of them.

Why it is written this way

Inside the Design a character you can reuse on every page prompt

Character consistency is the thing people most want from an image model and most often fail to get, and the reason is almost always workflow rather than capability. They write one prompt per illustration, each describing the character afresh, and are surprised when twelve prompts produce twelve characters. Each generation is an independent sample; nothing carries between them unless you carry it.

So this template is deliberately step one of two. The output is not meant to be a page in the book. It is a reference image, and every decision in the prompt serves that purpose. The background is plain because a plain background makes an unambiguous reference. There is one figure because a reference with two figures is a reference to what, exactly. There is no text because text in a reference gets carried into the next image as texture.

The character description is built out of countable nouns: a green hooded tunic, soft brown boots, a small belt pouch, a small wooden bow. That is the crucial property. When you write the second prompt you will repeat those four items verbatim, and after generating you will look at the image and confirm all four are present. Neither of those is possible with "a brave and gentle young hero" — you cannot repeat it usefully and you cannot check it.

The style block does the same job at a different level. "Hand-painted watercolor look, soft outlines, warm earthy colors" plus the proportion note gives you a drawing convention you can restate. In a series, style drift is as damaging as character drift and much easier to miss until you see the pages side by side.

The phrasing "inspired by a little forest outlaw" is worth noticing rather than skipping. It reaches for a familiar archetype without naming a character anyone owns. Naming one would have been shorter and would have worked, and it would also have produced a design that is somebody else's. For anything intended for publication, the constraint line "Original character (no copyrighted characters)" is not boilerplate.

The companion template — continuing the story with the same character — is the other half of this. Neither is much use alone.

4 questions

Design a character you can reuse on every page — common questions

Why generate the character alone before any scene?: Because you are building a reference, not an illustration. A plain single figure carries forward far more reliably than a character embedded in a busy scene.
How many defining details should I list?: Three to five countable ones. Enough to be recognisable, few enough that you will actually repeat all of them in every later prompt.
Can I just say "same character as before"?: It carries some context but drifts. OpenAI's guidance is to restate critical constraints, and in practice restating the countable details is what holds a series together.
What comes next?: Pass this image as the reference to the continue-the-story template, which describes a new scene while repeating the same appearance constraints.