GPT Image 2.5 prompt — continue a story with the same character

New scene, same four garments. The consistency block is copied forward word for word.

The same character kneeling beside a frightened squirrel in a snowy forest after a storm
The established character reference from the previous template
InputResult
Settings the Draw the next page without the character changing prompt ran at
ModeModelQualitySizeRatioCredits
Image to ImageThinkStandard1K2:310 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.

The forest hero from the previous template, now kneeling beside a frightened squirrel after a winter storm. The scene is entirely new; the character description is copied from the first prompt without a word changed. That repetition is not laziness — it is the mechanism. A model has no memory between requests, so continuity is something you carry, not something you invoke.

In full

The prompt, nothing cut

Continue the children's book story using the same character.

Scene:
The same young forest hero is gently helping a frightened squirrel
out of a fallen tree after a winter storm.
The character kneels beside the squirrel, offering reassurance.

Character Consistency:
- Same green hooded tunic
- Same facial features, proportions, and color palette
- Same gentle, heroic personality

Style:
Children's book watercolor illustration,
soft lighting, snowy forest environment,
warm and comforting mood.

Constraints:
- Do not redesign the character
- No text
- No watermarks
569 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.

PNG, JPEG or WebP · up to 25 MB

Up to 16 images. The first four are included; each one after that costs 3 credits.

0 / 4000
Keep intact — tick anything the model must not touch

We turn your ticks into OpenAI's own "change only X" phrasing, then re-render from your original file — not from the last output.

No card, no waitlist. Credits only start after the free one.

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 consistency block

    Copied forward from the establishing prompt, unchanged. Copy it into every page of the series — the moment you start paraphrasing it, the character starts drifting.

  • The scene block

    The only part that changes between pages. Keeping the changing part and the fixed part in separate labelled blocks is what makes a series maintainable.

  • "Do not redesign the character"

    A blunt instruction that catches what the itemised list misses. The list covers the tunic and the face; this covers everything you did not think to itemise.

Three ways to break it

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

  1. 01Writing "same character as before" and nothing else

    It carries some context and drifts steadily. OpenAI's guidance is that references like this can carry context but critical constraints should be restated if the result drifts — and across a picture book, it will.

  2. 02Paraphrasing the consistency block

    Every rewording is a small nudge. "A green hood" is not "the same green hooded tunic", and ten pages of small nudges is a different character by the end.

  3. 03Chaining page 5 off page 4 off page 3

    Drift compounds. Always reference the original establishing image, not the most recent page, or you are doing a slow game of telephone with your own protagonist.

Why it is written this way

Inside the Draw the next page without the character changing prompt

This is the second half of the character workflow and the half that does the actual work. The first prompt made a reference; this one uses it, and the technique is almost embarrassingly simple: attach the reference image, describe the new scene, and copy the character description across word for word.

The block structure is what makes it maintainable. Scene, Character Consistency, Style, Constraints. Only the Scene block changes between pages. The other three are copied. That means a twelve-page book is twelve short scene descriptions plus one shared boilerplate, which is both less work and more reliable than writing twelve full prompts — and it is far easier to audit, because you can diff two pages and see exactly what differs.

The reason the repetition is necessary is that there is no session state. Each request is independent, and "the same character" is only meaningful because an image is attached and a description is restated. OpenAI's guidance acknowledges that a reference like "same style as before" can carry context, but is careful to add that critical constraints should be restated if the result drifts. Across a picture book, treat that as "will drift" rather than "if".

The most valuable single line is the last constraint: "Do not redesign the character." The itemised list handles the things you thought of — tunic, face, proportions, palette. This handles the things you did not. Boot height, pouch position, bow size, the exact green. A model regenerating a character in a new pose is making a thousand small decisions, and this line biases all of them toward conservatism at once.

One workflow warning that costs people entire projects. Always generate from the original establishing image, never from the most recent page. Chaining page five off page four off page three compounds drift at every step, and because each step looks fine against the one before it, you do not notice until you lay all twelve pages out and the character on the last page is somebody else.

4 questions

Draw the next page without the character changing — common questions

Why repeat the character description on every page?
There is no memory between requests. "The same character" only means something because an image is attached and the description is restated.
Can I just say "same character as before"?
It carries some context and it drifts. Across a whole book, restate the countable details — the tunic, the palette, the proportions.
Should page 5 reference page 4?
No. Always reference the original establishing image. Chaining compounds drift, and each step looks fine until you see all the pages together.
How consistent does it actually get?
Good enough for a picture book, not good enough for animation. Silhouette, palette and outfit hold well; fine facial detail moves between poses.

Written and maintained by Andy SwiftPublished Last updated

Your first GPT Image 2.5 image is free

Generate my first image