The comic or graphic novel
The hero stays recognisable from the first panel to the last, as a realistic photo or in a drawn style described identically in every scene prompt.
Keep the same character across AI images
A selfie becomes a four-view reference sheet, then the same character, face and outfit included, in as many scenes as you want. The workflow is shown here, no account needed.
Real workflow: a selfie of Frank, his four-view sheet, then Frank at a cafe terrace in Lisbon and on a ridge at sunset. Same face, same overshirt, same boots in both scenes.
At a glance
- You provide
- 1 photo · a text or a brief
- What the workflow produces
- 3 images in 16:9, 4:3
- Canvas structure
- 8 nodes wired by 6 connections
- Models wired in
- Nano Banana Pro
- Cost of one full run
- about 57 credits, roughly $0.68
- Account
- The demo is visible without an account; generating requires a free account, trial credits included
A comic, an illustrated book, a campaign or a series of posts needs the same character from one image to the next. That is what AI does worst when you give it the selfie again each time: the face drifts, the outfit changes. This workflow first builds a reference sheet, then every scene starts from that sheet.
The sheet fixes the face from the front and in profile, the silhouette and a precise outfit. Scenes are wired to it, never to the photo, and each prompt repeats the outfit word for word. The face itself is not described: the model takes it from the image instead of rebuilding it.
The full guide is in our article on consistent characters. For a whole story, see the comic or the illustrated children's book. The sheet and two scenes cost fifty-seven credits.
Actual renders from the demo
These are the files of the workflow above, exactly as the models produced them, unretouched. The starting point is shown when there is one.
Starting point
Render
Who it is for
Comic and picture book authors, brands, content creators and storyboard artists who need the same character from one image to the next.
Which files to provide
A sharp front selfie, and the character's outfit described once and for all.
How it works
- 1
Start from a sharp front selfie
The whole face, well lit, no sunglasses. It is the only time the photo is used: everything else starts from the sheet.
- 2
Make the sheet with a fixed outfit
Face close-up, then front, three quarter and profile, on a plain background. Describe a precise outfit: it is what links the scenes.
- 3
Wire every scene to the sheet
Not to the selfie. Each Image node receives the sheet, and its prompt repeats the outfit word for word.
- 4
Do not describe the face
Hair and beard, yes; the shape of the eyes or nose, no. A described face is a rebuilt face, so another one.
Variations and use cases
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The children's book
The same little character goes through every page of the children's book, with the same outfit, colours and cuddly toy, the face unchanged from one illustration to the next.
The brand campaign
A human brand face or mascot appears in ten situations of the campaign, at the office, travelling, in a shop, without ten photo shoots or ten castings.
The content creator
The content creator places themselves in travel, sport or work settings for a series of posts, always recognisable by their audience.
The storyboard
The lead actor appears in every shot of a storyboard sequence, with the same face and costume, before the shoot or before moving to video.
The role-playing game or avatar
A role-playing character or an avatar is shown in several scenes for its profile, the presentation of its world and the campaign artwork.
Common mistakes
Giving the selfie to every scene
One angle, reinterpreted each time: the face drifts. The sheet gives every view at once.
Leaving the outfit open
Without an outfit repeated word for word, each scene dresses the character differently, and you no longer recognise them from afar.
Describing the face in detail
"Thin nose, green almond eyes" pushes the model to draw that face, not the one on the sheet.
Using someone's face without consent
For a real person other than you, ask for their permission before publishing images of them.
Recommended models
Nano Banana Pro
Seedream 5 Pro
Flux Kontext
GPT Image 2.5 SunburstFrequently asked questions
How many scenes can I make?
As many as you want: duplicate a Text node and an Image node wired to the same sheet. Each scene costs nineteen credits.
Can the outfit change?
Yes, but make a second sheet for the new outfit. Changing the outfit in a single scene is what makes the character drift.
Does it work for an invented character?
Yes: generate a portrait first, then the sheet from that portrait. The rest is the same.
How much does the demo cost?
Fifty-seven credits: nineteen for the sheet, nineteen per scene, on Nano Banana Pro.
Why Nano Banana Pro?
It keeps a face from a reference image better than the other image models in the catalogue, and it holds the described outfit. Measured on the demo.
Can I animate the character afterwards?
Yes: wire a scene as the first frame of a Video node. Wan 3.0 also accepts the sheet as a reference image.
What about a drawn style?
Yes, by describing it in the sheet AND in every scene: watercolour, manga, comic line art. The style must be the same everywhere.
Do I need a front selfie?
It is the safest. A profile or a photo in shadow lets the model invent half of the face.
Keep the same character across AI images
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.