AI Generation

Character Reference: Keeping a Subject Consistent Across Shots

Also called subject reference, character consistency, cref, sref, reference image

A character reference is one or more images supplied alongside a prompt so a model reproduces that specific person, creature or object instead of inventing a new one. It conditions identity rather than composition, which is what makes character consistency possible across separate generations.

What it conditions, and what it does not

Most of the difficulty in producing more than one shot of the same thing comes from a single fact: each generation starts fresh. Describe a character in words, run the prompt twice, and you get two different people who both match the description. A character reference exists to break that, by supplying pixels instead of adjectives.

What it constrains is identity: face structure, hair, distinguishing marks, often clothing and colour. What it deliberately does not constrain is the shot. Pose, framing, camera angle, lighting and environment stay under the prompt's control, which is the whole point. If the reference dictated composition too, you could not put the same person in a new scene.

This division is also where most failures come from. Hand a model a stylised painting as a character reference and it will carry the brushwork across, because it cannot tell which properties of the image you meant. Hand it a photograph shot in hard side light and the shadow becomes part of the identity.

Choosing reference images

The rules are unglamorous and they matter more than prompt wording.

  • Neutral, even light. Flat frontal or soft window light. Shadow gets baked in as a feature.
  • Multiple angles. Front, three-quarter, profile. Three to five images beats one, because the model has genuine information about the sides of a head rather than a guess.
  • Consistent subject across the set. Same haircut, same age, same wardrobe if wardrobe matters. A reference set that disagrees with itself averages out into a fourth person.
  • Plain background. Other faces in frame are the single most reliable way to get a hybrid.
  • Resolution and sharpness. Reference detail is the ceiling on output detail.

For a recurring subject, build the reference set once, keep it in the project, and reuse it for every shot. Consistency across a sequence comes from the reference set being fixed, not from any individual generation being good.

Prompting alongside a reference

The prompt's job changes when a reference is attached. Stop describing the subject and start describing everything else.

Write the scene, the framing and the light in full, then name the reference explicitly rather than redescribing its contents: the woman from the reference image, standing at a bus stop in the rain at night. Adding dark hair, green eyes, mid thirties on top of a reference does not reinforce the identity. It introduces a second, weaker description that competes with the pixels and pulls the face toward the average of both.

Then pin the attributes that must survive: keep her face, hair length and jacket exactly as in the reference. Models treat clothing as far more negotiable than faces, so wardrobe continuity usually needs to be named.

Where you can, keep the subject at roughly the scale it occupies in the reference. Identity holds well from medium shot to medium close-up and degrades quickly in wide shots, where a face is a few dozen pixels and there is nothing to condition.

When to move to a LoRA instead

A character reference is the right tool for a handful of shots, a pitch, or a one-off. It costs nothing, needs no training, and works immediately.

It has a ceiling. Identity is suggested rather than learned, so it loosens under unusual poses, profile turns the reference did not cover, and long video clips. If the same character appears across dozens of shots, training a LoRA on twenty to thirty images holds the likeness through angles and expressions no reference set can cover, and removes the per-shot drift entirely. The practical trigger for switching is repetition: once you are regenerating shots because the face moved, the training run is cheaper than the re-rolls.

Models that support this

Pulled from the live ZOOOP model catalog, so this list stays current as new models ship.

The prompt for this

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

The woman from the reference image, standing at a bus stop in the rain at night, wet coat, neon reflections on the pavement, medium shot at eye level, keep her face, hair length and jacket exactly as in the reference

Try Character Reference yourself

Open the generator with a starting point already filled in.

Frequently asked questions

What is the difference between a character reference and a style reference?
A character reference says who or what is in the shot. A style reference says how the shot should look: palette, rendering, texture, era. They are separate conditioning channels on most models, and giving a stylised illustration as a character reference is the classic mistake, because the model copies the rendering along with the identity.
How many reference images should I give?
One clean frontal image is enough for a plausible likeness. Three to five, covering front, profile and a three-quarter angle in consistent lighting, is where character consistency becomes reliable, because the model has real information about the parts of the head your prompt will rotate into view.
Why does the face drift over a sequence of shots?
Because each generation is independent. The reference constrains identity but does not pin it, so every shot lands slightly differently, and small offsets accumulate as you look across a sequence. Regenerating drifted shots against the same reference set is more effective than trying to describe the face more precisely in words.
Is a character reference the same as training a LoRA?
No. A reference conditions a single generation at inference time and costs nothing extra. A LoRA changes the model's weights by training on a set of images, holds identity far more tightly, and survives poses and angles the reference never showed. Use references for a handful of shots and a LoRA for a recurring character.
Does a character reference work in video as well as images?
Yes, and it is where most of the value is. Video models that accept reference images can carry a subject through a shot they were never photographed in, which is otherwise the hardest part of building continuity. Identity holds best when the subject stays roughly the scale it was in the reference.
What makes a bad reference image?
Heavy shadow across the face, extreme wide-angle distortion, sunglasses or a hat, motion blur, a busy background with other faces in it, and low resolution. Anything the model has to guess at becomes something it invents freely on the way out.

Related terms