AI Character Consistency Prompts & Reference Templates

Character reference sheet, single subject repeated across a clean grid, 6 panels on a neutral light grey seamless background (#e6e6e6).

SUBJECT: 28-year-old [gender] with [hair colour and cut], [eye colour] eyes, [skin tone] skin, [one distinctive permanent feature: freckles across nose / small scar above left eyebrow / silver septum ring], wearing [exact outfit: charcoal wool overcoat, black crew neck, dark denim].

PANELS: front full body, three-quarter left, profile left, back, waist-up neutral expression, waist-up subtle smile.

Consistent flat studio lighting from a large softbox at 45 degrees camera left, no coloured gels, no shadows on the background. Shot on 85mm lens, f/5.6, sharp focus edge to edge, photorealistic skin texture with visible pores, natural colour grading.

Identical face geometry, identical hairstyle, identical wardrobe in every panel. No text, no watermarks, no logos, no additional people, no props.

Aspect ratio: 3:2.

Using the attached reference image as the exact identity, generate a new photorealistic portrait of the same person. Do not change facial structure, eye shape, nose, jawline, hairline, or skin tone.

NEW SCENE: [describe: standing on a rain-soaked city street at night, neon signs reflecting in puddles behind them].
WARDROBE: [describe exactly].
LIGHTING: soft key light from camera left, cool rim light from behind right, practical neon sources visible in the background bokeh.
CAMERA: 50mm lens, f/1.8, shallow depth of field, subject sharp, background compressed and blurred.
FINISH: natural skin texture with pores and fine hairs visible, no beauty smoothing, no plastic sheen, subtle film grain, colour graded with cool shadows and warm highlights.

Keep the identity 100% consistent with the reference. Photorealistic, editorial photography quality, 4K detail.

Photorealistic lifestyle photograph of the same person as the attached reference. Identity must remain identical — same face, same hair, same body proportions.

SCENE: [e.g. sitting at a corner table in a sunlit Lisbon café, iced coffee on the table, tiled wall behind].
WARDROBE: [exact outfit].
POSE: candid, mid-gesture, looking slightly off camera, natural body language, not posed for the lens.
CAMERA: 35mm lens, f/2.2, eye level, subject placed on the right third.
LIGHT: warm natural window light from camera left, soft falloff, gentle highlight on the cheekbone.
FINISH: real skin texture with pores and fine hair, slight lens vignette, natural colour science like a Fujifilm sensor, no beauty filter, no HDR look.

Instagram editorial quality, believable as a real photo. No text, no logos, no extra hands or fingers.

Aspect ratio: 4:5.

Write a reusable identity lock paragraph for a recurring faceless-channel character.

INPUTS:
Niche: [PASTE NICHE]
Gender / age range: [PASTE]
Vibe: [PASTE, e.g. calm, authoritative, weathered]

Produce a single dense paragraph of 90-120 words that specifies, in this order: age, ethnicity and skin tone, face shape, jaw and cheekbone structure, eye colour and shape, eyebrow shape, nose shape, lip shape, exact hairstyle including length, part and texture, facial hair, three permanent distinguishing marks, build and height, posture, and the fixed core wardrobe.

Rules:
- Every attribute must be concrete and visual. No adjectives like "beautiful" or "striking".
- No emotion, no pose, no lighting, no background — those change per shot.
- Written so it can be pasted unchanged into any image or video prompt.

Then output a 12-word short version for use when the prompt budget is tight.

Character turnaround sheet, 5 full-body views of the same character in a single horizontal row on a flat neutral grey background.

VIEWS, left to right: front, three-quarter front left, profile left, three-quarter back left, straight back.

CHARACTER: [PASTE YOUR IDENTITY LOCK PARAGRAPH HERE]

RULES: identical height and proportions across all five views, aligned to the same horizon and ground line, feet on the same baseline, T-pose relaxed with arms slightly away from the body, neutral expression, identical wardrobe and accessories in every view.

LIGHTING: flat even studio lighting, no dramatic shadows, no coloured light.
RENDER: photorealistic, 85mm lens compression, sharp throughout.

No text, no labels, no measurement lines, no extra characters, no background objects.

Aspect ratio: 16:9.

[IDENTITY LOCK — paste the exact same paragraph every single time, word for word]

SHOT: [wide / medium / close-up / extreme close-up]
SCENE: [location, time of day, season, weather]
WARDROBE VARIATION: [only change layers and colours, never the core silhouette]
POSE / ACTION: [one clear action]
EXPRESSION: [one emotion, described physically — brow lowered, jaw set]
CAMERA: [focal length], [aperture], [angle: eye level / low / high], subject on the [left / centre / right] third
LIGHTING: key [direction and quality], fill [strength], rim [colour and direction], practicals [list]
GRADE: [palette and contrast]

Consistency rules: identical facial geometry, hairline, eye colour and permanent marks as described in the identity lock. No aging, no weight change, no hairstyle change.

Photorealistic, no text, no watermark, no extra people.

Define a reusable narrator voice profile for a faceless channel.

NICHE: [PASTE NICHE]
AUDIENCE: [PASTE]

Deliver:
1. Voice description in 40 words: age impression, timbre, accent, pace in words per minute, energy level, breathiness, and the emotional default.
2. Delivery rules: sentence length preference, where the narrator pauses, which syllables get weight, what the narrator never does.
3. Five sample lines written the way this narrator would say them, with pause markers.
4. The TTS settings to approximate it: recommended stability, similarity and style values, plus what to adjust if the read sounds flat.
5. The three phrases that become this narrator's signature across every video.

Build a cast bible for a faceless narrative channel with recurring characters.

PREMISE: [PASTE]
CAST SIZE: [number]

For each character deliver: name, role in the series, the identity lock paragraph (90-120 words of purely visual, concrete attributes), the fixed wardrobe core, one prop that always appears with them, their speech pattern in one sentence, and their relationship to every other character in one line each.

Then deliver:
- A visual contrast check confirming no two characters share silhouette, hair colour, or dominant wardrobe colour.
- The three recurring locations described so they can be regenerated consistently.
- A continuity checklist to run before generating any new episode.