An AI influencer prompt that holds a face has three parts: a character line that never changes, a scene line that changes every post and a look line you keep per shoot type. This pack gives you the template, three example character lines and 100 scene lines across ten shoot types, with the changes each image model needs. They are templates to adapt, not fixed recipes.
An AI influencer prompt that keeps one face across a feed has three parts: a character line you never change, a scene line that changes with every post, and a look line you keep per shoot type. Write the character line once, paste it in front of any of the 100 scene lines below, add the look line for that shoot, and you have a full prompt.
Model notes checked October 2026 against Black Forest Labs' FLUX.2 prompting guide, Midjourney's Version and Edit model docs, Google's Gemini image generation docs and Nano Banana 2.1 model page, and ComfyUI's text-to-image tutorial.
Most prompt lists are unrelated one-offs: a new face in every image, the opposite of what a persona account needs. This pack is organised by shoot type, with the part that must stay identical separated from the part that should change. The prompts follow the vendors' current prompt guides; they were not scored in a test, and output varies by model, version and seed, so treat each line as a starting point. For the whole build, start with how to create an AI influencer; this page is the prompt layer of that system.
The AI influencer prompt formula
Every prompt in the pack is the same three lines in the same order. The order matters: Black Forest Labs says FLUX.2 gives more weight to earlier words and recommends putting the main subject first, and a persona is always the subject.
[CHARACTER LINE: same words, same order, every prompt], [SCENE LINE: action, wardrobe, place, one light or framing note]. [LOOK LINE: photo type, light, lens, texture. One per shoot type]
- 1Write the character line once
Age, hair, eyes, two distinguishing marks, build. Five to seven traits, no clothing, no mood.
- 2Pick a shoot type
Each of the ten has its own look line, so a whole shoot shares one photographic style.
- 3Choose a scene line
Action, wardrobe and place for this post. This is the only line you edit freely.
- 4Join them in order
Character, then scene, then look. Add model parameters last.
- 5Generate a small batch
Four to eight images per scene is enough to judge the line.
- 6Check against the reference set
Keep only outputs where eyes, nose, jaw and hairline match. Save the prompt with each keeper.
If you would rather click than type, the free AI influencer prompt generator builds the same structure and formats it for Midjourney V8.x, FLUX.2, Nano Banana or SDXL. If you have nowhere to paste a prompt yet, our list of free AI influencer generators without login shows which tools run without an account and what their free images can be used for.
The character line: the part you keep identical
The character line is a short physical description of a fictional adult, written once and never reworded. Black Forest Labs' guidance for recurring characters is to define the character in detail and maintain that description across every image, repeating the same specifics each time. Three examples:
A 27-year-old woman with shoulder-length wavy dark brown hair, hazel eyes, light freckles across the nose, a small beauty mark under the left eye, slim build A 31-year-old man with short black curly hair, dark brown eyes, a trimmed beard, a thin scar through the right eyebrow, athletic build A 34-year-old woman with a sleek black bob, brown eyes, round tortoiseshell glasses, a gap between the front teeth, medium build
- Include two distinguishing marks. Freckles, a scar, glasses, a tooth gap. Generic faces drift because nothing anchors them; marks give you something to check.
- Leave clothing out. Wardrobe belongs in the scene line. A character line that says "wearing a white dress" fights every other outfit.
- Leave judgement out. Words such as beautiful, perfect or flawless pull every model toward the same averaged face.
- Freeze the wording. Store it in a text file and paste it. Swapping "wavy" for "curly" on a Tuesday is a new character.
The reference methods are covered in depth elsewhere: our consistent character AI guide compares them, and the AI face consistency guide lists the causes of drift when a face slips anyway.
AI influencer prompt examples: 100 scene lines by shoot type
Each shoot type below has one look line and ten scene lines. A full prompt is your character line, one scene line, then the look line. All scenes are written for an adult persona and kept brand-safe; change wardrobe and pronouns to suit yours.
| Shoot type | Prompts | Use it for |
|---|---|---|
| Reference set | 1 to 10 | Run these first. The ten approved outputs become the reference images and the start of a LoRA dataset. |
| At home | 11 to 20 | The everyday filler between bigger shoots: low effort, high relatability. |
| Cafe and city | 21 to 30 | Street scenes test whether the face holds at a distance and in mixed light. |
| Fitness and wellness | 31 to 40 | Movement poses. Expect more rejects here: hands, feet and equipment are where models slip. |
| Travel | 41 to 50 | Location-led posts. Name a kind of place rather than a landmark if you want fewer postcard cliches. |
| Fashion and outfit | 51 to 60 | Outfit posts. The wardrobe is the subject, so list every garment and keep the background plain. |
| Food and cooking | 61 to 70 | Hands plus props. Keep one prop per scene so the model does not invent cutlery. |
| Work and creator desk | 71 to 80 | Behind-the-scenes posts that give the persona a job and a routine. |
| Product in hand | 81 to 90 | Practice for sponsored-style posts. Keep products unbranded; real labels and logos are added from real photos, not invented. |
| Seasonal and events | 91 to 100 | Calendar posts. Batch these a few weeks ahead so the review is not rushed. |
One scene, assembled three ways:
Natural language (FLUX.2, Nano Banana): A 27-year-old woman with shoulder-length wavy dark brown hair, hazel eyes, light freckles across the nose, a small beauty mark under the left eye, slim build, stirring a flat white at a small marble cafe table, camel coat over a black turtleneck. Street photograph, shallow depth of field, overcast daylight, muted colours, 50mm lens. Midjourney V8.2: the same text, then --ar 4:5 --raw SDXL: the same text as comma-separated phrases, plus a short negative prompt in its own field
1 to 10: Reference set
Run these first. The ten approved outputs become the reference images and the start of a LoRA dataset.
Look line: studio portrait photograph, light grey seamless backdrop, soft even light, plain white t-shirt, natural skin texture, 85mm lens 1. front-facing head and shoulders, neutral expression, looking at the camera 2. three-quarter view facing left, relaxed closed-mouth smile 3. three-quarter view facing right, neutral expression 4. full left profile, hair tucked behind the ear 5. full-body standing pose, straight blue jeans and white trainers, arms relaxed 6. head and shoulders, laughing with teeth showing 7. looking back over one shoulder at the camera 8. head tilted slightly down, eyes up to the camera, slight smile 9. waist-up, arms crossed, confident expression 10. seated on a stool, hands in lap, full body in frame
11 to 20: At home
The everyday filler between bigger shoots: low effort, high relatability.
Look line: candid lifestyle photograph, bright Scandinavian apartment, soft window light, 35mm lens, natural skin texture 11. sitting cross-legged on a linen sofa holding a mug, oversized cream sweater 12. slicing a lemon at the kitchen counter, grey t-shirt and apron 13. reading a paperback in an armchair, striped long-sleeve top, evening lamp light 14. watering a monstera on the windowsill, denim shirt with rolled sleeves 15. shaking out a white duvet while making the bed, cotton pyjama set, hazy morning light 16. sitting on the floor with a laptop on a low table, hoodie and joggers 17. hallway mirror selfie holding a phone at chest height, beige trench coat over jeans 18. laughing while folding laundry on the bed, white t-shirt and linen trousers 19. tying hair up at the bathroom mirror, grey sweatshirt 20. curled up on a window seat watching rain, chunky cardigan, cool grey daylight
21 to 30: Cafe and city
Street scenes test whether the face holds at a distance and in mixed light.
Look line: street photograph, shallow depth of field, overcast daylight, muted colours, 50mm lens 21. stirring a flat white at a small marble cafe table, camel coat over a black turtleneck 22. walking toward the camera on a zebra crossing, leather jacket and straight jeans 23. leaning on a bridge railing looking at the river, navy wool coat and scarf, blue hour 24. pointing at a pastry at a bakery counter, denim jacket over a white tee 25. waiting on a metro platform, headphones around the neck, oversized blazer 26. flipping through records in a vinyl shop, plain vintage tee and wide trousers 27. sitting on museum steps with a takeaway coffee, trench coat and loafers 28. holding a paper tray at a night-market food stall, puffer jacket, string lights behind 29. looking out of a tram window with chin on hand, knit beanie and wool coat 30. crossing a rainy street under a clear umbrella, long black coat, reflections on wet pavement
31 to 40: Fitness and wellness
Movement poses. Expect more rejects here: hands, feet and equipment are where models slip.
Look line: sports lifestyle photograph, early morning light, natural skin texture, 50mm lens 31. tying running shoes on a park bench, running jacket and leggings, sunrise 32. holding a plank on a yoga mat, long-sleeve training top, bright studio with a wooden floor 33. drinking from a steel water bottle after a run, breathable t-shirt, riverside path 34. swinging a kettlebell with both hands, grey training tee and joggers, concrete-walled gym 35. stretching a hamstring on a running track, quarter-zip top and leggings 36. sitting cross-legged with eyes closed, soft cotton set, minimalist room with a plant 37. rolling up a yoga mat and smiling at someone off camera, zip hoodie 38. hiking up a mountain trail with a small backpack, fleece jacket, wide shot 39. blending a green smoothie at a kitchen island, training tee, bright window light 40. checking a sports watch with a towel on one shoulder, gym hallway, close-up
41 to 50: Travel
Location-led posts. Name a kind of place rather than a landmark if you want fewer postcard cliches.
Look line: travel photograph, warm natural light, true-to-life colour, 35mm lens, full scene visible 41. walking down a whitewashed alley with a woven bag, linen shirt and wide trousers, looking back 42. leaning on a ferry railing with wind in the hair, striped top, open sea behind 43. sitting on a stone wall above a bay at sunset, flowing midi dress and sandals 44. standing under red shrine gates looking up, light trench coat and trainers 45. pulling a carry-on through an airport terminal, cap, hoodie and joggers, mid-stride 46. eating gelato on a piazza step, sundress and denim jacket, golden hour 47. holding a small coffee on a rooftop terrace, linen blazer, city skyline at sunrise 48. browsing spices at a covered market stall, loose cotton shirt and crossbody bag 49. watching hills go by from a train window seat, knit jumper, close-up 50. walking along the waterline carrying sandals, linen shirt and rolled trousers, sunset backlight
51 to 60: Fashion and outfit
Outfit posts. The wardrobe is the subject, so list every garment and keep the background plain.
Look line: fashion editorial photograph, full outfit visible head to toe, soft directional light, crisp fabric detail, 85mm lens 51. standing square to the camera by a concrete wall, oversized grey blazer, white tee, straight jeans, loafers 52. walking past a stone building with a tote bag, long camel coat and ankle boots 53. bedroom mirror outfit check holding a phone, wide-leg trousers and fitted knit 54. sitting on a wooden stool, black tailored suit and white trainers, white seamless backdrop 55. adjusting sunglasses on a sunny street, striped shirt tucked into linen trousers 56. leaning against a vintage car door, leather jacket, white tee and jeans, golden hour 57. fastening a watch, cuff of a cream knit sweater in frame, close-up on hands 58. standing on gallery stairs looking down at the camera, satin midi skirt and oversized jumper 59. holding up two hangers and choosing between them, plain tee and jeans, open wardrobe behind 60. walking down park steps as the wind catches a trench coat, scarf and boots, autumn
61 to 70: Food and cooking
Hands plus props. Keep one prop per scene so the model does not invent cutlery.
Look line: food and lifestyle photograph, warm interior light, shallow depth of field, hands and face both visible, 50mm lens 61. plating pasta with tongs at a kitchen counter, striped apron over a white tee 62. lifting ramen with chopsticks at a narrow noodle bar, black knit jumper 63. pouring filter coffee from a glass carafe, linen shirt, morning window light 64. picking tomatoes at a farmers' market stall, canvas tote and denim jacket, morning sun 65. laughing across a bistro table with a shared plate in front, silk blouse, candlelight 66. cracking an egg into a mixing bowl, flour on the counter, grey t-shirt and apron 67. eating a croissant on a park bench with a paper bag beside, trench coat 68. carrying a board of cheese and fruit to a garden table, summer dress, string lights at dusk 69. tasting soup from a wooden spoon at the hob, rolled-sleeve shirt, steam rising 70. holding an iced matcha toward the camera, white tee, cafe terrace, drink in sharp focus
71 to 80: Work and creator desk
Behind-the-scenes posts that give the persona a job and a routine.
Look line: editorial workplace photograph, tidy background, soft window light from one side, 35mm lens, natural colour 71. typing on a laptop at a wooden desk, cream knit, glasses pushed into the hair 72. writing in a notebook at a cafe window table, shirt with rolled sleeves, over-the-shoulder angle 73. smiling at a laptop screen on a video call, earbuds in, plain blazer over a tee 74. setting up a camera on a tripod next to a softbox, black tee and jeans 75. sketching on a tablet with a stylus at a drafting table, oversized shirt, profile view 76. pinning photos to a wall moodboard, cardigan and jeans, three-quarter view from behind 77. carrying a laptop and coffee through a co-working space, knit polo and smart trousers 78. unboxing a parcel at the desk with tissue paper around, plain sweatshirt 79. sorting printed photos spread across a studio table, linen shirt, high angle 80. leaning back in a desk chair mid-stretch, hoodie, desk lamp and screen glow in the evening
81 to 90: Product in hand
Practice for sponsored-style posts. Keep products unbranded; real labels and logos are added from real photos, not invented.
Look line: user-generated-style product photograph, unbranded product in sharp focus, soft front light, clean background, 50mm lens 81. holding an unlabelled amber skincare bottle beside the cheek, white tee, cream background 82. pouring water from a plain steel bottle into a glass, training top, bright kitchen 83. showing a plain canvas tote on one shoulder, striped shirt and jeans, pale outdoor wall 84. wearing over-ear headphones with one hand on the ear cup, grey hoodie, minimal desk behind 85. holding a paperback with a blank cover toward the camera, cardigan, reading corner 86. squeezing hand cream from a plain tube, cream knit sleeves, marble counter, close-up on hands 87. lacing plain white trainers on a townhouse doorstep, jeans and tee, low angle 88. holding a ceramic mug toward the camera with both hands, oversized sweater, morning kitchen 89. opening a matte black gift box on a clean table, silk blouse, over-the-shoulder angle 90. checking a plain smartwatch mid-walk, running jacket, city park path
91 to 100: Seasonal and events
Calendar posts. Batch these a few weeks ahead so the review is not rushed.
Look line: seasonal lifestyle photograph, light that fits the moment, shallow depth of field, 50mm lens, candid framing 91. wrapping a scarf on a snowy street, wool coat and knit beanie, shop windows glowing at blue hour 92. holding a sparkler at arm's length and laughing, dark coat, rooftop at night 93. carrying tulips wrapped in paper, light trench coat, spring street with blossom trees 94. sitting on a picnic blanket beside a basket, sundress and straw hat, bright summer lawn 95. kicking up fallen leaves on a park avenue, rust knit and jeans, low golden sun 96. hanging ornaments on a Christmas tree, cream knit jumper, warm fairy lights 97. blowing out candles on a small birthday cake, satin blouse, candlelight on the face 98. carving a pumpkin at a kitchen table, flannel shirt, newspaper under the pumpkin 99. watching fireworks in profile, puffer jacket and scarf, coloured light on the face 100. smiling up from under an umbrella in summer rain, denim jacket, wet city street
AI influencer prompt ideas: how to get past 100
You need more combinations, not more prompts. Each scene line has three slots (action, wardrobe, place); change one and you have a new post from the same shoot. Scene 21 in a rain jacket at a station kiosk is a different image, and the persona has not moved.
- 01Reference set
Prompts 1 to 10, approved once
- 02Pick a shoot
One look line for the whole batch
- 03Vary one slot
Action, wardrobe or place
- 04Generate
Small batch per scene
- 05Review
Against the reference set
- 06Schedule
Save prompt with the image
Weight the shoots toward the persona's niche instead of running all ten evenly: a fitness persona might draw mostly on prompts 31 to 40 and 61 to 70. If the niche is still open, our AI influencer niche guide helps you choose before you generate hundreds of images in the wrong one.
Weak versus strong: the same idea, written twice
- beautiful instagram model, perfect face, cafe, 8k, ultra realistic, masterpiece
- No fixed traits, so the face is rerolled on every run
- Quality keywords in place of a light source and a lens
- Nothing to check the output against
- Character line, then: stirring a flat white at a small marble cafe table, camel coat over a black turtleneck
- Then the look line: street photograph, overcast daylight, 50mm lens
- Same character line as every other post
- Two distinguishing marks you can verify in the result
The weak version describes a category of person. The strong version describes one person in one photograph: who she is, what she is doing, what she is wearing, where, and how the picture was taken. The FLUX.2 guide makes the same point about photographic language, advising camera models, lenses and film stocks over a generic request for a professional photo.
AI model prompt dialects: one pack, four models
The three lines stay the same across tools. What changes is punctuation, where the reference images go and how you exclude things.
| Model | How to write it | How to hold the face | Exclusions |
|---|---|---|---|
| Midjourney V8.2 | Description first, parameters at the end: --ar 4:5 --raw | Edit model: attach up to four reference images | --no item, item |
| FLUX.2 | Subject first; word order carries weight. 30 to 80 words suits most prompts | Multi-reference editing: up to eight images on [pro] at 1MP output | Not supported: describe what you want |
| Nano Banana 2.1 (Gemini) | Full sentences that describe the scene in rich detail | Up to 14 reference images; consistency for up to four characters | No field: describe what you want |
| SDXL checkpoints | Comma-separated description, same three-line order | A character LoRA, or an IP-Adapter reference | Separate negative prompt field |
- Midjourney. V8.2 has been the default since July 24, 2026. Its Edit model replaces Omni Reference and Character Reference: attach images in the Imagine bar on the web, or add
--editand the image URLs in Discord. It matches the first image's aspect ratio unless you add--ar. More in our Midjourney prompts guide. - FLUX.2. No negative prompts: the guide says to write "sharp focus throughout" rather than "no blur". Our Flux consistent character guide covers the reference workflow.
- Nano Banana. Google lists Nano Banana 2.1 as its current primary image model, with improved multi-turn character consistency. Feed approved outputs back in as references when you move to a new scene.
- SDXL. In ComfyUI the positive and negative prompts are two separate text nodes feeding the sampler. Keep the negative list short and specific; our negative prompt guide for realistic people explains what each term does and gives a short template.
Prompts get a persona started; they do not run an account. The AI Influencers program goes deeper where this pack stops, with lessons on prompt engineering, creating the persona, LoRA training for consistent faces and photorealism.
Before you publish an image
- Character line was pasted unchanged, not retyped
- Eyes, nose, jaw and hairline match the reference set at full size
- Both distinguishing marks are present and on the correct side
- Hands, teeth and jewellery look right at 100% zoom
- No invented logos, brand names or readable text in the frame
- The persona is fictional and does not resemble a real, identifiable person
- AI disclosure is in place where the platform or a brand agreement requires it
- Prompt, seed and reference images are saved next to the final file
The last two items are the ones people skip. Disclosure rules differ by platform and country; our AI influencer legal and ethics guide covers labelling, likeness and copyright. And a saved prompt is what lets you reshoot a scene months later when a brand asks for "one more like that".
AI influencer prompts: FAQ
What is a good AI influencer prompt?
A good AI influencer prompt has three parts in a fixed order: a character line that describes the persona with the same five to seven traits every time, a scene line with the action, wardrobe and place for this post, and a look line that sets the photographic style. Only the scene line changes between posts. Write it as a description of a photo, not a list of quality keywords.
How do I keep the same face across different prompts?
Keep the character line word for word identical, then add a reference: attach approved images of the persona in Midjourney's Edit model, Nano Banana or FLUX.2 multi-reference editing, or train a LoRA on an open model. Text alone narrows the face but does not lock it. Check every output against your reference set and reject the ones where the eyes, nose, jaw or hairline moved.
Do these prompts work in Midjourney, Flux and Nano Banana?
Yes, with small changes. Midjourney takes the description followed by parameters such as --ar 4:5 --raw. FLUX.2 wants the subject first and has no negative prompt. Google's docs for Nano Banana advise describing the scene in rich detail, so join the three lines into full sentences. SDXL takes the same description plus a separate negative prompt field. Checked October 2026.
Should I use a negative prompt for an AI influencer?
Only where the model has one. SDXL and other Stable Diffusion checkpoints take a separate negative prompt, and a short list aimed at the failures you actually see works better than a long pasted one. FLUX.2 does not support negative prompts, so describe what you want instead, such as natural skin texture rather than no plastic skin. Midjourney uses --no at the end of the prompt.
How many prompts do I need to run an AI influencer account?
Fewer than you would think, because the scene line is the only part that changes. The ten shoot types in this pack give 100 scene lines; rotating wardrobe, place and light inside each line multiplies that without touching the character. What limits output is not prompt ideas but review time: every image still has to be checked against the reference set before it is posted.
Can I base the character line on a real person?
No. Describe a fictional adult. Do not put a real person's name in the prompt, and do not upload photos of a real person as references without their consent; likeness and publicity rights apply to AI images as they do to photographs, and each image tool sets its own rules on uploading pictures of real people. Build the persona from a written description and your own generated reference set instead.
The prompt is line one. The persona is the system behind it.
AI Influencers, included in All Access, covers persona design, prompt engineering, LoRA training for a consistent face and photorealism, with the other three programs, live coaching and the private community in one subscription.
Build your prompt in a click
Use the free prompt generator to format a character and scene for your image model, then join the free Telegram channel for the prompts and tools we are using right now.