Genvizu

AI Fashion Photography Prompts and Outfit Examples

AI fashion photography prompts on this page come from real Genvizu runs, each shown with the prompt that produced it. The ones that hold up name six things: who the subject is, the garments and their fabrics, the setting, the light, the camera position and crop, and the finish. Copy one, replace the outfit lines with your own pieces, and keep the framing sentence — it is what stops the model from recomposing the shot.

Verified on 2026-09-12

Start free · no credit card

Best for

  • Lookbook and campaign stills where the garment is the subject: name fabric, fit, closure, layering and footwear separately instead of writing "stylish outfit".
  • Editorial portraits with a stated camera position — "low angle below her hips, shooting upward" changes the picture more than any style adjective does.
  • Series that must repeat one model, one outfit or one fabric: carry reference images from run to run, up to 16 in the GPT Image 2 image-to-image channel, 14 on Nano Banana 2, 5 on Seedream 5.0 Pro.
  • Feed and lookbook crops: 3:4, 2:3 and 9:16 are selectable on every image model here, while 4:5 exists only on GPT Image 2 and Nano Banana 2.
  • Briefs written in Chinese or any other language — Midjourney 8.1 is the only image model here that requires English.

Not for

  • Reproducing a real garment exactly. Reference images steer cut, colour and drape; they do not reproduce a print, a label or a stitch line, so a real product still needs a real photo.
  • Images of an identifiable real person. Uploading such a photo requires the rights or written consent, and publicity rights are yours to clear before publication.
  • Locking one face across a long series from text alone. Without references, the same description drifts between runs, because no seed is exposed.
  • Motion. A runway walk or a fabric-in-motion clip needs a video model, not an image model.

Specifications and limits

FieldValue
Typical inputsSubject brief, garment list with fabric and fit, setting, light, camera position and crop, finish
Reference imagesUp to 16 (GPT Image 2 / 2.5 image-to-image), 14 (Nano Banana 2), 5 (Seedream 5.0 Pro), 1 (Midjourney 8.1), none (FLUX.2 Klein)
Feed ratios3:4, 2:3 and 9:16 on every image model here; 4:5 only on GPT Image 2 and Nano Banana 2
Prompt budget20,000 characters on GPT Image 2 and GPT Image 2.5, 3,000 on Seedream 5.0 Pro, 1,024 on Midjourney 8.1
Delivery size4K on GPT Image 2, GPT Image 2.5 and Nano Banana 2; Seedream 5.0 Pro, Midjourney 8.1 and FLUX.2 Klein stop at 2K
Prompt languageAny language, except Midjourney 8.1 which requires English
Output per runOne image on GPT Image 2, GPT Image 2.5, Nano Banana 2 and Seedream 5.0 Pro; four variants on Midjourney 8.1
RightsPhotos of identifiable people need rights or written consent before upload; commercial clearance stays with you
Failed runsRefunded automatically

Prompt examples

High-end fashion portrait with the crop fixed first

Prompt

Masterpiece, ultra-realistic high-end fashion portrait photography, an elegant adult East Asian woman, approximately 25 years old, seated gracefully in a dark traditional East Asian interior. Vertical three-quarter composition from head to mid-thigh. Her body is turned slightly sideways while her face tilts downward, eyes softly lowered, creating a poised, reserved and subtly alluring expression.

The crop is written as a range — "from head to mid-thigh" — before any styling detail, which is what keeps the frame from sliding out to a full-body scene. Stating that the subject is an adult with an approximate age is also the habit to keep: age left implicit is the most common reason a fashion prompt is refused by moderation.

Outfit written as a swappable garment list

Prompt

A cute minimalist portrait of a young woman sitting casually on the floor in a clean studio setting with a light grey background. She wears an oversized dark charcoal hoodie, cream straight-leg pants, white socks, and chunky white sneakers. Relaxed seated pose with crossed legs, one eye slightly closed in a playful wink and a soft smile. Behind her, a large black-and-white doodle character hugs her from behind, wrapping its cartoon arms around her shoulders and gently squeezing her cheek. The doodle figure has a simple round head, tiny eyes, and a hand-drawn sketch style.

Every garment carries a fit and a colour — "oversized dark charcoal hoodie", "cream straight-leg pants" — so one piece can be swapped without rewriting the prompt. That is what makes a prompt reusable across a whole drop: the pose, studio and background sentences stay, only the garment line moves.

Lifestyle framing driven by camera position

Prompt

A fresh high-key lifestyle portrait set against a solid pure-white backdrop fading gently to pale off-white gray. The central subject is a young woman in her twenties standing before this bright gradient white background. The camera is positioned close-up at a low angle below her hips, shooting upward. She sits perfectly centered in a vertical 3:4 or 4:5 frame, framed from the top of her head down to below her hips with her calves cropped out of the shot.

Camera height, distance and crop are written as instructions rather than as a mood, which is why the result reads as one specific photograph. One caveat: the ratio still comes from the panel, and 4:5 is available on GPT Image 2 and Nano Banana 2 but not on GPT Image 2.5, Seedream 5.0 Pro, Midjourney 8.1 or FLUX.2 Klein — 3:4 is the one every image model here offers.

How it compares

Nano Banana 2

Pick it when the same model or the same garment has to repeat across a set: it takes up to 14 reference images per run, offers ten ratios including 4:5 and 5:4, and delivers at 1K, 2K or 4K.

Portrait

Go there when the face is the deliverable rather than the clothes — that page is organised around identity and skin rendering as the acceptance criteria.

Poster

Go there when the fashion image also has to carry a headline and typographic slots: those prompts are built around layout, empty space and quoted strings.

Examples from the gallery

Common failures and fixes

The fabric turns generic, or the garment is redesigned instead of reproduced.

Attach real photos of the piece and list what must not change — cut, colour, closure, print placement — in one sentence, with the scene described in a separate sentence. When the text and the references disagree, the text wins.

The face drifts between images in the same series.

Feed two or three references of the same person from different angles rather than one, and repeat the fixed traits in words. There is no seed to lock, so consistency comes from references plus repeated wording.

Hands, fingers or jewellery come out mangled.

Write the hand position explicitly — in a pocket, resting on a knee, fingers relaxed — instead of leaving it to the model, and avoid poses where fingers overlap the garment. Then re-run: with no seed, each run is a fresh draw and a different one usually resolves it.

4:5 is not selectable on the model you picked.

It exists on GPT Image 2 and Nano Banana 2 only. Elsewhere, generate at 3:4 — available on every image model here — and crop to 4:5 afterwards, since changing the ratio recomposes the shot rather than trimming it.

A long styling brief is rejected or truncated.

Check the model budget: Seedream 5.0 Pro caps prompts at 3,000 characters and Midjourney 8.1 at 1,024, while GPT Image 2 and GPT Image 2.5 allow 20,000. Trim the adjectives before you trim the garment list.

Frequently asked questions

How do I write an AI fashion photography prompt?

Cover six things in order: the subject, the garments with fabric and fit, the setting, the light, the camera position and crop, and the finish. Write each garment as a concrete noun with a colour and a fit, and state the crop as a range such as "from head to mid-thigh" so the framing survives the run.

How do I keep the same model and outfit across a series of images?

Carry reference images from run to run and repeat the fixed traits in words. The image-to-image channel on GPT Image 2 and GPT Image 2.5 accepts up to 16 references, Nano Banana 2 up to 14 and Seedream 5.0 Pro up to 5. None of these models expose a seed, so references plus repeated wording are what hold a series together.

Which aspect ratio should I use for fashion images?

Use 3:4 or 2:3 for lookbook and feed placements and 9:16 for stories; all of them are selectable on every image model here. The 4:5 crop is available on GPT Image 2 and Nano Banana 2 only, so on other models generate at 3:4 and crop afterwards.

Can I use AI fashion images for a real clothing brand?

Only after your own rights clearance. A generated garment resembles your product, it does not reproduce it, so product pages still need real photography. Generated output also carries no guarantee that it is free of third-party intellectual-property, trademark or publicity rights, and Midjourney output is additionally restricted for commercial use by Midjourney’s terms.

Do AI outfit prompts have to be written in English?

No. Midjourney 8.1 is the only image model here that requires English; GPT Image 2, GPT Image 2.5, Nano Banana 2, Seedream 5.0 Pro and FLUX.2 Klein accept Chinese and other languages. Garment names are worth writing in English anyway, since cut and fabric vocabulary is more precise there.

Sources