Fashion AI School frameworks
Three frameworks for AI creative work
Named methods we teach and use in our studios. Each one fixes a common reason AI images look fake or inconsistent.
The 4-Point Commercial Prompt Formula
Subject + Surface/Rig + Lighting Technique + Lens/Optics.
The 4-Point Commercial Prompt Formula is a prompt structure by Fashion AI School for realistic commercial product and fashion images. Every prompt names four things in order: the subject, the surface or set it sits on, the lighting technique, and the lens. Leaving out any of the four is the most common reason AI product images look fake.
1
Subject
The exact product or person, with material and colour. "Amber glass serum bottle with a gold dropper."
2
Surface / rig
What it stands on and what surrounds it. "Wet black slate, scattered rose petals."
3
Lighting technique
A named light setup, not "good lighting". "Single softbox from the left, black card fill, soft rim."
4
Lens / optics
Focal length and depth. "100mm macro, f/4, shallow depth of field."
Amber glass serum bottle with a gold dropper, on wet black slate with scattered rose petals, single softbox from the left with a soft rim light, 100mm macro lens at f/4, commercial product photography, true colour, sharp label.
The Ghost-Mannequin to Campaign Pipeline
Garment photo → studio stage → model try-on → 9:16 motion.
The Ghost-Mannequin to Campaign Pipeline is a four-stage workflow by Fashion AI School that turns one flat lay, hanger or ghost-mannequin photo into a full campaign: a clean studio packshot, on-model images, and a vertical video ad. The garment is locked from the first photo, so each stage only changes the scene, never the product.
1
Garment photo
One sharp photo of the real garment on a plain background: flat lay, hanger or ghost mannequin.
2
Studio stage
Place the garment in a clean, consistent set, or change the background.
3
Model try-on
Put the same garment on an AI model in the chosen setting, keeping colour, print and cut.
4
9:16 motion
Animate the best on-model frame into a short vertical clip for Reels and TikTok.
Use the garment in [garment photo] exactly: same colour, print, buttons and cut. Show it worn by [model description] in [setting], [lighting], full-length, natural drape and folds, realistic fabric texture, e-commerce quality.
The 3-Angle Identity Anchor
Front, 45° and profile references for a consistent face.
The 3-Angle Identity Anchor is a consistency method by Fashion AI School: before making any campaign image, create three reference images of the same face (front, 45-degree three-quarter, and side profile) in neutral light. Every later image is generated from those three references, which keeps the person recognisable across shots, outfits and settings.
1
Front
Straight to camera, neutral expression, even soft light, plain background.
2
45° three-quarter
Same person turned halfway, same light and background.
3
Profile
Full side view showing nose, jaw and ear shape.
4
Generate from all three
Attach the three references to every new prompt and change only outfit, pose and scene.
Use the person in [front], [45°] and [profile] references as the same individual: identical face shape, eyes, nose, skin tone and hairline. Show them wearing [outfit] in [setting], [lighting], [lens].
Which AI model for which job
A plain-language comparison from public documentation and our own use. Models change often, so test on your own products.
| Model | Best for | Product accuracy from a photo | Video | Access |
|---|---|---|---|---|
| Midjourney | Mood, concepts, editorial style | Weaker: tends to redraw the product | Short animations | Subscription |
| Flux | Photorealism, following long prompts | Good with reference tools | No (image model) | API or apps |
| Gemini image (Nano Banana) | Editing a real photo, keeping the product | Strong: edits around the original | No (image model) | Used in our studios, pay per image |
| Higgsfield | Fashion video, camera moves, AI ambassadors | Good when started from a photo | Yes, main strength | Used in our studios, pay per clip |
Photoshoot vs AI workflow: where the cost goes
Shoot prices vary a lot by city and team. Use the cost calculator with your own numbers.
| Cost line | Traditional photoshoot | AI workflow in our studios |
|---|---|---|
| Set up | Studio hire, photographer, model, stylist, travel | One garment or product photo |
| Per image | Shared across a full shoot day | €0.20 per image (testing price), refunded if it fails |
| Changes after the shoot | Usually a reshoot | Generate a new version |
| Video | Separate crew or editor | Pay per clip from the same balance |
| Best use | Master packshots, exact fit, hero campaigns | Variations, lifestyle scenes, social and ad formats |
Guides: commercial photography · creative direction · video production · character consistency · e-commerce imagery
Your next step