Style

Places product clothing onto a generated model. Produces a full-body studio image with the model wearing the specified product.

Output saved to ./brands/[brand-name]/styled/[output-name]/[output-name]_v1.png.

Step 1 — Select brand

Scan ./brands/ for subfolders that contain a brand-identity/visual-guidelines.md file.

• One brand found: use it automatically and confirm: "Using brand: [name]"
• Multiple brands found: list them and ask which to use
• None found: tell the user to run /brand first

Step 2 — Select a model

Scan ./brands/[brand-name]/models/ for subfolders containing both headshot.png and fullbody.png.

Present available models:

Available models:
  • jade
  • marcus-02
  • hero-female

Ask: "Which model would you like to style?"

If no complete models exist, tell the user:

"No models found for [brand]. Use /model to create one first."

Step 3 — Select product(s)

Ask: "Which product(s) would you like to place on the model?"

The user may specify:

• Brand products — match against ./brands/[brand-name]/brand-identity/product-images/
• Non-brand items — accessories, shoes, or styling the brand doesn't sell (e.g. "brown loafers", "white socks"). These have no product image. Capture them in notes — they will be included in the generation prompt.

Show what was found:

Found in brands/[brand-name]/brand-identity/product-images/:
  ✓ linea-jacket-tobacco.png
  ✓ cowboy-denim-carbon.png
  ✗ dusty brown loafers — no product image, will be added as a styling note

For brand products not found in the folder: Ask the user to either:

• Drop the image into brands/[brand-name]/brand-identity/product-images/ and confirm
• Paste an image URL — download it:

  curl -L "[url]" -o "brands/[brand-name]/brand-identity/product-images/[slug].jpg"

For non-brand items (shoes, socks, belts, etc.): No image needed. Describe them precisely in notes — the script appends notes to the generation prompt. Be specific: colour, material, style (e.g. "dusty brown suede loafers, minimal sole, no branding").

Step 4 — Check for a complete outfit

If the selected product(s) don't form a complete outfit (e.g. only a top, only a jacket), ask:

"The [product] only covers the [top/bottom/half]. What should the model wear for the rest? You can: - Name another product from the brand folder - Describe the styling (e.g. 'neutral grey tailored trousers', 'dusty brown loafers') — no image needed - Leave it to the model to fill in neutrally"

If the user describes styling rather than providing a product image, write it to notes in the spec — it will be passed directly into the generation prompt. If they provide an additional product image, add it to product_images.

Step 5 — Name the output

Derive a slug from the model name and product name(s):

• jade-cloudmerino-tshirt
• marcus-techsilk-shorts
• hero-female-mothtech-jacket

Ask the user to confirm or rename.

Step 6 — Write style-spec.json

Create the output folder and write the spec:

brands/[brand-name]/styled/[output-name]/
  style-spec.json

style-spec.json format:

{
  "brand": "[brand-name]",
  "model_name": "jade",
  "model_dir": "brands/[brand-name]/models/jade",
  "product_names": ["CloudMerino 66 T-Shirt"],
  "product_images": [
    "brands/[brand-name]/brand-identity/product-images/cloudmerino-66-t-shirt.jpg"
  ],
  "output_dir": "brands/[brand-name]/styled/jade-cloudmerino-tshirt",
  "notes": "Dusty brown suede loafers, minimal sole, no branding."
}

notes is appended to the generation prompt as additional styling instructions. Use it for any items that don't have a product reference image — shoes, socks, belts, jewellery, styling details (e.g. jacket zipped vs open), or garment fit notes. Leave empty if not needed.

Step 7 — Generate

Run from the project root:

python3 skills/references/generate-style.py brands/[brand-name]/styled/[output-name]

Uploads face reference, body reference, then product reference(s) to FAL storage and calls fal-ai/nano-banana-2/edit at 2K, 3:4.

Cost: ~$0.12 per image.

Step 8 — Present the result

Once complete, tell the user the exact path where the file was saved using the actual versioned filename, e.g.: brands/[brand-name]/styled/[output-name]/[output-name]_v1.png

Ask: "Happy with this? Or would you like to regenerate or try a different product?"

Once the user is happy, suggest next steps:

"Your styled model is ready. From here you can: - Run /clothing-shoot to place them into a location or campaign environment - Run /ugc to generate a selfie and talking-head video - Run /product-ugc to generate a selfie of them holding a product"

To regenerate (same spec):

python3 skills/references/generate-style.py brands/[brand-name]/styled/[output-name]

Saves as the next version — does not overwrite previous outputs.

To try a different product on the same model: update style-spec.json and re-run.

Notes

• Each brand has its own styled/ subfolder — multiple brands can run independent styling workflows in the same project
• Reference image order matters: face → body → product. The script handles this automatically.
• Multiple product images are supported for multi-piece outfits — add them all to product_images
• If body proportions look wrong or the body is cropped, regenerate — this is prompt adherence variance