Clothing shoot

Clothing Shoot

Places a styled model into a location environment. Produces a campaign-quality fashion photograph with the model grounded in the scene.

Three reference images drive the output:

1. Environment reference — location, lighting, perspective, depth of field
2. Styled image — the model's outfit and proportions
3. Face reference — identity consistency (headshot)

Output saved to ./brands/[brand-name]/clothing-shoot/[output-name]/shoot_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 styled model

Scan ./brands/[brand-name]/styled/ for subfolders containing a versioned image matching [folder-name]_v1.png (e.g. jade-cloudmerino-tshirt_v1.png). The script auto-detects the latest version.

Present available options:

Available styled models:
  • jade-cloudmerino-tshirt
  • marcus-techsilk-shorts
  • hero-female-mothtech-jacket

Ask: "Which styled model would you like to use for this shoot?"

If no styled outputs exist:

"No styled models found for [brand]. Use /model to create a model first, then /style to dress them."

Once selected, read brands/[brand-name]/styled/[name]/style-spec.json to get model_dir for the face reference.

---

Step 3 — Get the environment reference image

At the start of every clothing-shoot session, create the environment references folder if it doesn't exist:

mkdir -p brands/[brand-name]/clothing-shoot/environment-references

Then list what's in it:

ls brands/[brand-name]/clothing-shoot/environment-references/

If images are present, show them and ask which to use:

Environment references available:
  1. jazz-room.jpg
  2. concrete-alley.jpg
  3. rooftop-golden-hour.jpg

Which would you like to use? (number or name)

If the folder is empty, tell the user:

"Drop your environment reference images into brands/[brand-name]/clothing-shoot/environment-references/ — then let me know and I'll continue."

Wait for confirmation before proceeding. Do not generate without a saved reference image.

---

Step 4 — Get pose direction

Ask: "How should the model be posed or what should they be doing in this scene? Be specific — the more direction you give, the better the pose will suit the environment."

Examples:

• Walking through the space, mid-stride, looking slightly off-camera
• Leaning against the wall, arms crossed, direct gaze
• Caught mid-turn, one hand in pocket, looking back over shoulder

This is required — don't proceed without it.

---

Step 5 — Aspect ratio

Ask: "What aspect ratio do you want? Options: 1:1, 4:5, 3:4, 2:3, 9:16, 16:9, 3:2"

Default to 4:5 if the user isn't sure.

---

Step 6 — Name the output

Derive a slug from the styled model name and a short environment descriptor:

• jade-cloudmerino-urban-alley
• marcus-techsilk-rooftop
• hero-female-mothtech-forest-trail

Ask the user to confirm or rename.

---

Step 7 — Write shoot-spec.json

Create the output folder and write the spec:

brands/[brand-name]/clothing-shoot/[output-name]/
  shoot-spec.json
  env-reference.jpg  (if downloaded)

shoot-spec.json format:

{
  "brand": "[brand-name]",
  "output_name": "jade-cloudmerino-urban-alley",
  "styled_dir": "brands/[brand-name]/styled/jade-cloudmerino-tshirt",
  "model_dir": "brands/[brand-name]/models/jade",
  "env_reference": "brands/[brand-name]/clothing-shoot/jade-cloudmerino-urban-alley/env-reference.jpg",
  "pose_direction": "Walking through the space, mid-stride, looking slightly off-camera",
  "aspect_ratio": "4:5",
  "output_dir": "brands/[brand-name]/clothing-shoot/jade-cloudmerino-urban-alley"
}

model_dir is pulled from the styled output's style-spec.json — no need to ask the user.

---

Step 8 — Generate

Run from the project root:

python3 skills/references/generate-clothing-shoot.py brands/[brand-name]/clothing-shoot/[output-name]

Uploads environment, styled image, and face reference to FAL storage in that order, then calls fal-ai/nano-banana-2/edit at 2K in the specified aspect ratio.

Cost: ~$0.12 per image.

---

Step 9 — Present the result

Once complete, tell the user the exact path where the file was saved, e.g.: brands/[brand-name]/clothing-shoot/[output-name]/shoot_v1.png

Ask: "Happy with this? Or would you like to regenerate, adjust the pose, or try a different environment?"

Once the user is happy, suggest the next step:

"Looking good. If you'd like to bring this image to life, run /add-motion to animate it into a short video clip."

To regenerate (same spec):

python3 skills/references/generate-clothing-shoot.py brands/[brand-name]/clothing-shoot/[output-name]

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

To adjust pose: update pose_direction in shoot-spec.json and re-run.

To try a different environment: replace the env-reference image, update the path in shoot-spec.json if needed, re-run.

To use a different styled model in the same environment: update styled_dir and model_dir in shoot-spec.json, re-run.

---

Notes

• Each brand has its own clothing-shoot/ subfolder — multiple brands can run independent shoot workflows in the same project
• Reference image order matters: environment → styled → face. The script handles this automatically.
• A strong, clearly lit environment reference produces significantly better results than a vague one
• If the model appears floating or unnaturally placed, regenerate — organic placement varies by scene complexity