1950s American Diner Portrait · skill pública Open Skill

1950s American Diner Portrait

1950s American Diner Portrait

@GG_Bone

Internal

0 0 0

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4150474/2026-10-15/arxc3ryr.jpg

# 1950s American Diner Editorial Portrait

Generate a 1950s American diner editorial portrait from the user's text (optionally plus one reference image): warm retro film look, vintage Kodachrome color, 3:4 vertical framing. When a reference image is provided, preserve the person's face and identity and rebuild only the wardrobe, hairstyle, pose and setting.

## When to use

- The user wants a 1950s American diner / soda fountain / retro milkshake themed portrait, photoshoot or editorial photo.
- The user uploads or points to a photo of a person and wants "still the same person" but restyled into a 1950s retro look.
- The user wants to change only one element: wardrobe, hairstyle, milkshake flavor, setting details, lighting or color grading.

## Steps

1. **Determine the input and the channel**
   - Reference image available (an image the user uploaded this turn, an image asset on the canvas, or an image the user explicitly points to) → image-to-image: `model='i2i'`, put the reference image in `input_urls` (the first one is authoritative).
   - No reference image → text-to-image: `model='t2i'`, do not pass `input_urls`.
2. **Fill in the description**: use whatever the user specified; fill the rest from the defaults below. Do not keep asking about details.
   - Subject: an adult woman. If the user specifies another adult subject, follow the user.
   - Wardrobe / hairstyle / milkshake flavor / setting: pick one classic combination from each vocabulary list below.
3. **Rewrite into an English prompt**, assembling in this order: subject → wardrobe → hairstyle → pose and props → setting → lighting → film style → aspect ratio. Do not copy the user's original phrasing verbatim.
4. **Call `generate_image`** with these conventions:
   - `prompt`: the English prompt from the previous step
   - `model`: `i2i` (with a reference image) or `t2i` (text only)
   - `input_urls`: pass the reference image only for `i2i`
   - `name`: a short title, e.g. `1950s Diner Portrait`
   - `entity_type='characters'`
   - Framing is always vertical 3:4: `orientation='portrait'` together with `aspect_ratio='3:4'`
   - Do not pass `resolution` (use the default); pass `1080P` only when the user explicitly asks for high resolution.
5. **Reply**: one or two natural-language sentences describing the character setup and the retro style; do not paste long URLs. When the user wants a change, regenerate with only that one element changed and keep everything else the same.

## Prompt template

Use this as the skeleton when there is no reference image:

> Editorial portrait of an adult woman in a 1950s American diner, <wardrobe>, <hair>, <pose and props>, holding a tall frosty <flavor> milkshake with a striped straw, <background elements>, <lighting>, shot on vintage Kodachrome film, warm amber and soft teal palette, subtle magenta cast in the shadows, gentle film grain, soft halation on highlights, shallow depth of field, retro editorial fashion photography, 3:4 vertical composition, no modern objects or contemporary styling

When a reference image is provided, prepend an identity constraint to the same prompt:

> Keep the facial features, face shape and identity of the person in the reference image exactly; change only the wardrobe, hairstyle, pose and setting.

### Vocabulary (pick from these; the user's explicit choice wins)

- **Wardrobe**: red polka-dot swing dress with a white cardigan / poodle skirt with petticoat and satin blouse / sleeveless halter dress with a thin belt / fitted knit cardigan and high-waisted A-line skirt / white diner waitress uniform with apron and headscarf; shoes: `white saddle shoes`
- **Accessories**: `cat-eye glasses`, `silk neck scarf`, `pearl stud earrings`, `ribbon headband`, `short white gloves`
- **Hairstyle**: `victory rolls`, `pin curls`, `high ponytail with a ribbon`, `bouffant`, `finger waves`
- **Pose and milkshake**: `seated at the counter, elbows on the chrome top`, `leaning back in a red vinyl booth`, `sipping through a s