Pajama Outfit Swap · skill pública de Open Skill

Pajama Outfit Swap

Pajama Outfit Swap es una Open Skill pública en Digen. Ábrela para ver qué crea este flujo.

@wwws

TryOn

0 0 4

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4284863/2026-10-20/767yfb8j.jpg

# Pajama Outfit Swap

Recreates the "pajama outfit swap" experience: take the person from the user's uploaded photo and restyle their clothing into thin, lightweight nightwear, while the face, identity and body stay exactly the same person. Output is one finished image.

Inputs: a reference photo (required) plus optional text describing the pajama style, color, fabric, pose or setting.
Output: one image-to-image result shown as a canvas asset.

## When to use

- The user uploads a portrait/photo and asks to change the outfit into pajamas, nightwear, sleepwear, a nightgown, a loungewear set, and so on.
- The user wants "the same person" but in a cozy bedtime look, and expects the face to stay identical.
- The user refines a previous result (different pajama color, pattern, fabric, room, pose, framing).

Do not use this Skill for full scene/character design from scratch with no reference photo, for video, or for requests that sexualize the subject.

## Steps

1. **Collect inputs.** Take the reference image from the conversation or canvas (asset id / `s3://` uri / https url are all fine). Take any text the user wrote about pajama type, color, pattern, fabric, pose, room or mood. If no reference photo exists, ask the user to upload one — do not invent a person. Only if the user explicitly wants a generic model with no photo at all, switch to text-to-image with `model='t2i'` and say so.

2. **Read the reference photo (optional but recommended).** Call `recognize_image` on the reference with a short English prompt asking for: subject gender presentation and apparent age, hair (length, color, style), current clothing, pose and hand position, framing (close-up / half body / full body), camera angle, background, and lighting. Use the answer only to describe what is really in the photo — never add details the photo does not show.

3. **Load the visual rules.** Call `read_skill_file` with path `references/style-and-prompt.md`. Follow its Style, Subject, Layout, Must keep, Must avoid and prompt-template sections exactly.

4. **Write the prompt.** Compose one English prompt from the template in `references/style-and-prompt.md`, filling every slot with facts from step 2 and the user's text (garment type, fabric, color, pattern, pose, setting, lighting). If the user wrote nothing about the pajamas, use the default look defined in that file. Keep the identity-lock sentence in every prompt.

5. **Run the pre-flight checks.** Call `read_skill_file` with path `references/step-details.md` and apply its checklist and branches (aspect ratio, multi-person photos, missing details, disallowed content, retry rules) before generating.

6. **Generate the image.** Call `generate_image` with:
   - `prompt`: the English prompt from step 4,
   - `input_urls`: `[reference image]` — the uploaded photo first,
   - `model='i2i.hd.lite'` (image-to-image; the default channel for this Skill — never a physical model name),
   - `orientation`: `portrait` by default for standing / half-body / full-body photos, `square` for tight head-and-shoulders crops, `landscape` only when the source photo is clearly landscape,
   - `name`: a short title such as `Pajama Swap – Cotton Set`,
   - `entity_type='characters'`.
   Do not pass `resolution` unless the user asks for a specific sharpness level; the default is fine.

7. **Report and offer next steps.** Describe the result in one or two natural sentences (what pajama style, what setting), mention that the face and identity were kept from the uploaded photo. Do not paste long URLs. If the user wants another variant, go back to step 4 with the changed slot only and generate again. When the task is finished, optionally call `set_guidance` in the same turn with 2–4 concrete follow-ups (e.g. different fabric, different room, full-body framing).

## Notes

- Identity is the product: never change the face, facial features, hair, skin tone, apparent age, gender presentation or body proportions of the uploaded person.
- Only restyle wh