Side-Lying Swimsuit Pose · skill pubblica Open Skill
Side-Lying Swimsuit Pose
Side-Lying Swimsuit Pose è una Open Skill pubblica su Digen. Aprila per vedere cosa crea questo workflow.
Custom
1 0 0
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4150474/2026-10-15/g7ecttrp.jpg
# Side-Lying Swimsuit Pose
Turn a user-uploaded photo into a side-lying swimsuit image by re-posing the people in
the photo into one specific reclining pose. There is no structured project state; the
result is a single generated image.
## When to use
- The user uploads an image and asks for the people in it to be re-posed into a
side-lying / reclining pose, especially the "leg kicked straight up" pose.
- The user asks for a swimsuit, beachwear, or pin-up version of an uploaded photo.
- The user says things like "make this pose", "turn them into this pose",
"swimsuit version", "side-lying swimsuit", "repose this photo".
- The user wants to modify only the pose (keeping the person's identity roughly
recognizable) of an already generated image.
## Reference pose to enforce
The target pose is fixed. Read it carefully and encode it in every prompt.
- Body is on the ground, resting on its **side**, weight on the lower hip and
forearm; the whole silhouette is **horizontal along the bottom of the frame**.
- **Head**: propped up on one bent arm, hand near the temple/cheek — a relaxed,
self-assured "propping the head up" gesture. Face turned toward the viewer.
- **Lower leg**: extended flat along the ground, pointing away from the head.
- **Upper leg**: bent at the knee, then kicked **straight up vertically**, so the
calf and foot point at the top of the frame. This vertical leg is the signature
shape and must dominate the upper-left area of the image.
- **Free arm**: resting along the torso/hip, relaxed fingers.
- Frame is **tall / portrait**, tall enough to contain the raised foot; the body
occupies the lower two-thirds, the raised leg crosses the upper-left.
## Steps
1. **Read the uploaded image.** Call `recognize_image(image_url=<the uploaded
asset url>, prompt="Describe the person(s): number of people, apparent age,
gender presentation, hair, skin tone, body type, clothing, and the current
pose.")` so you know whose likeness you are preserving and what the current
pose is. Also note the existing background/lighting, which you may keep or
simplify.
2. **If the person(s) are clearly minors, do not generate.** Politely decline and
stop. Only proceed for clearly adult subjects in adult-appropriate framing.
3. **Translate to the target pose in one clean prompt.** Write the prompt in
English. Use this template and fill in the subject details gathered in step 1:
```
Photograph of a <body type / hair / skin tone / age description> woman,
lying on her side on a <surface, e.g. sandy beach / towel / studio floor>,
wearing a <color and cut, e.g. simple one-piece> swimsuit.
Pose: body horizontal along the ground, resting on her side, weight on the
lower hip and forearm; head propped up on one bent arm, hand near her temple,
face turned toward the camera with a calm confident expression; the lower leg
extended flat along the ground; the upper leg bent at the knee and raised
straight up vertically, calf and foot pointing toward the top of the frame;
the free arm relaxed along her torso.
Style: glossy swimwear editorial, soft warm daylight, natural skin texture,
clean simple background that does not compete with the figure, shallow depth
of field, high detail on the swimsuit material.
```
Rules for the prompt:
- Keep the **same person** as the input: carry over hair color/length, skin
tone, body type, and any accessory that defines identity.
- Always state the pose geometry explicitly (horizontal side-rest + vertical
raised leg). Do not just write "lying down".
- Always state the swimsuit as the only garment; remove the original clothing.
- Do not describe clothing, objects, or a background that contradicts the
single theme; the whole frame serves one look.
4. **Generate with the image-to-image channel.** Call:
`generate_image(prompt=<prompt from step 3>, input_urls=[<uploaded image url>],
model='i2i', orientation='portra