Futuristic Asian Portrait · публичный навык Open Skill

Futuristic Asian Portrait

Futuristic Asian Portrait — публичный Open Skill на Digen. Откройте, чтобы увидеть, что создаёт этот процесс.

@Vjagstya

Anime

6 1 3

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4136188/2026-09-22/vhz6vtff.jpg

# Futuristic Asian Portrait

Generates a cinematic closeup portrait of a futuristic Asian character with a signature mood: standing before a waterfall, wind streaming through the hair, water glistening on the face (drops running from the eyebrow down the cheek). Supports on-request variations, and can animate the still into a short clip.

## Reference files
- `references/portrait-style.md`: The master cinematic style template + the default prompt. Read it before the first generation so the base look stays consistent.

## When to use
- The user asks for this futuristic Asian closeup portrait (waterfall, wind, water drops on the cheek).
- The user asks for a variation: different expression, hair style/color, clothing, camera angle, lighting, background swap, or a redo.
- The user asks to animate an existing still of this character into a short motion clip.

## Steps

1. **Check intent.** Ask briefly whether they want a fresh variant (tell me what to change) or the default signature look; if they just name the idea, use the default.

2. **Generate the image** (`generate_image`).
   - Read `references/portrait-style.md` and build the prompt from the master template, adjusting only what the user asked to change (expression, hair, clothing, angle, lighting, background).
   - Default to the signature look: futuristic Asian character, front of a waterfall, wind through hair, water drops from eyebrow to cheek, closeup.
   - Parameters: `model='t2i'` (text-to-image, no reference), `orientation='portrait'` (9:16 closeup), `resolution` default `720P` (raise to `1080P` if the user wants higher detail). `name='Futuristic Asian Portrait'`, `entity_type='characters'`.
   - Show the result and ask if they want any tweak.

3. **Video clip (optional step — only if requested).** Animate a chosen still with `generate_video`.
   - Pass the generated image asset as `input_urls` (the still becomes the start frame), describe the motion in the `prompt`: wind blowing the hair, waterfall flowing, water drops sliding down the face.
   - Parameters: `model='i2v'` (image-to-video), `orientation='portrait'`, `resolution` `720P`, short `duration` (e.g. 4–5s). `name='Waterfall Portrait Clip'`, `entity_type='final_videos'`.
   - Report the clip result in natural language.

4. **Reply** in natural language; describe what was generated, and offer the next variation or the animation step. On failure, honestly report what went wrong and offer to retry — never fabricate a URL.

## Notes
- Keep the character consistent across variations and the animation by reusing the same generated still/facial description as reference where possible.
- Non-default changes (e.g. different ethnicity request, expanded scene, bigger resolution) are welcome as long as the request is appropriate; politely decline any prohibited/adult content.
- If the user wants a different established style (cartoon, cyberpunk poster, etc.), adjust the style wording in the template accordingly.