Cozy Romantic Vlog Style · compétence Open Skill

Cozy Romantic Vlog Style

Cozy Romantic Vlog Style est une Open Skill publique sur Digen. Ouvrez-la pour voir ce que ce workflow crée.

@saharasaine

ShortDrama

8 2 1

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4164563/2026-10-05/rrwk3v6y.jpg

# Cozy Romantic Vlog Style

## When to use
- The user wants a cozy, romantic, vlog-style short video: a gorgeous curvy plus-size woman in a warm dimly lit room, with a sweet & flirty voiceover.
- The user wants to redo or tweak one part: change the scene/visual, change the dialogue line, or regenerate the final video.

## Default creative brief
Use these as the defaults unless the user changes them.

**Visual prompt (scene image):**
A cinematic, high-resolution vlog-style video still of a gorgeous, curvy, plus-size Japanese woman sitting in a warm, dimly lit room with soft ambient lighting. She has long dark hair, a sweet and affectionate expression, wearing a cozy oversized knit sweater off one shoulder. Moody cinematic lighting, shallow depth of field, 35mm lens effect, soft skin tones, natural movement.

**Dialogue / voiceover script (sweet & flirty tone):**
"Hey, come a little closer... you're always way over there. Don't you know what you do to me when you look at me like that? Mm, stay right here tonight, okay? I promise it'll be fun."

## Steps
1. First use `write_todos` to list the three stages (scene image → voiceover → final video) and update it as you progress.
2. **Scene image** — call `generate_image`:
   - `model='t2i'` (no reference image; use bare `t2i`, not `.hd`, to keep the soft cinematic look and allow the cozy/romantic mood).
   - `prompt` = the visual prompt above (or the user's edited version).
   - `orientation='portrait'` (9:16 vlog framing).
   - `name='cozy scene'`, `entity_type='storyboard_images'`.
3. **Voiceover** — call `generate_tts`:
   - `model='tts.design'` (voice design, no reference audio needed).
   - `text` = the dialogue script above (or the user's edited line).
   - `voice_instruction` = "playful, teasing, flirty and lighthearted female voice, warm and inviting with a mischievous, giggly tone".
   - `name='flirty voiceover'`, `entity_type='voiceovers'`.
4. **Final video** — call `generate_video`:
   - `model='ai2v'` (image + audio → video, scene driven by the voiceover).
   - `input_urls` = [the scene image asset_id from step 2].
   - `audio_url` = the voiceover asset_id from step 3.
   - `orientation='portrait'`, `resolution='720P'`.
   - `name='cozy romantic vlog'`, `entity_type='final_videos'`.
5. Reply in natural language summarizing the three assets (scene, voiceover, final video). Don't paste long URLs.

## Notes
- If a generation fails or returns an error, report it honestly and retry once with the same parameters; never fabricate a URL or claim success.
- If the user only wants one stage (e.g. just the scene image, or just the voiceover), run only that step and stop.
- The final video is driven by the voiceover audio, so the dialogue line and the scene should match in mood.
- Politely decline requests for explicit/NSFW content — this skill is for cozy, romantic, tasteful vlog-style content only.