Elevator Welcome Clip — Open Skill

Elevator Welcome Clip

Elevator Welcome Clip is a public Open Skill on Digen. Open it to see what this workflow creates.

@toopchat

Beauty

1 1 1

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4239354/2026-10-20/hhr8cu6x.jpg

# Elevator Welcome Clip

## When to use
- The user asks for the elevator-hallway clip: closed modern elevator doors, a woman in modest formal attire with a headscarf taking a selfie inside, noticing the camera, smiling, lowering the phone, stepping back as the camera enters, and inviting the camera inside with eye contact only.
- The user wants to redo or restyle this same shot (different mood, outfit detail, framing, aspect ratio).
- The user supplies a reference photo of the woman — use it as the identity reference and go through the reference-image route.

## Steps

1. **Pick the route**
   - No reference image → **Route A (text-to-video)**: go straight to step 2 with `model='t2v'`.
   - Reference image supplied (uploaded asset or preset asset key) → **Route B (reference-to-video)**: `model='ref2v'` with the image(s) in `input_urls` (all inputs are references, never start/end frames).

2. **Build the prompt** — one dense English paragraph covering, in order:
   - Style: photorealistic, cinematic shot filmed by an **independent third-party camera** held by another person in the hallway — the filming camera is **not** the woman's own phone; no mirror, no reflection, no selfie-stick, no phone-screen POV. 24fps, single continuous shot, no cuts, no edits.
   - Setting: modern hallway, camera standing still facing closed modern elevator doors.
   - Action 1: the doors slide open.
   - Character: beautiful young Iranian woman, long modest formal office dress with a headscarf, fully covered; inside the elevator, her smartphone held low at waist level, briefly raised for a selfie.
   - Action 2: she notices the camera at the entrance, looks up, surprised for a moment.
   - Action 3: she lowers the phone slightly in a small shy, modest gesture and a soft warm smile spreads across her face.
   - Action 4: the camera slowly moves forward and enters the elevator; as the camera approaches her, she takes one small step **backward, away from the camera**, to make room, keeping a warm gentle smile and looking straight into the camera lens.
   - Action 5 (the invitation): the invitation to come inside is carried **only by her eye contact and facial expression** — a warm, welcoming, slightly bashful gaze straight into the lens, as if silently saying "please come in"; her eyes stay locked on the camera the whole time the camera moves in. No speech, no words, no audible or visible talking, no lip movement, no subtitles, no on-screen text, and no gesture other than the single small lowering of the phone.
   - Look: realistic elevator interior lighting, natural skin texture, highly detailed, gentle friendly mood.
   - **Negative constraints (always append as a final exclusion sentence):** the woman's face and identity stay perfectly consistent and never morph or shift; her clothing and headscarf stay exactly the same throughout — no wardrobe change, no revealing or loose clothing, and the outfit remains fully modest and covered; she does not spin, twirl, or dance, and takes only one small backward step, never moving toward the camera; no extra people appear and no one else enters the hallway or elevator; photorealistic live-action look — not cartoon, not 3D animation, not illustration; sharp focus throughout, never blurry or out of focus; hands and fingers stay anatomically correct and steady, never distorted, warped, or extra; she never talks, never mouths or pronounces words, no speech, no lip movement, no laughter with sound, no subtitles or text on screen; her gaze stays on the camera the entire time, never darting away or looking off to the side; no mirror, no reflection, no selfie-stick, no phone-screen POV.

3. **Call `generate_video`** with:
   - `model`: `'t2v'` (Route A) or `'ref2v'` (Route B)
   - `input_urls`: `[]` (Route A) or the reference image asset_id(s) (Route B)
   - `prompt`: the paragraph from step 2
   - `orientation`: `'portrait'` for a 9:16 short (default for this format); use `'landscape'` only if the user asks for widesc