Hotel guest welcome video: an avatar in your lobby photo, 8 properties

For a hotel welcome video, create one avatar for the brand and pass each property's lobby photo as scene: { "type": "photo", "image_url": ... }, so the same host greets guests in a different place per site. Eight properties at 25 seconds each cost $36.80 on Standard, $49.00 on Plus and $110.00 on Max.

The scene field and the 4-60 second limit are in Sume's avatar video guide, read 2026-10-04. Scene photos must be public HTTPS image URLs; signed, private or non-image URLs are rejected before the job is submitted.

One host, many scenes

A consistent presenter makes the brand feel coherent across properties, and the scene photo carries what is local. Keep the same avatar handle in every request and change only scene.image_url and the property facts in the script.

Per-property request

Use the property code in the Idempotency-Key so a retry cannot double-bill the same site.

curl -X POST https://api.sume.com/v1/avatar-1.0/talking-video \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: welcome-prop-03" \
  -d '{"avatar_handle": "brand_host", "aspect_ratio": "16:9", "quality": "plus", "scene": {"type": "photo", "image_url": "https://example.com/properties/03/lobby.jpg"}, "script": "Welcome. Breakfast is served from six thirty to ten in the garden room, and the front desk is open all night if you need anything."}'

Keep the facts in data, not in the script

Store opening hours and amenities in a small table per property and fill the script from it. When a fact changes, re-render only that property. A spoken fact cannot be edited later, only re-rendered.

What the guest should know

Use the clip on a welcome page or in the pre-arrival email. Say in the clip, the caption or the page around it that the presenter is AI-generated; the API does not add that label for you.