Real estate agent AI video: one avatar for every listing

A real estate agent can make AI videos without filming each one: create an AI avatar once from a photo of yourself, then turn each script, such as a listing intro, an open-house invite, or a market update, into a talking video of that avatar. On Sume that is Avatar 1.0, with up to 60 seconds per video and a listing photo as the scene reference. Know two things first: in current code the avatar is a generated likeness based on your photo, and it speaks English in a generated voice, not yours.

Facts come from Sume's Create new avatar, Generate avatar video, and Avatar video previews docs and the Sume API reference, read on 2026-09-27; anything called current behavior is read from Sume's code. For a tour made from the listing photos themselves, see Real estate listing video from photos.

How do I make an AI video of myself as a real estate agent?

• Create the avatar once with POST /v1/avatar-1.0/generate: a stable avatar_handle and input: { "type": "photo", "image_url": "https://…" }, a public HTTPS photo of you. Creation is a job; when it completes, reuse the handle for every video. Create a reusable AI avatar shows the request.
• Write each script so Sume estimates it at 4–60 seconds, and split a longer one into several videos.
• Send the script and the handle to POST /v1/avatar-1.0/talking-video. aspect_ratio takes 1:1, 3:4, 9:16, 4:3, or 16:9 (default 9:16), and 720p is the documented resolution.

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: listing-intro-001" \
  -d '{
    "avatar_handle": "acme",
    "script": "Welcome to 12 Harbor Lane: three bedrooms, a sunny kitchen, and a garden made for summer evenings. Book a tour this weekend.",
    "scene": { "type": "photo", "image_url": "https://example.com/listing/kitchen.jpg" },
    "aspect_ratio": "9:16"
  }'

Can I put a listing photo behind the avatar?

Yes, as a reference: scene: { "type": "photo", "image_url": … } is what the docs call a photo scene reference, and the photo must be at a public HTTPS URL. The docs don't promise the room appears exactly as photographed, and a video resolves to one shared scene, so plan one room per video. Keep your real listing photos as the record of the property. AI avatar video with your product and background covers the scene fields.

How do I check the first frame before the full render?

Send the same body to POST /v1/avatar-video-previews first. It generates the first-frame still, preview_image_url, without starting the full render, so you see the avatar in front of the backdrop. POST /v1/avatar-video-previews/:id/regenerate refreshes the stills, and POST /v1/avatar-video-previews/:id/generate-video starts the final video, reusing the preview's first frame when available. Put captions on the preview create; they are burned at generate-video, and in current code the talking-video create doesn't accept them. Avatar video previews covers the flow.

Will the avatar look and sound like me?

Not exactly. In current code, a photo avatar is redrawn by an image model from your photo, so it is a generated likeness rather than your picture; its voice is generated to match that look and cloned as English; and it speaks English only. The talking-video request has no field for choosing a voice.

To appear with your own face and voice, or to reach clients in another language, clone your voice in the Sume app, speak the script with TTS 1.0 (set language for a non-English script), and lip-sync your own photo to that audio with VEED Fabric 1.0. How to clone yourself with AI covers that route. Use only your own likeness, or a colleague's with their consent.

What does an AI real estate video cost?

Avatar creation is a one-time charge of $0.95 per avatar. Each video bills per second by quality tier, and quality defaults to plus. Every amount is plus a 5.5% agent fee by default.