Real Estate Listing Videos From One Photo With Wan 3.0
Listings with video hold attention longer than listings without. Most agents have neither a videographer nor a second visit to the property, but they do have a folder of listing stills. Image-to-video turns each of those into a short motion clip, one photo per room, and alibaba/wan-3.0-i2v currently runs a ten-second 1080p clip for $1.60 per run.
Key Takeaways
- One still per room becomes one clip; no reshoot, no gimbal.
- A 10-second 1080p run costs $1.60; a 2-second 480p test costs $0.08.
- Subtle camera motion only. Invented geometry is a compliance problem, not a style choice.
- Disclose AI-generated motion in the listing.
Which stills work
The source photo does most of the work. What holds up:
- Wide room shots with clear depth, shot straight rather than at an angle
- Clean verticals — door frames and wall corners that are actually vertical
- Even daylight, curtains open, lights on
What fails: mirrors and glass with reflections the model will reinterpret, heavy clutter, low light, and anything mid-renovation. A still that a buyer would question is a still that becomes a clip a buyer will question.
Prompting for property motion
Each element controls one thing. Testing shows that naming the static elements explicitly matters more here than in any other use case, because the failure mode is not an ugly clip, it is a misleading one.
Camera move. Pick one and keep it slow. Slow push-in toward the far window, steady and even. Or gentle pan from left to right across the room, constant speed. Handheld shake and fast moves both read as cheap.
What must not move. State it every time. Furniture, walls, fixtures and window frames remain completely static. This is the clause that keeps the model from redecorating.
Light. Preserve the existing lighting and time of day. Do not relight the room. Relighting is the most common silent change, and it is also the one most likely to misrepresent a north-facing room.
Duration. Five seconds for an interior room, ten for a hero exterior or a long hallway where the move needs room to breathe.
Stacked: Slow push-in toward the far window, steady and even. Furniture, walls, fixtures and window frames remain completely static. Preserve the existing lighting and time of day.

The line that matters in this industry
Motion is fine. Invention is not. A generated clip must not add a room, remove a defect, widen a space, or change the light in a way that makes the property look different from what a buyer will walk into. In many jurisdictions listing imagery counts as advertising, and an enhanced photo that misleads is treated as such regardless of how it was produced.
Two practical habits cover most of it: keep the static clause in every prompt, and disclose that motion was AI generated in the listing description. Neither costs anything, and both are cheaper than a complaint.
What a listing costs
Cost per run on alibaba/wan-3.0-i2v:
Run | Cost |
2s / 480p | $0.08 |
5s / 720p | $0.40 |
5s / 1080p | $0.80 |
10s / 1080p | $1.60 |
30s / 1080p | $4.80 |
Siray's site banner currently reads "Wan 3.0 Launch Month — Prices Dropped Again. Now Up to 33% Lower," and these are the discounted rows. A twelve-room property at five seconds each, plus one ten-second exterior, comes to under $12. Test at 2 seconds and 480p first; at $0.08 a run there is no reason to debug a prompt at full resolution.
Wan 3.0 covers other verticals with the same workflow: product video ads for small business, marketing hero video and animated explainers. Rate history is in the Wan 3.0 launch pricing post; check the console for current numbers.
Start generating
Create your free Siray account and turn your listing photos into video with Wan 3.0 now.