Wan3.0 for Real Estate: Turn a Listing PDF into a 30-Second Video

Short answer: attach the listing brochure as the document input, then write a director's brief that tells Wan3.0 which pages are the visual truth, in what order the rooms appear, what each caption says and what it must never change. Done that way, a 20-second 1080p listing film costs $4 and takes one pass. Done as "make a video from this PDF", the model renders your floor plan as a room and adds a bedroom you do not have.

What Wan3.0 does that older models could not

• Documents as input. PDF, DOC, PPT, XLS, TXT, MD, Keynote, Pages, Numbers — one file or link, up to 100 MB and 50 pages. A brochure, an exposé or the listing deck goes in directly.
• 2 to 30 seconds in one pass, 480p, 720p or 1080p, with the prompt acting as the control layer that decides what the document becomes.
• API-only. Alibaba Cloud Model Studio and aggregators such as OpenRouter; there are no open weights. For local GPUs, Wan 2.2 stays the last open release.
• Per-second pricing: $0.05 (480p), $0.10 (720p), $0.20 (1080p) list price in September 2026, with a 30% launch discount until September 23.

Prepare the brochure first

The video can only be as honest as the document. Before you upload: put the six to ten photos you want on screen in the order you want them, one room per page, hero exterior first; keep the floor plan on its own page; put the facts you want as captions in one table (size, rooms, year, price, viewing date); remove agent portraits, logos and QR codes unless you want them rendered; and keep it under 50 pages. If a photo is virtually staged, its label must already be on the photo (see the labeling section).

The brief structure

Wan3.0 follows the same 2026 rule as Seedance 2.5 and Kling 3.0: structured blocks, one visible action per beat, no keyword soup. For a document source one block changes — REFERENCE ROLES becomes SOURCE ROLES, where you assign a job to each page.

1. FORMAT — length, aspect ratio, resolution, where it will run.
2. SOURCE ROLES — which pages are visual truth, which page only orders the rooms, which page feeds the captions, and what must never be rendered.
3. STARTING STATE — where the camera and light are at 0 s.
4. TIMELINE — second ranges, one room and one camera move per beat, the caption for that beat, and where the last move settles.
5. CAMERA — one move type, height, lens, what is forbidden (drone, whip pans).
6. CONTINUITY — rooms, furniture, finishes and views fixed to the photos; consistent light and caption style.
7. AUDIO — music bed or voice-over, or explicitly none.
8. CONSTRAINTS — no structural changes, no invented rooms, no people, no extra text.

A complete example (20 s, 16:9, 1080p)

FORMAT: one continuous 20 s listing film, 16:9, 1080p, for the agency website and YouTube.
SOURCE ROLES: the attached PDF is the listing brochure. Use its photos as the only visual truth: page 1 hero exterior, pages 2-3 living room and kitchen, page 4 master bedroom, page 5 terrace. Use the floor plan on page 6 only to order the rooms; never render the plan itself. Use the facts table on page 7 for the captions.
STARTING STATE: exterior at golden hour, the house centred, camera at street level about 15 m away, still.
TIMELINE:
0-4 s - slow push-in on the facade; caption "Family villa, 186 m2, Bratislava - Ruzinov".
4-9 s - living room: the camera glides forward from the doorway toward the terrace windows; caption "Open living, 48 m2".
9-13 s - kitchen: a slow lateral move along the island; caption "Kitchen with island, renovated 2024".
13-17 s - master bedroom: gentle push-in toward the window; caption "3 bedrooms, 2 bathrooms".
17-20 s - terrace at sunset: the camera settles wide and holds; caption "689 000 EUR - viewings from Monday".
CAMERA: one smooth gimbal move per room, eye level 1.5 m, 24 mm, no whip pans, no drone.
CONTINUITY: rooms, furniture, finishes and window views exactly as in the photos; the same warm evening light throughout; captions in the same white sans-serif, lower left.
AUDIO: soft acoustic guitar bed, no voice-over, no lyrics.
CONSTRAINTS: do not add, remove or move walls, windows, doors or fixtures; do not invent rooms that are not in the brochure; no people, no pets; no on-screen text except the captions above; no watermark.

What a listing video costs

Workflow that keeps the budget flat: render the full film once at 480p for the seller's approval, fix the brief, then render 1080p once. Two passes, $7.50 for a 30-second film.

The labeling line every listing video needs

Two rules apply in 2026. In California, AB 723 (since January 1, 2026) requires every virtually staged or digitally altered listing image to carry a visible label and the original to be available on request — a video built from staged photos inherits that duty, so keep the "Virtually staged" caption on those beats. In the EU, Article 50 of the AI Act (since August 2, 2026) requires AI-generated or AI-edited video to be machine-readably marked and disclosed. Keep the original export from the model and disclose in the listing text. The full checklist is in our AI content labeling rules 2026.

Wan3.0, Seedance 2.5 or Veo 3.1?

• Wan3.0 — the source is a document and you want one 30-second pass with captions. Cheapest full listing film.
• Seedance 2.5 — you drive from photos with explicit @Image roles and need control of every shot; 30 seconds, up to 50 references. See the Seedance 2.5 prompt guide.
• Veo 3.1 — 8-second hero clips with native audio and the most realistic light; best for the social hook, not the whole tour. See AI real-estate video tour prompts.

Common mistakes

• No SOURCE ROLES — the floor plan gets rendered as a room.
• Eight rooms in 15 seconds — two seconds per room reads as a slideshow, not a film.
• "Drone shot" on an interior brochure — the model invents an exterior you never photographed.
• Captions left to the model — prices and sizes drift; write every caption yourself.
• Skipping the 480p draft — the 1080p render costs four times as much to redo.

FAQ

The Home & Property atelier in GoldenPrompts writes listing-video briefs in this structure — source roles, timeline with captions, camera, continuity, constraints and the labeling line — from a few clicks. Free to start: 24 hours of everything, no card.