For Real estate agents & brokerages
Turn a phone walkthrough into a listing video worth posting
Turn a phone-shot walkthrough into a finished listing video: a generated establishing shot to open it, burned-in captions for feeds that autoplay muted, and a translated audio track for buyers who don't read your listing's base language. A 5-second AI establishing shot costs 50 credits, $1.23 to $2.25 — the kind of add-on a traditional videographer bills as a line item, if they offer it at all. None of it touches the property itself: the walkthrough has to be real footage you shot, and AI only generates what surrounds it.
The workflow
- 1
Film the property yourself — this step has no AI in it
Shoot the real walkthrough on a phone or camera: every room, the exterior, the street approach. This is the one part of this workflow that stays entirely real. Everything below adds to that footage; nothing in this toolkit should generate or edit what the walkthrough actually shows — that is the one hard rule for using AI on a listing, covered in the disclosure section below.
- 2
Generate an establishing shot that matches the setting, not the address
Describe the kind of scene the listing sits in — a tree-lined suburban street at golden hour, a coastal boardwalk, a downtown skyline at dusk — for 4 to 15 seconds on Seedance 2.0, at 16:9, 9:16, 1:1, 4:3 or 3:4, in 720p or 1080p. Seedance 2.0 is the fast default; switch to Veo for a premium hero shot (4, 6 or 8 seconds, 16:9 or 9:16). At 10 credits per second, a 5-second clip is 5 × 10 = 50 credits. Describe the type of place, never the specific street or building — this clip is atmosphere, not evidence of what the property looks like.
Open the Video Generator → - 3
Generate a still for the social post, not the listing photos
A "Just Listed" or "Open House" title card, or a lifestyle frame to open a Reel, comes from a text prompt — 8 credits per image, at 1:1, 16:9, 9:16, 3:2 or 2:3. Keep this to marketing graphics and generic scenes; never generate a version of a room or the exterior and use it anywhere a buyer could mistake it for the actual property.
Open the Image Generator → - 4
Cut it together and write the disclosure line
In any editor: AI establishing shot → real walkthrough → AI lifestyle still as a closing card. Add a line to the description or an end card — "Establishing shots are AI-generated; all interior and exterior footage is the real property, unedited." — before you post anywhere. The rule behind it is in the disclosure section below.
- 5
Burn in captions for feeds that autoplay muted
Upload the finished cut (MP4, MOV or WebM, up to 30 minutes) and leave translation off to caption it in its original language. 2 credits per second: a 55-second cut is 55 × 2 = 110 credits. The Subtitler burns captions into the MP4 only — there is no separate SRT file; if your brokerage site needs a transcript for accessibility, run the walkthrough's audio through OpenAI's Whisper, free and open source, instead.
Open the Subtitler → - 6
Translate the walkthrough for buyers who don't read your base language
Upload the same cut and pick one of 25 target languages, including regional variants such as Spanish for Spain or Latin America. You get back a new video with a translated, time-aligned audio track — 2 credits per second, so the same 55-second cut is 110 credits per language. One job makes one language; a second market means a second job.
Open the Video Translator → - 7
Caption the translated version too
Run the translated file back through the Subtitler with its new language set as the spoken language, so captions match what's actually playing — another 110 credits for a 55-second cut. The Subtitler covers 20 of the Translator's 25 languages: not Dutch, Polish, Swedish, Filipino or Malay.
Open the Subtitler → - 8
Upscale only the footage that actually needs it
An old listing shot on a years-old phone, or archived drone footage, upscales to 4K as a last step on whichever finished version you'll publish — 3 credits per second, so a 55-second cut is 55 × 3 = 165 credits. Skip it for anything filmed on a recent phone; most walkthroughs don't need it.
Open the 4K Upscaler →
What a listing video actually costs
Live credit rates as of October 2026: video generation 10 credits per second, image generation 8 credits per image, captioning and translation 2 credits per second each, 4K upscaling 3. A credit costs from $0.045 (Starter Pack: $9 ÷ 200) down to $0.0245 (Pro plan: $49 ÷ 2,000 a month).
Example: a 45-second real walkthrough opened with a 5-second AI establishing shot and closed with a 5-second AI lifestyle shot (55 seconds total), plus one social still and burned-in captions on the finished cut. B-roll: 2 × 5 × 10 = 100 credits. Still: 8 credits. Captions on the 55-second cut: 55 × 2 = 110 credits. Total: 218 credits, $5.34 to $9.81 — inside the Starter plan's 600 monthly credits, with 382 left over for a second listing's package that same month.
Add one international-buyer language: translate the same 55-second cut (110 credits) and caption that translated version too (110 credits), 220 credits per added language, $5.39 to $9.90. Domestic plus one extra language comes to 438 credits, $10.73 to $19.71 — still fits inside the Starter plan's 600 monthly credits, with 162 left over.
2026 real estate videography pricing guides put a standard professional listing video at $300 to $1,500 (a basic walkthrough $150–$250, a cinematic walkthrough with drone footage $300–$500 plus $150–$500 for the drone separately) — before a translated version, which traditional videographers rarely offer at all. Premium, multi-day cinematic production for luxury listings runs $2,000 to $10,000 or more; this workflow isn't a substitute for that tier, since the AI layer only adds B-roll, captions and translation around footage you still have to shoot well.
The property must be filmed — AI only generates what's around it
Can: a generated establishing or lifestyle shot, a social graphic, burned-in captions, a translated audio track, a sharper upscaled file. Cannot: a generated or edited version of the property's actual interior or exterior, under any circumstance. Making a room look bigger, replacing flooring, removing clutter that's actually there, or adding a feature the home doesn't have is a misrepresentation risk everywhere, not just a compliance technicality — and it's also outside what every tool in this workflow does, since they generate new scenes or re-render audio, not edit an uploaded photo of a real room.
The one tool this workflow deliberately leaves out is the Video Editor (/edit), which changes what's visibly in an uploaded clip from a text prompt. Running the property's own walkthrough footage through it to change what a room shows would cross from "generating B-roll around the property" into "altering the property" — exactly what the rule above forbids, and exactly what the disclosure law below is aimed at. Use /edit, if at all, only on the generated B-roll, never on real footage of the listing.
Disclosure: what 2026 actually requires
California's AB 723, effective January 1, 2026, requires any photo of a property that's digitally altered to add, remove or modify a physical feature — virtual staging, an AI redesign, a color, flooring or landscaping change, object removal — to carry a visible "Virtually Staged" or "Digitally Altered" label: high-contrast text, roughly 4 to 6 percent of the image's height, in a corner where it survives syndication to portals like Zillow and Redfin. It also requires the unaltered original to be available, a written note in the listing remarks, and the label to travel with the image everywhere it's reposted — MLS, brokerage site, social, ads. Standard editing (brightness, contrast, white balance, HDR, lens correction) and removing temporary non-property clutter are exempt. As of October 2026, California is the only state with a law this specific; check your own state and MLS rules before assuming it doesn't apply to you.
Strictly, AB 723 is about altered photographs of the property — which this workflow must never produce in the first place. The generated establishing shot and lifestyle still here aren't photos of the home, so the letter of the law likely doesn't reach them. The National Association of Realtors' Code of Ethics does, though: Articles 2 and 12 require presenting a "true picture" in advertising and bar "misleading images," without carving out an exception for AI, and that duty applies nationwide regardless of which state has passed a specific statute. One industry audit found that over 90% of AI-altered listing photos currently carry no visible disclosure anywhere online — that's a measure of how little this is enforced today, not a sign that it's safe to skip. Add the disclosure line from step four of the workflow to every video that contains a generated clip, whether or not your state has a law on the books yet.
Pre-publish QA checklist
Check every listing video against this before it goes anywhere public:
- The disclosure line is in the description or on an end card, naming exactly which shots are AI-generated.
- No generated clip could be mistaken for the property itself — review it as a buyer would, not as the person who wrote the prompt.
- Captions match the numbers spoken in the walkthrough: address, square footage, price, HOA fees, offer deadline.
- A translated version's numbers, currency and any legal or as-is language are checked by a native speaker, not just back-translated.
- The aspect ratio matches where it's posted — 9:16 for Reels and Stories, 16:9 for a listing page or YouTube.
- Your MLS and brokerage's own photo and video policy is checked separately — this page covers California law and NAR's general duty, not every local MLS rule.
Limits to know before you commit
Plan a listing's video package around these:
- Generated clips run up to 15 seconds each, with the exact lengths set by the model (Veo 3.1: 4, 6 or 8 seconds); assemble several in your editor for anything longer.
- One target language per Video Translator job, and up to 30 minutes per file on the Translator, Subtitler and 4K Upscaler.
- The Subtitler burns captions into the MP4 only — no SRT or VTT file; for an editable transcript, use Whisper (free, open source) instead.
- Dutch, Polish, Swedish, Filipino and Malay are available for translated audio but not for captions.
- Neither tool translates on-screen text, so a price or "Open House" overlay you added needs its own swap per language.
- Generated video tops out at 1080p; upscaling only sharpens real footage you upload, it doesn't add detail that was never captured.
Questions
Can AI generate or edit photos of the property itself?
No — not with any tool in this workflow, and not as a matter of policy either. The Video and Image Generators create new scenes from a text prompt; they don't take a photo of your listing as input and alter it. Treat that as a hard rule, not just a tool limitation: a generated or edited version of the actual interior or exterior is a misrepresentation risk under NAR's Code of Ethics everywhere, and specifically regulated in California under AB 723 if it ever happened to a real photo.
Do we have to disclose AI-generated B-roll in a listing video?
There's no law yet that names "AI establishing shots in a listing video" specifically — California's AB 723 (effective January 1, 2026) covers altered photographs of the property, which this workflow never produces. But NAR's Code of Ethics requires a "true picture" in advertising nationwide, and a viewer who can't tell a generated skyline from the actual view has been misled regardless of which law applies. Add the disclosure line every time, in every state.
Can one Video Translator job produce more than one language?
No. Each job produces one language, chosen from 25 targets including regional variants of English, Spanish, Portuguese and Chinese. Marketing to three buyer segments in three languages means three jobs and three times the credits.
Do we get a separate caption or transcript file?
Not from the Subtitler — it delivers captions burned into the MP4, nothing separate. If your brokerage site needs a transcript for accessibility or SEO, run the walkthrough's audio through OpenAI's Whisper, free and open source, for an editable SRT on your own computer.
What does a test listing video cost?
The 20 free credits at sign-up don't cover a video clip — the shortest, 3 seconds, costs 30 — but they cover two still images (2 × 8 = 16) to check a style, with no watermark. For a real listing, two $9 Starter Packs (400 credits) cover the full example package above (218 credits); the Starter plan, $19 for 600 credits a month, covers it with 382 credits left for captions or a translated language on the next listing. Failed jobs are refunded.
Does this replace hiring a videographer?
Not for every listing. It's built for the ones that wouldn't get a video at all otherwise — lower-priced listings where a $300–$1,500 production fee doesn't pencil out. For a luxury listing where cinematic quality is itself part of the pitch, a professional shoot is still worth it; use this workflow's captions and translation on top of that footage instead of replacing it.
Read next
Pay only for what you process
Credits are charged per second of video or per image, and failed jobs are refunded.
See pricing