
Veo 3 is Google's flagship AI video model, and in 2026 it is the strongest all-round option for property video. It generates high-resolution clips with synchronized native audio, a first for a Google video product, and, the part that matters for listings, it does image-to-video: hand it one photo and it animates a cinematic camera move through the scene. No drone, no crew, no filming day. The Veo 3.1 update (October 2025) sharpened the image-to-video path further.
Our articles are still written by humans!
Get human written articles in your Google feed.
The catch is that Veo 3 is a general video model, not a real estate tool. It will happily generate anything, which means the burden of getting a clean, believable listing clip is on you: what you feed it, what you type, and one step almost everyone skips. Below is exactly how to turn a listing photo into a walkthrough with Veo 3, the prompts that work per room, what it costs, and the faster path if you would rather not juggle any of it.
Quick answer: To make a real estate video with Veo 3, stage or clean up your listing photo first, upload it in image-to-video mode, and prompt one slow camera move (a dolly through the entry, an orbit around the kitchen island), four to eight seconds, no people in frame. Repeat per room and stitch the clips into a 15 to 30 second tour. The biggest quality factor is the photo you start with, not the prompt.
FreeUpload photo · free to startYou have seen the bad AI videos: warped faces, seven-fingered hands, people gliding like they are on ice. That is real, and it is why plenty of agents do not trust AI video yet. So here is the thing worth understanding before you spend a credit.
Almost every weakness of AI video comes from two things: people and fast motion. Faces, limbs, crowds, quick physics. A listing video has none of that. It is a slow camera move across a still, empty room, which is the single easiest thing a video model can do. The exact failure modes that make AI video look fake barely touch property footage. That is why Veo 3, pointed at an empty staged room, produces clips that genuinely read as filmed.
Veo 3 adds two things on top that suit real estate specifically. It can synthesize matched ambient audio, the soft room tone of an unfurnished space or footsteps moving from hardwood to rug, which sells the realism. And it handles image-to-video cleanly, so you start from your real listing photo rather than an invented room. The whole game reduces to two rules: keep people out of frame, and keep the camera move slow and singular.
The workflow is image-to-video, and it is quick once you have a photo worth animating. For one room:
Which camera move for which room is most of the craft. This table is the shorthand:
| Room or shot | Veo 3 camera move to prompt |
|---|---|
| Entry / hallway | Slow dolly forward through the doorway into the room beyond |
| Kitchen | Gentle orbit around the island or a glide past the countertop |
| Living room | Slow dolly out to reveal the full size of an open-plan space |
| Primary bedroom | Soft dolly in toward the bed and the window light |
| Bathroom | Slow pan across the vanity and tub, steady and even |
| Exterior | Crane up over the facade at golden hour, or a low drone push toward the entrance |
If you would rather see the image-to-video workflow in motion first, this short walkthrough runs real estate photos through Veo 3 end to end, a useful check on what the output actually looks like before you try it on your own listing.

A Veo 3 real estate workflow, listing photos to video. Third-party tutorial by The Social Theory on YouTube.
The takeaway is the one this guide keeps returning to: keep people out of frame, keep the camera move slow, and above all, the clip is only as good as the photo you feed it, which is what the staging step below is about.
Paste one of these into Veo 3's image-to-video prompt box after uploading the room photo. Keep the room well lit, the move slow, and the frame empty of people. We later ran 29 longer real estate video prompts through Veo 3.1 and published every clip, including the eight that failed.
Rule of thumb: one slow, well-chosen move per room beats a dozen flashy ones. If a clip looks off, it is almost always the move that was too fast or the photo that was too busy, not the wording.
Here is the honest truth that separates a listing video that sells from one that gets scrolled past. Veo 3 animates whatever is in your photo. If the room is empty, dated, or cluttered, you get a beautifully animated empty, dated, or cluttered room. The camera move cannot fix the space, it only moves through it.

So stage the photo before you ever open Veo 3. This is where MeltFlex does the work no video model does: upload a photo of the actual room, empty or dated, and it furnishes or restyles it into something worth filming while keeping your real walls, windows, and proportions. You get a clean, room-accurate image to feed straight into Veo 3.

Same room, same layout, now a scene a buyer wants to walk through. Stage first in MeltFlex's create tool, then animate. Skip this step and no prompt will save the clip.
Veo 3 is powerful, but for a listing you are stitching together three separate jobs: staging the room somewhere, writing prompts, and generating and assembling clips one at a time. If you make videos occasionally that is fine. If you list every week, it is a lot of steps for a 20 second tour.
This is why we built the MeltFlex AI video walkthrough. It wraps the same class of image-to-video technology into one flow made for property: you upload a room photo, MeltFlex can stage it first, and then you pick a camera style, cinematic, walkthrough, orbit, drone, golden hour, instead of writing a prompt. One click and you get the clip.
A MeltFlex walkthrough generated from a single interior photo. Camera style picked, no prompt written.
The same one-photo-in approach works for exteriors and architectural renders, useful for marketing a project before it is built.
The trade-off is honest: clips are short, so a full tour is several stitched together, and it is built for rooms and buildings rather than open-ended filmmaking. If you want to direct a complex multi-shot scene, a raw model like Veo 3 gives you more rope. For a listing, that rope is mostly unnecessary. You can try the walkthrough tool here and see the difference on your own photo.
Google lists 22 October 2026 as the shutdown date for the Veo 3.1 preview models on the Gemini API and points developers to Gemini Omni 1.1 Flash instead. In Google Flow, subscription credits now run on Gemini Omni Flash too. The prompts and the staging rule in this guide carry over; the model name and the price per second change.
| Model | Status on the Gemini API | Price per second of video |
|---|---|---|
Veo 3.1 (veo-3.1-generate-preview) | Shutdown date 22 Oct 2026 | $0.40 at 720p or 1080p, $0.60 at 4K |
Veo 3.1 Fast (veo-3.1-fast-generate-preview) | Shutdown date 22 Oct 2026 | $0.10 at 720p, $0.12 at 1080p, $0.30 at 4K |
Veo 3.1 Lite (veo-3.1-lite-generate-preview) | Shutdown date 22 Oct 2026 | $0.05 at 720p, $0.08 at 1080p |
Gemini Omni 1.1 Flash (gemini-omni-1.1-flash) | The replacement | About $0.10 at 720p ($17.50 per 1M video tokens) |
Read 14 October 2026 from the Gemini API pricing and the Gemini API deprecations page. We moved our own video walkthrough tool from Veo 3.1 Fast to Gemini Omni 1.1 Flash on 10 October 2026. What we measured: an 8 second 720p clip used about 47,000 output tokens (roughly 82 cents) and an 8 second 1080p clip about 70,000 (roughly 1.23 dollars); clips came back in 30 to 60 seconds; and there is no duration or negative prompt field, so the clip length and the things to avoid go into the prompt text (“A 4-second video.”, “Avoid: people, text”). An end frame worked at both 4 and 8 seconds, which Veo 3.1 Fast only allowed at 8.
On the Gemini API, yes: Google lists 22 October 2026 as the shutdown date for all three Veo 3.1 preview models (Standard, Fast and Lite) and recommends Gemini Omni 1.1 Flash as the replacement, at about 0.10 dollars per second of 720p. In Google Flow, subscription credits already run on Gemini Omni Flash. The prompts in this guide work on Omni too; state the clip length in the prompt, because Omni has no duration setting.
Veo 3 runs on a Google AI subscription or the Gemini API, where a short clip is about a dollar in 2026 (8 seconds of 720p on Gemini Omni 1.1 Flash cost us roughly 82 cents), so a stitched tour is a few dollars of compute against the 1,000 to 1,500 dollars a filmed listing video used to cost. Pricing moves often, so check Google's current rates before you budget. The MeltFlex walkthrough folds into a normal plan with the staging, so there is no separate video model to meter. If Veo's per-clip price is what stops you, the budget route is Kuaishou's model instead: our guide to making listing videos with Kling AI runs the same pipeline at roughly a tenth of the cost per second.
The reason to bother is demand-side, though not for the number you usually read. The claim that listings with video get 403 percent more inquiries is credited to the National Association of Realtors almost everywhere, but it is not an NAR figure; we traced it to a single Australian agency in 2012 in our guide to the best AI video generator for interior design. What holds up is the cost gap: only a minority of agents produce listing videos because traditional videography is slow and expensive, and a walkthrough from a single staged photo removes that cost, which is the whole point of doing it with AI.
Veo 3 is a genuinely strong way to turn a listing photo into a cinematic walkthrough, as long as you respect the two rules, keep people out and the move slow, and, above all, stage the room before you animate it. Do that and the results look filmed.
If you want the shortest path, stage your room in MeltFlex and generate the clip with the AI video walkthrough in one place. If you want to see how Veo 3 stacks up against Kling, Runway for real estate video and the rest before you pick, our roundup of the best AI video generators for real estate tests them side by side, and if you prefer a preset-driven model with fine camera control, our guide to Higgsfield for real estate covers that approach.
Yes. Veo 3 supports image-to-video, so you upload a single photo of a room or the exterior and it animates a cinematic camera move through the scene, no filming and no drone. It does not walk through a real space, it invents believable motion around the still image you give it. That is why the photo you start with matters more than the prompt: a clean, staged, well-lit shot produces a premium clip, while an empty or cluttered one produces an empty or cluttered clip.
On the Gemini API, Veo 3.1 Fast costs 0.10 dollars per second at 720p and 0.12 at 1080p up to the Veo 3.1 preview shutdown date Google lists, 22 October 2026; the replacement, Gemini Omni 1.1 Flash, is about 0.10 dollars per second at 720p, and in Google Flow subscription credits now run on Gemini Omni Flash. A short listing clip lands at roughly one dollar, so a full tour stitched from several clips is a few dollars rather than the 1,000 to 1,500 dollars a filmed listing video traditionally costs. Pricing changes often, so check Google's current plans before you commit a budget.
The best Veo 3 real estate prompt names the room, the camera move, the speed, and the light, and nothing else. For example: "slow dolly forward through the front entry into a bright open-plan living room, eye-level, smooth steady gimbal motion, warm natural daylight, photorealistic, no people". Keep the move slow and singular, keep people out of frame, and describe the lighting. Fast moves and crowds are exactly what makes AI video look fake, and a listing has no need for either.
Yes, and it is the step most people skip. Veo 3 animates whatever is in your photo, so a dated, empty, or messy room becomes a beautifully animated dated, empty, or messy room. Stage or restyle the photo first, then animate the clean result. MeltFlex furnishes or restyles a real room photo in seconds and keeps your actual walls, windows, and layout, which gives Veo 3 a scene worth moving through.
Veo 3 gives you the most raw control and the highest ceiling on quality, but it is a general video model: you stage elsewhere, write prompts, generate clips, and stitch them yourself. If you just want to upload a room photo and get a walkthrough without any of that, an all-in-one tool like MeltFlex is faster, because it stages the room and produces the walkthrough in one place with a camera-style picker instead of a prompt. Many agents use MeltFlex for volume and reach for a raw model like Veo when they want a single hero clip.
Probably, but be careful which number you repeat. The 403 percent more inquiries figure is usually credited to the National Association of Realtors, but it is not an NAR study; it traces back to a single Australian agency quoted in the trade press around 2012. The case that holds up is simpler: a clip holds attention on a portal longer than a photo carousel, and an AI walkthrough from one photo costs a few dollars against a videographer, which is why the cost gap is the real argument.