For the feed, not the product page
AI product ads from your photos
An ad is a different job from a product video. It is vertical, it is seen with the sound on, and it has about a second to be worth not scrolling past. The modes below are the three that produce something you could actually run, and the last one has a rule attached.
One subscription, every mode, cancel in two clicks. The clips below were rendered by this tool from real storefront photographs.
The modes for this
3 of them apply here
Cinematic Commercial
3 CR · 5sYour product as the star of a premium commercial.
An AI model animates your photos. It looks like a commercial and it can quietly change small text, logo detail and edge proportions — always check the result before you publish it. Adding a second and third angle is the best defence against that.
Product in Real Life
6 CR · 10sYour product in a cinematic lifestyle scene.
The model builds a scene around your product and puts it in context. The scene is invented — it is a setting your product could plausibly be in, not a place it has been — and a person's hands in frame are generated too. Over ten seconds it can also redraw small text and logo detail on the product itself, so read your label in the last frame before you publish it.
Before & After
6 CR · 10sA visual transformation that fits what you actually sell.
This one shows a result, so it needs a sentence from you about what your product does — we will not guess a transformation from a photo of a bottle. What it produces is a dramatisation for advertising, not a test, a measurement or a promise of what a customer will get. Over ten seconds it can also redraw small text and logo detail on the product itself, so read your label in the last frame before you publish it.
Before you spend a credit
What makes one of these usable as an ad
Render vertical first. Nine by sixteen is the slot every feed gives you, and a landscape clip dropped into it spends most of its height on black bars instead of on the product.
Decide about sound deliberately. The model scores what it rendered, which is what stops a clip reading as a tech demo — and a merchant laying their own music and voiceover over it wants a clean picture instead. Each mode starts where it usually belongs and you can change it.
Pick a style that matches the product rather than the one that looks best in isolation. The style sets the set, the lens and the grade, and a cosmetics palette around a tyre cleaner is the kind of mismatch a viewer reads as fake without being able to say why.
Keep the claim in the copy, not in the frame. The video shows the product; the sentence that promises something is yours to write and yours to stand behind.
Output shapes
Render it in the shape you are going to post
The shape is an instruction to the renderer rather than a crop afterwards, so a vertical clip is composed vertically instead of being a landscape one with the sides cut off. Best first for this job:
- 9:16
Vertical · 1080×1920 · TikTok, Reels, Shorts
- 1:1
Square · 1080×1080 · Feed ads, Shopify grid
- 16:9
Landscape · 1920×1080 · Site hero, YouTube
The gallery
Finished clips, each beside the photo it was made from
Press one to play it; hold the button under it to see the original photograph. These are not reels we commissioned — they came out of this tool.
Bookshelf speaker
A matte cabinet beside a grille cloth — two very different surfaces, both lit by the same set rather than by a model's idea of a speaker.
Soda can
The hardest case for an AI video model and the easiest here: the artwork wraps the can, and an image-to-video model asked to move it would re-letter the whole thing.
Coffee bag
A soft pouch with small type down the front. Every word on the label is the merchant's own pixels, at every frame.
Supplement bottle
White bottle, white label, white backdrop. The cut works because it fills inward from the frame edge and stops at the bottle, instead of deleting every white pixel.
Roll-on applicator
Gloss black, which is why the set behind it is light: the renderer measures the product against the backdrop and refuses to leave a dark product on a dark one. Every line of the 8% complex panel is readable in every frame.
Oral care kit box
A box is the shape this mode suits best: flat faces and straight edges take a travelling key light the way the real thing would.
Every clip above is a Product Showcase render, made by this tool from the single public product photo behind it — hold any tile to see that photo. The shops those photos came from are not customers, have not seen this, and haven’t endorsed anything; they are here because using a real catalogue photo is the only honest way to show what this does to one. The AI modes have no published examples yet.
These are the shapes Product Showcase is good at: bottles, cans, boxes, bags and hard goods with a face worth reading. It films one photograph from a moving camera, so it never shows you a side the photograph didn’t contain — and the two things it is honestly worse at are flat-laid clothing, which has no edge for the light to catch, and products photographed against a busy background, which cut out badly. Try it on your own photo before you decide.
What this will not do
One of these modes shows a result, and it will not guess one. A model given a photograph of a bottle cannot know whether it is a tyre cleaner, a leather conditioner or a drain unblocker — so that mode asks what your product does, in your words, and what it produces is a dramatisation for advertising rather than a test, a measurement or a promise of what a customer will get. You are the one publishing it.
Pricing
One plan, every mode
StoreFix Pro
$19 / month
20 credits a month for the modes that run on a video model. Product Showcase renders on your own device and takes nothing out of them. Cancel from the billing portal.
FAQ
Questions
Do the videos come with sound?
Some modes do by default and every mode lets you choose. The soundtrack is generated to match the picture — room tone, a whoosh under a camera move — and it is the difference between a clip that reads as an advertisement and one that reads as a demo.
Turn it off if you are laying your own audio over the top, which most people running paid social are.
Can it show a before and after?
Yes, and it is the one mode with a condition on it: you have to say what your product does before it will render, because nothing in a photograph tells a model what a transformation should look like.
What comes out is a dramatisation made for advertising. It is not evidence, and it should not be captioned as if it were.
Is there a watermark on the output?
We add nothing of our own to your video — no logo, no badge, no bar across the corner. The AI modes are also told explicitly not to generate a watermark or burned-in captions of their own, which is a strong instruction rather than a guarantee, so look at the clip before you spend money putting it in front of people.
Will my label, logo or packaging change?
Not in product showcase. Nothing in that mode is generated by a model — your photo's own pixels are cut out and filmed on a lit set, so every word on your label is the word you photographed, at every frame. That is the whole reason the mode is built this way instead of handed to a video model.
The cinematic commercial, product in real life and before & after modes do run on a video model, which means small drift in fine print and tight logo detail is possible. Every one of those renders is instructed to leave your label, logo and proportions alone and to add no text of its own — but an instruction is not a guarantee, so the results are labelled AI-generated and you should look before you publish.
What happens if a generation fails?
The credits go back. A failed generation is refunded automatically, and a request that is retried — by you, by a flaky connection, or by the browser — is charged once, because the charge is keyed to the request rather than to the attempt.
A failure also tells you what failed. A video provider being down and a photo being unusable are different problems with different answers, and a studio that says "something went wrong" to both is a studio you cannot act on.
You already have the photo.
Product Showcase turns it into a finished mp4 in about the length of the clip, on your own machine, with no queue to wait in.