AI ad maker: image and video ads from one product photo
This AI ad maker starts from the product photo you already have. Upload it, answer two questions, and it returns the three ad formats that fit the product, ready to make as an image or a short video.
Step 1Your product photo
One product per photo. If a person is in it, only use a photo you have their permission to use.
Step 2What the ad has to do
2 credits, and it reads the photo once however many formats come back.
How the match is made
The photo answers what the product is. A classifier reads the category, whether it is worn or held, and whether there is a face in the frame. Those are all constraints: a kettle cannot be a try-on however much you would like it to be.
The two dropdowns answer what the ad is for and who you can put on camera. Neither is in the picture. A knit top is equally compatible with selling it on TikTok Shop, showing how soft it is, launching a brand and clearing old stock, and those are four different ads.
Formats score on those three. Category and job are worth the same, because a try-on for a kettle and a testimonial for a dress are both wrong and neither is more wrong. Ties break on the library's own order, so the same photo and the same answers give the same three every time. When nothing in the library is built for your product you get hands-only demos and a line saying so, rather than a consolation prize dressed up as a match.
What comes out
Real generations from the library, not mock-ups. Hover to play.
Loading samples…
What this AI ad maker makes from one photo
One product photo can become three kinds of ad here. Each has its own page, and each shows its price before it runs.
A product image ad
Your product placed in a clean scene, square or vertical
3 credits ($0.11) per image
Make an image adA short video ad
A 5 to 15 second vertical ad in a format matched to the product
40 to 120 credits ($1.49 to $4.48)
Make a video adA remake of a trending format
A template that is already working, retaken with your product
120 credits ($4.48) for 15 seconds
Browse templatesImage ad or video ad: which to make first
| If the ad is for | Make | Why |
|---|---|---|
| Filling a product listing or a catalog feed | Image | Marketplaces and catalog ads show stills; a clean scene does the job |
| Retargeting people who already visited | Image | They know the product; a reminder does not need a hook |
| Reaching people who have never seen the product | Video | The first seconds of motion are what stops the scroll |
| Showing what the product does | Video | A demo, an unboxing or a before and after needs time to happen |
Starting from a listing photo? The photo-to-video guide covers what one photo can and cannot become, and the product photo prompt library has GPT Image prompts for the still shots.
Five checks before you run an AI-made ad
- The product in the ad matches the listing: color, shape, size, what is in the box.
- No logo, sign or label you do not own appears anywhere in the frame.
- Nobody in the ad is a real, identifiable person you do not have permission from.
- The ad is labelled as AI-generated where the platform asks for it.
- The shape fits the placement: 9:16 for Reels, Shorts and TikTok, 4:5 or 1:1 for feeds.
The first check is the one platforms enforce most directly. TikTok, for one, treats an AI ad whose product no longer matches the listing as misleading; the AI UGC for TikTok Shop guide quotes its rules.
Questions about the AI ad maker
What can an AI ad maker make from a single product photo?
Here, three things. A product image ad, where the product is placed into a clean scene you choose. A short vertical video ad, where the agent matches the photo to a format such as a demo, an unboxing or a before and after, and renders it. And a remake of a trending template with your product swapped in. All three start from the same photo, and each shows its credit price before anything is generated, so you can try an image first and move to video once you know which angle you want to push.
Should my first AI ad be an image or a video?
It depends on who is going to see it. If the ad is for people who already know the product, for example retargeting or a catalog feed, an image is cheaper and does the job. If it is for people who have never heard of you, video usually earns its higher price, because motion in the first second is what stops someone scrolling past. A sensible order is one image to check that the product reads well in a scene, then two or three short videos with different hooks.
Why does this AI ad maker suggest formats instead of asking for a prompt?
Because many sellers do not know what to type, and the format decides more about an ad than the wording does. The finder at the top of the page looks at the photo, asks what the ad has to do, and returns the three formats that fit that product, each with its length and price. Formats written around one kind of product, such as a lipstick or a jacket, are withheld from products that are not that kind, so a serum is never offered a format whose shots assume a lipstick.
Will the AI ad maker keep my product exactly as it is?
It is built to, and it usually does, but check every result before you spend on it. Image and video models can shift a color, a cap shape or a label, and on TikTok Shop an ad whose product no longer matches the listing is treated as misleading. The practical check takes seconds: compare the first and last frame, or the finished image, with your listing photo. A tight, well-lit photo of the product on a plain surface is the single biggest thing you can do to keep it accurate.
How is an AI ad maker different from an ad template editor?
A template editor gives you a designed layout and you fill it in: your photo goes into a frame, your text into a box, and the result looks like the template. An AI ad maker generates new footage or a new scene around the product, so the ad can show the product being used or in a setting you never photographed. The trade-off is control: an editor never changes your product, while a generator can, which is why the accuracy check above matters more here than in an editor.