You already paid for product photography. It sits on your product pages doing its job, and it does nothing at all in a feed, where a still image is close to invisible. The obvious move is to turn those photographs into video, and the obvious objection is that this normally means an editor, a week, and a budget.
What actually happens
You upload a photograph, or paste a product URL and let the importer pull the imagery. The AI reads the product, then writes a plan before rendering anything: hook, benefit, call to action, with a decision about what the camera does at each moment. That plan is what gets rendered.
This is the part that distinguishes a video ad from a slideshow. A tool that skips planning gives you your photographs with a pan and some music, which looks like video and behaves like a gallery. A planned sequence poses a question in the first second and answers it in the last.
Which photographs work best
- Clean, uncluttered background. A product isolated on white or a plain surface gives the AI the clearest subject to work with.
- Sharp and reasonably large. Roughly 1000 pixels on the shorter side or better. A thumbnail upscaled produces a soft video, and no amount of planning fixes that.
- Product centred and fully in frame. Cropped edges become odd artefacts in motion.
- No text or watermarks baked in. Overlaid text distorts as the frame moves, and it competes with whatever the ad wants to say.
- One product per image. A photograph containing three variants asks the AI to decide which one is the subject, and it may not choose the one you meant.
If your best photograph fails one of these, it is usually faster to reshoot that single image on a phone against a plain wall than to fight the result.
Getting several videos from one photograph
Each generation plans its own angle rather than recolouring the last one, so running the same photograph twice gives you two genuinely different ads. This matters for testing: comparing two crops of one idea teaches you nothing, while comparing two ideas tells you which one your audience responds to.
A practical routine is to generate two or three, run them against each other with a modest budget, keep the winner, then generate new challengers against it.
Formats and what you get
Choose 9:16 for Reels, Stories and TikTok, 1:1 for feed placements, 16:9 for YouTube and landing pages. Generate for the placement rather than cropping into it later; a cropped video puts your product exactly where the interface sits.
You get a standard MP4, 720p by default and 1080p on the premium models. The €5 entry offer produces one short clip of about 12 seconds so you can judge the result on your own photographs before committing. The paid plans produce multi-chapter videos of 24, 32 or 36 seconds.
The honest limit
The AI works with what is in the image. It cannot show you an angle the photograph does not contain, and it cannot invent a detail that was never photographed. If your product has a feature that matters and no photograph of it exists, take that photograph first. Everything downstream depends on it.
Related reading: Ways to create your video ad and Video ads without an editor.