Short Brand Clips: 9:16, Sound, and the Parts That Still Need a Human

Picture the brief most small shops actually have. There is a candle, a mug or a jacket on the counter, one decent photo of it, and a feed that wants something moving by Friday. Nobody is booking a videographer for a seasonal post, and nobody has an afternoon to learn an editing suite.

That brief is where AI video is most useful, and also where it is easiest to oversell. So this comes in two halves: what is genuinely simple here, and the parts we would still leave to a person.

What is genuinely easy

Starting from your own photo.Fellowi Video can animate a still you upload, and that is the path a brand wants: the product in frame is the one you photographed, not one the model imagined from a description. Steam rising off the mug, the jacket turning slowly on its hanger, light moving across a candle. The model's job shrinks to adding motion, and the difference between the two paths is laid out in text to video or image to video.

Vertical from the start. 9:16 is one of the aspect ratios you pick before generating, so the clip is composed for a phone rather than cropped down from a wide frame afterwards. If your photo is already vertical, the auto setting keeps its framing as it is. The craft of the vertical frame itself is in our Reels and TikTok guide.

Sound if you want it. A generated soundtrack is a box you tick, and it stays off unless you do. On a brand feed you will usually leave it off and choose a track inside the platform, but a clip that goes out exactly as it is, in a newsletter or on a product page, can carry its own.

What it costs, and what it does not ask of you

Video is paid in coins, bought once. There is no subscription to start, no monthly minimum, and coins do not expire, so a shop that makes three clips before the holidays and nothing in January is not paying for January. Each clip is priced by model, resolution and length; the current figures are on the product page, which reads them from the same code that charges you, so we do not repeat them here.

Two things make that price the whole price. A generation that fails returns its coins automatically, as described in what a failed generation costs you. And our terms place no restriction on commercial use: post the clip on your shop's account, run it as an ad or put it on a product page, with no share and no credit line. The details are in the commercial-use post.

How fast “fast” is

Faster than a shoot, slower than a photo filter. A clip takes minutes rather than seconds to render, and sometimes noticeably longer when our provider is busy. We published a month of real wait times in how long a generation takes. The rhythm that works is to start a few short drafts, go and serve a customer, and come back to choose.

Clips that pass, and clips that read as cheap

The clips that hold up are small, physical and slow: steam, fabric settling, a reflection sliding across glass, a gentle push towards the product, snow falling past a shop window for a seasonal loop. They look like a moment rather than a production, and short ones loop cleanly on a feed.

The ones that give themselves away ask the model to invent something complicated: hands picking the product up and using it, a person trying the jacket on, fast camera moves around a small object, anything where the viewer knows exactly how the real thing should behave. If a shot needs choreography, it needs a camera.

The parts that still need a human

The hero shot is still a photo shoot. Image to video moves the picture you give it; it does not improve it. A dim, cluttered phone photo becomes a dim, cluttered clip. The one good photograph of the product is still your job, and it sets the ceiling for everything the generator does with it.

Logos and lettering are not reproduced faithfully. Text on packaging, a printed logo, a label: generation drifts on exactly these, and motion makes it worse, because the lettering has to survive every frame. Watch the whole clip before posting, frame the product so the logo is not the subject, and add your name, price and call to action as a text layer in the editor rather than trusting the model to keep them legible.

Real people need permission. If the start photo shows a customer, a member of staff or a model, you need their consent and the rights to the image. Turning it into a clip does not create a release.

Adult output is not for your shop's feed. Fellowi allows adult generation for your own use; public social platforms do not host it, however it was made. A brand account has more to lose than most.

The studio is not an archive. Clips are kept for a limited period and then deleted; the window is stated on the product page. Download what you post and keep your own copies.

No copyright warranty, no exclusivity. We claim no ownership of what you generate, but machine-made work may not be protectable at all in some countries, and someone else can get a similar result from a similar prompt. A seasonal teaser is fine on those terms. A brand mascot you intend to defend is a different question, covered in who owns AI-generated images.

A small shop's first batch

Photograph the product once, well, upright, against a plain background. Open Fellowi Video, upload the photo as the start frame, choose 9:16 and describe only what moves. Make two or three short drafts at the cheapest settings, keep the one where the label survives, add your text and music in the platform's own editor, and download the file the same day.

For mood clips that do not show the product at all, such as a snowy street for a December post, Fellowi Images can make the start frame first. The product itself stays a photograph.