AI couple video styles, from a kiss to a slow dance

These are the moments people usually mean when they search for an AI kiss or couple video. The second column is what the clip shows, as it is described to the video model, so it is a fair picture of what you will get rather than a sales line.

Not every pair is a couple. These styles are for friends, siblings, a parent and a grown-up child, or the hug your younger self trend, where nothing romantic should happen at all.

The last chip in the row is Custom. It opens a text box where you describe the action in your own words, between 3 and 200 characters, and that description replaces the built-in one. The two people, the frame, the length and the price stay exactly as they are for every other style.

What works is one continuous action that two people can do in a few seconds: "they clink champagne glasses and laugh", "she whispers something in his ear and he smiles". What does not work is a story with several scenes, a change of location, extra people, text on screen, or anything explicit, which the Terms rule out anyway. Write it the way you would describe a single shot to a camera operator.

A rough guide, from what the styles actually show. None of it is a rule: the photos are the same, so trying a second style costs one more generation and nothing else.

The style changes only the action. Everything else is shared: the same upload (two photos, or one with both people), the same frame shapes (4:3 and 9:16), the same lengths (5, 10 or 15 seconds), the same 768p MP4 with a sound track the model generates itself, and the same credit cost for a given length whichever style you pick.

Failures are handled the same way too. If a generation fails, the credits it used go back to your balance automatically, so a style that does not suit a particular pair of photos costs you a retry, not a purchase. The photo requirements guide explains the usual reasons a clip goes wrong.

Hands are the hardest part of any AI video, so the styles built around hands (Hand kiss, Hold hands, High five, Fist bump, Proposal) are the ones most likely to show an odd finger or a ring box that changes shape. A photo with both arms visible helps; a tight face crop makes it worse.

Styles with big movement, like Slow dance and Laugh together, can drift further from the original faces than a still, close style like Forehead kiss, because more of the frame is invented. If likeness matters most, start with a close style, and read why an AI video may not look like you.