AI Kiss App Features: An Honest Tour

Feature lists are usually written to make a product sound larger than it is. This one is written the other way round: here is everything the app actually contains, why each part is there, and the things it deliberately does not do.

The short version is that it is a focused tool. Photos in, a five-second clip out, one payment per video. Everything below is detail on that.

Two ways to give it photos

The app accepts your images in one of two modes, and the choice changes the quality of the result more than most people expect.

Two photos is the mode most people start with. You add one image per person. Before anything is generated, those two images are composed into a single frame: one canvas, each photo fitted into its half. That single frame is what the model actually receives, which is why the guidance is one person per photo, and why lighting that matches between the two makes such a difference.

One photo with two people takes a single picture that already contains both faces. If you have a good photo of the two of you together, use it. The lighting, the camera, the background and the perspective already agree with each other, so the model has far less to reconcile and the output tends to hold together better.

Accepted formats are JPG, PNG and WebP. Images are resized in your own browser before upload, so a large photo from a modern phone is fine.

Three styles

The same pipeline produces different motion depending on the instruction that goes with your image.

Kiss is the default: a lean in, contact, a short hold. It is the most forgiving of imperfect source photos, and the right pick when you are not sure.

French kiss asks for something more intimate and a longer moment of contact. It is a heavier request, so it rewards clear, front-facing portraits with plenty of face in frame. There is a page on what that style changes.

Hug moves shoulders and arms rather than only heads. It is the style behind reunion clips, family posts and the younger-self trend, and it suits any pair of people where a romantic framing would be wrong. See the hug video page.

Trying a different style on the same two photos is often a better second attempt than swapping the photos.

Two aspect ratios

You choose 4:3 or 9:16 before generating, and the finished video inherits that shape.

Pick 9:16 for Reels, TikTok and Stories, where it fills a phone screen with no bars. Pick 4:3 when the clip is going next to photographs, into a wider edit, or anywhere that a more classic frame suits better.

This choice is made at the start rather than at the end because it decides the shape of the frame the model starts from. There is no lossless way to change it afterwards, so it is worth thirty seconds of thought about where the video is going.

Your gallery

Every job you start appears in a gallery view with its current status: pending while it renders, then finished or failed. You can leave the page and come back: the work continues on our side rather than in your tab.

Sorting is newest first, so a job you started a minute ago is at the top. A pending item refreshes on its own while you wait, which is why leaving the page open is optional rather than necessary.

Finished items open into a player with a download button, and downloading is the step worth doing immediately. Because the app uses an anonymous browser session rather than an account, the gallery belongs to that browser on that device. Clearing site data or moving to another phone means starting fresh. The video file on your device is the permanent copy.

Paying for a video

One payment, one video, charged once, with no subscription and nothing to cancel later. Apple Pay, Google Pay and cards are all accepted, and the current figure is on the pricing page.

There is no account to create and no email address to hand over before you can pay. The full breakdown is on the pricing page, and the reasoning behind the no-login approach is on the no-login generator page.

Because a credit is bought and spent in one go, there is no balance quietly draining in the background and no renewal date to keep track of. If you make one video this year and never come back, that is the whole transaction.

Retry, and the automatic refund

Generative models fail sometimes. When a job fails here, your credit is returned to your balance automatically and a retry option appears on that item.

This matters more than it sounds. It means a failed run costs you a minute rather than a payment, and you do not have to write to anyone to sort it out. It also means the sensible response to a failure is to look at your source photos, fix the obvious problem, and go again, the photo requirements guide covers what to look for.

To be precise about the boundary: the automatic credit applies to a generation that fails. A clip that renders successfully but is not what you had in mind is a finished video, and the fix there is different photos or a different style.

Promo codes and the invite link

There is a promo code field alongside payment. If you have been given a code, it applies before you pay.

There is also an invite link in settings, unique to your session, which you can copy or share so a friend arrives at the app through you.

Neither of these is a growth scheme you have to opt into. They are two small fields you can ignore entirely if they are not relevant to you.

Works on any device

The app is a web page. It runs in Safari on an iPhone, in Chrome on Android, and in any current desktop browser on Windows or Mac.

Nothing to install, nothing taking up storage, no update to apply. The practical consequence is that you can shoot a photo on your phone and make the clip on the same device without moving files around, which is most of the reason the browser approach is the right one for a tool used a handful of times. There is more on that in the app overview.

What the app does not do

This is the part most feature pages leave out.

• No editing suite. No trimming, no transitions, no text overlays, no colour grading. It generates the clip and hands you an MP4.
• No music library. Add sound in the app you are posting to, which keeps you inside that platform's audio catalogue.
• No filters or effects. The three styles are the controls. There is no separate look to apply on top.
• No text-to-video. It animates photographs you provide. It cannot invent people from a written description.
• No long clips. Around five seconds, because that is where current image-to-video models hold a face together reliably.
• No cross-device history. Anonymous sessions do not sync. Download what you want to keep.
• No batch mode. One video per job.
• No control over the exact motion. You choose a style, not a camera path or a timing curve. The model decides the rest, which is why two runs from the same photos are never quite identical.

If that list reads as short, that is the intent. The app does one thing, and walking through it takes about five minutes. You can open it at the create screen and see the whole interface at once.