One still · Locked camera · Motion that never resolves
An ordinary AI clip has a beginning and an end, which is exactly what a wallpaper cannot have. This runs the other kind: locked camera, ambient motion, and an ending that lands back where it started.
Generated on this page's flow, from one vertical image and one paragraph of instruction. Every output is AI-generated, so yours will vary with the still and the prompt — treat this as the shape of the result rather than a fixed one.

Chosen deliberately, because it is the easiest class of subject to loop. Everything in the frame that could move — the drops, the shimmer on the lights — is motion that repeats in reality, so the model does not have to invent an arc. There is no person in it and nothing at the edges waiting to enter, which removes the two commonest ways a loop breaks.
Start from a photo like thisDrops run down and merge, the bokeh behind pulses gently, and the framing does not shift by a pixel. Watch it twice and try to catch the seam — the point of this demo is not that it is beautiful but that it is boring in exactly the right way. That is the property a looping wallpaper needs and an ordinary clip never has.
See it loopSame models, opposite instructions. Most of the craft is in what you tell it not to do.
Image-to-video is tuned to be interesting. Given a still it will invent a camera move, start an action, build to something — all the things that make a five-second clip worth watching once. A phone wallpaper is watched a hundred times a day out of the corner of an eye, and every one of those instincts becomes an irritation at that frequency.
So the prompt has to actively suppress them: no camera movement, no entrances or exits, no action with a start and a finish, and a composition at the end that matches the composition at the beginning.
The pair above is our own run. The still is a rain-covered window with a city thrown out of focus behind it, and the clip does three things and nothing else: drops run and merge on the glass, the bokeh behind them shimmers slightly, and the frame stays exactly where it was. Compare the first frame against the last — they are close enough that the repeat is hard to catch, which is the only quality test that matters for this.
Subjects like rain, water, fog and out-of-focus light are the easiest wins because their motion is genuinely cyclical in real life.
Two practical notes. What comes back is an MP4, which is what Android wallpaper apps and desktop tools like Lively want directly; iOS needs one more step, because a lock screen there takes a Live Photo rather than a video, and a free converter app does that in a few taps. And a loop is short by design — a few seconds of ambient motion, not a scene.
If you want the photo to actually do something instead, bring photos to life is the page for a clip with a beginning and an end.
Creative engine
Start from a vertical still, ask for ambient motion with a locked camera, then check the first and last frames.
Water, weather, fire, crowds at distance, city lights. Anything whose real-world motion repeats will loop without you having to force it.
In those words. It is the single instruction that separates a wallpaper from a clip, and models will add a drift if you leave the door open.
Each of these is the opposite of what you would ask for in a normal clip.
Shoot or generate the still at 9:16. Cropping a wide frame afterwards throws away the composition, and a lock screen crops hard around the clock and the notifications.

"The camera is completely locked off and never moves or zooms." Say it in the prompt even though it feels redundant. It is the instruction most likely to be quietly ignored if you only imply it.
Nothing walks in, nothing leaves, no person appears. An object that enters the frame has to leave it before the loop restarts, and it never quite does.
Scrub to the start, then to the end, and compare. If the composition drifted, rerun rather than editing — a wallpaper that jumps every five seconds is the one thing users notice immediately.
Phones are the obvious one, and they are not the biggest one.
The original reason anyone searches for this. A loop built from a photo you took means a moving wallpaper nobody else has, instead of the same handful of files that ship with the theme apps.
Animated desktop tools take an MP4 straight. The requirement is the same as a phone and the failure is worse — a background that resolves something every few seconds behind your work is genuinely hard to ignore.
A looping plate behind a talking head, on a shop screen, or on a stand at an event. Same discipline, larger canvas, and it pairs with a background plate built for the same eye level.
A still with a little ambient motion outperforms a flat image in a feed without ever looking like an ad. Vertical already, and short enough to autoplay twice before anyone scrolls.
The practical questions between a generated loop and a working lock screen.
iOS wants a Live Photo rather than a video, so there is one conversion step in between. Free apps in the App Store do it — you import the MP4, they output a Live Photo, and you set it from the photo library with Live Photo enabled. Android is more direct: a video live wallpaper app takes the file as it is.
Short. A few seconds is right, because the shorter the clip the less chance the composition has to drift away from where it started. Long loops sound better in theory and are much harder to make seamless in practice.
It loops well rather than mathematically. The prompt drives the end back toward the start, and on ambient subjects the join is usually invisible; on anything with a clear moving object it is not. If the seam bothers you, choosing a more cyclical subject helps far more than another rerun.
Some, on any phone — an animated lock screen is doing work a static one is not. In practice most people find a short quiet loop unnoticeable next to normal screen-on time, but it is a real cost and worth knowing before you blame something else.
Rain, water, fog, fire, snow, city lights out of focus, slow crowds at distance. The common factor is motion that repeats in the real world. Portraits are the hardest case, because a face that blinks or turns creates exactly the beginning and end a loop cannot absorb.
Yes, that is the main path — upload the still and describe the ambient motion you want in it. Photos you upload travel over HTTPS, land in your own history and go nowhere else. Every output is AI-generated.
One credit pool covers Nano Banana images and Seedance and Veo video. The cost shows before every run, and failed jobs are not charged. Use a subscription for ongoing work, or a one-time pack when you just need to top up.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
One-time top-ups — buy extra credits any time you run low.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
Charges appear as “KAVEL AI” on your card statement. You can cancel any time from Settings → Billing; cancellation takes effect at the end of the period you already paid for. Operator details, the full model list, and the refund window are on the about page.
Powered by
Use one balance across every supported image and video model — all of them listed below — and check the credit cost before each request.
Video models
Image models
Model credit guide
You're only charged for successful generations. The exact estimate in the generator varies by model, length, resolution, audio, and number of images.
| Type | Model | Credit cost |
|---|---|---|
| Video | Seedance 2.5 | The long-take tier, priced per second: 5s 480p ≈ 211 credits, 5s 720p ≈ 473. A full 30s take runs ≈ 1,260 at 480p and ≈ 2,835 at 720p. Supplying a reference clip lowers the per-second rate. |
| Video | Seedance 2.0 | 5s 720p image-to-video ≈ 188 credits; text-to-video ≈ 308 credits. Scales with resolution and length. |
| Video | Seedance 2 Fast | Faster and lower cost. 5s 720p text-to-video ≈ 248 credits. |
| Video | Seedance 2 Mini | The cheapest Seedance tier. 5s 720p ≈ 154 credits, 5s 480p ≈ 72 credits. |
| Video | Seedance 1.5 Pro | Audio doubles the rate. 5s 720p ≈ 27 credits silent, ≈ 53 with audio; 1080p ≈ 57 and ≈ 113. |
| Video | Veo 3.1 | Billed per video, not per second (Lite tier). About 45 credits at 720p and 53 at 1080p. |
| Video | Kling 3.0 | Audio raises the rate. 5s 720p ≈ 105 credits silent, ≈ 150 with audio; 1080p ≈ 135 and ≈ 203. |
| Video | MiniMax H3 | Fixed 2K, no resolution ladder. Priced per second — a 5s clip ≈ 158 credits. |
| Image | Nano Banana 2 | Generate or edit from text and images. 20 credits per 1K image, 30 at 2K, 45 at 4K. |
| Image | Nano Banana Pro | Consistent run times across generations. 30 credits per 1K or 2K image, 50 at 4K. |
| Image | Nano Banana 2 Lite | Faster, simpler variant, and the everyday editing price: a flat 8 credits per image at every resolution. |
| Image | GPT Image 2 | The lowest-cost premium image model here. 10 credits per 1K image, 15 at 2K, 30 at 4K. |
| Image | Seedream 5.0 Lite | Flat pricing — 20 credits per image whether you generate or edit. |
Start from a vertical still, keep the camera still, and check the last frame against the first — or make a clip that actually goes somewhere.