How-to
Kiss video prompts that work: a guide with sample clips
Make an AI kissing video from one or two photos with no prompt, or write your own. A prompt structure, three copyable examples, sample clips and what to do when a run fails.
Published November 6, 2026
You can make an AI kissing video from one or two photos without writing a prompt, or you can write the motion yourself for an image-to-video model. The first route takes about a minute. The second needs a structure, which most of this page lays out, plus three prompts you can copy.
Upload one couple photo or two portraits, pick a style, and generate. No prompt needed.
The one-tap route: how The Kiss works
The Kiss takes either a single photo of a couple or two separate portraits. A couple photo is animated as it is. With two portraits, a first step stages both people in one frame, and that frame becomes the opening shot of the video. Faces, hair and clothing carry over from your photos, and in most styles so does the setting of the first photo.
Styles decide how the moment plays out. Slow Burn, Hug & Kiss, French Kiss, Pocky Kiss, Neck Kiss, Wall Pin and Backseat are some of them. Five of them (Slow Burn, Hug & Kiss, French Kiss, Pocky Kiss and Neck Kiss) keep the setting of your first photo and have a scene list (Beach sunset, Flower field, Rainy street, Dinner date): pick a scene and your photo's background is replaced, and a couple photo goes through the staging step too. Wall Pin keeps your setting and adds a wall to it, and Backseat stages its own (the back seat of a parked car at dusk). Those two have no scene list.
Clips come out in portrait. You choose 5, 10 or 15 seconds and 480p or 720p. The free setting is 5 seconds at 480p; longer or sharper clips use credits, and the pricing page has the current numbers.
When you want to write the prompt yourself
The effect writes its own motion text. If you want a different moment, use image-to-video in Studio with a start photo and your own words. The photo already holds the faces, clothes and light, so the prompt only has to say what moves and when. These rules come from our hug and dance clips, which we wrote and tested the same way, and we apply them to kisses.
- Start from the moment before. Put the two people close together and looking at each other, not yet touching. If the first frame already shows the kiss, the clip has nowhere to go. In our hug runs, a start frame that already showed the embrace left the model with only "keep embracing".
- Split the clip by seconds and give each segment one main action. A single paragraph of prose gets executed from the front and the rest is dropped. Lines like 0-2s, 2-4s and 4-6s keep every beat in play.
- Use physical verbs. Models turn intensity down, so "leans in" tends to come out as a nudge. "Tilts her chin up" and "lowers his head slowly" give the model something to draw.
- Do not describe the photo. Clothes, location and lighting are already in the first frame, and repeating them spends prompt space that the motion needs.
- Lock the camera, then push in once. A still camera holds a pause, and one slow push-in on the last beat marks the payoff.
- Keep hands simple. In two of our clips, fine hand work (a zip pull, a door handle) rendered as vague reaching. Large movements read better, and tension can live in a pause rather than a gesture.
- End on a small gesture that shows how it landed, such as a forehead against a jaw. If your model generates sound with the video, finish with one line about it.
Three prompts to copy
These are ours, written for a start photo of two adults standing close and facing each other. The Slow Burn clip above came from the effect, not from these exact words, so treat them as starting points and expect to adjust.
[0-2s] They stand a breath apart, looking at each other. Neither moves. Camera locked, completely still. [2-4s] She tilts her chin up. He lowers his head slowly. [4-6s] Their foreheads touch, her eyes close, and they share a short, gentle kiss. Camera pushes in slowly. Natural continuous motion. No talking. AUDIO: room tone, one soft piano note.
[0-3s] She says something quiet and he laughs, shoulders shaking. She grins and looks down. Camera locked. [3-6s] She looks up. He steps in until they are almost touching. [6-10s] They lean in together and kiss briefly, then pull back, still smiling, foreheads touching. Camera pushes in slowly. No talking. AUDIO: a soft laugh, light acoustic guitar.
[0-2s] He stands still, looking down at her. She tilts her head up. [2-5s] He bends and presses a kiss to her forehead. Her eyes close and her shoulders drop. He rests his chin on top of her head. Camera locked. No talking. AUDIO: low strings, one slow breath.
What goes wrong
If the staging step turns a pair of photos down, the card shows "The image engine turned this pair down. Try different photos. Your credits are back." A retry would send the same pair to the same image engine, so swap in different photos instead.
Results will not match our samples exactly. The sample clips on our pages come from our own cast portraits at 720p, with the best take kept, while the free setting is 5 seconds at 480p.
For the best start, pick photos where both faces are visible and the light is even. We have not measured how much that changes the success rate, so take it as a habit rather than a rule.
Questions
Do I need a prompt for The Kiss effect?
Can I use one photo instead of two?
How long are the clips, and what does it cost?
Will it keep my clothes and background?
Why does my clip look different from the samples?
Try The Kiss with your own photos, or go to Studio if you would rather write the motion.