AI kiss generator
Upload two photos and get a short clip of the two people leaning in for a kiss. The prompt below is already filled in and the frame is set to vertical 9:16 for TikTok and Reels. Swap in your two photos and render a five-second take. Use photos of yourself and of people who agreed to this.
MiniMax H3 is available on a paid plan. See plans
Made with this model
Every clip below was rendered here, on the settings shown under it. The caption is the exact prompt that produced it.
MiniMax H3 · 2K · 9:16 · 5s
Make one in three steps
Upload two single-person photos
One person per photo, face clearly visible, roughly facing the camera, decent light. The prompt calls them @Image1 and @Image2, so the order only matters if you edit the prompt to say who stands where.
Keep the prompt, adjust the mood
The prefilled prompt sets the framing, the slow lean-in and a brief kiss. Change the setting or the light (a rainy street, a beach at sunset) and leave the rest alone until you have a draft that works.
Start with five seconds
Every clip renders at 2K, so there is no cheap draft tier. A five-second take is 130 credits and is enough for this format. Check that both faces still look like the photos before you try a longer one.
Specs at a glance
| Model | Duration | Resolution | Native audio | Reference images | From |
|---|---|---|---|---|---|
| MiniMax H3 | 5–15 seconds | 2K | No | Up to 9 | 130 credits for 5 seconds |
Two photos or one
Most kiss generators ask for two separate photos, and so does this page. Reference mode treats each photo as an identity to keep, then builds a new scene around both people. That is what lets you put two people who were never photographed together in the same shot.
If you already have a photo of the two of you together, image to video is the better tool: it animates that exact frame, so the pose, the background and the light stay as they are. You only have to describe the lean-in.
Either way, close-up portraits work better than full-body shots. The model has more of each face to hold on to, and faces are what people notice first.
Animate one photo with image to videoWhat a kiss clip costs here, and why this model
MiniMax H3 bills a flat 26 credits a second at 2K, whatever the frame shape. Five seconds is 130 credits, ten seconds is 260. The first five reference images are included, so two photos cost nothing extra.
In currency our credits cost roughly 3.3 to 4.6 cents each depending on the plan or pack, so a five-second kiss is about $4.30 to $6. Figures checked against the model catalogue on 30 October 2026.
We tried the cheaper Seedance 2.0 reference mode first. Its provider rejected our two reference portraits as possible likenesses of real people, even though both were AI-generated. H3 accepted the same two photos, and the sample above is that render: the first take, not a pick from several.
See credit packs and plansConsent, celebrities and what moderation does not check
Our moderation screens prompts, uploads and results for NSFW content. It does not know who is in your photos. That part is your responsibility.
Use photos of yourself, and of a partner or friend who has said yes to this. Do not use celebrities, public figures, exes or anyone under 18, and do not post a clip of someone who has not seen and approved it. In many places making or sharing intimate content of a real person without consent is a crime and it breaks our terms of service.
Fictional characters you drew or generated yourself are fine.
How reference to video worksWhat to check in the draft
Faces drifting. One face slowly turns into the other, or neither looks like its photo by the end.
The kiss never lands. The two lean in and freeze a few centimetres apart. Make the action explicit in the prompt rather than raising the resolution.
Merged noses and chins at the moment of contact. A slightly turned head angle in the prompt usually helps.
Clothing or hair swapping between the two. Photos with clearly different outfits make this less likely.