Paloma Blanca AI is a 10-second video in which you are the dancer who bounds in with the bouncing kick-step behind a line of walking friends, made from one photo. You pick a format, press Generate and download an MP4 with the song already on it. This guide covers the details behind the Paloma Blanca AI trend page: how the dance is built, what the model changes, what to check before you post, and what it costs.
Key takeaways
- Paloma Blanca AI is a 10-second clip in which you replace the dancer who bounds in with a kick-step behind a line of three walking men.
- One photo of you is enough; head to knees, sharp and evenly lit works best.
- Only the one dancer is you. The three men in the line stay as in the reference clip.
- The sound is a copyrighted George Baker Selection recording: safest on TikTok, check other platforms.
- A standard video is 10 credits, and a failed render returns its credits.
In this guide
- How is the kick-step structured in the 10 seconds?
- Why does the straight-faced walk-in work so well?
- What does Seedance edit, and what does it invent?
- Can I post the Paloma Blanca AI video with the song?
- Which photo gives the best result?
- How is it different from filming the dance myself?
- Can I put my friends in the line too?
- Why not just write a text prompt in a general video model?
- What can go wrong, and what happens to my credits?
- What does one video cost in real money?
- Is it still worth posting now that the trend is fading?
How is the kick-step structured in the 10 seconds?
The clip runs in five beats. From 0 to 2.5 seconds, three men in blue stroll toward a low camera on a park path with their hands on their hips, and a fourth follows out of sight behind them. From 2.5 to 4.5 seconds, a long-haired man in a grey cap peeks out from behind the line and bobs his head to the song.
The kick-step itself starts at about 4.5 seconds, when he steps out between the men. The three men turn to watch, and from 6 seconds they break into the same bounce, knees up and arms pumping beside him. The last beat, from 8 to 9.7 seconds, is the leap toward the lens while the group dances on behind. The structure is simple: one dancer starts, the others catch on, and then everyone is moving together.
On the Paloma Blanca AI page, you replace the dancer in the grey cap, with your own face, hair and clothes from your photo. The model keeps the timing of all five beats because it copies them from a 10-second motion reference, not from a text description.
Why does the straight-faced walk-in work so well?
The joke is contrast. For the first few seconds the clip looks like four ordinary people walking down a path, and the hands on the hips make it look almost bored. Then the bounce arrives with no warning, and the gap between the calm walk and the sudden kick-step is what people laugh at.
That is also why the format spread to friends, office teams, schoolmates and dance troupes: anyone can do the walk, and the surprise is built into the order of the beats.
What does Seedance edit, and what does it invent?
Seedance, the ByteDance video model, works in a photo plus motion reference mode here. Your photo is the only source for the dancer's face, hair, build and clothes. The reference clip supplies the walk-in, the kick-step, the timing and the low fixed camera.
The prompt only states who replaces whom: the long-haired dancer in the grey cap is replaced by the person from your photo, and everything else in the reference stays the same. That is why the park path and the three men in the line are unchanged in your result.
What the model still has to invent is everything your photo does not show. A head-to-knees photo gives it a face and a torso, but the movement of your legs mid kick, the back of your head and how your clothes fold are all generated. This is an AI clip of you, not a recording of a real dance, and likeness is not guaranteed.
Can I post the Paloma Blanca AI video with the song?
The clip carries a George Baker Selection recording of "Una Paloma Blanca" as its sound. That is a copyrighted song. TikTok licenses it in its own sound library, but other platforms can mute a video, block it or file a claim against it, and commercial use is not covered.
A practical way to handle it: post on TikTok first, since the song is licensed there, and turn on the platform's AI-generated label. If you plan to post on YouTube or Instagram, check the result after upload. If a platform objects, add a sound from that platform's own library instead, or use a track you are licensed to use. The visuals do not depend on the audio, so the clip still works with a different track under it, although the kick-step is timed to the original beat.
Which photo gives the best result?
The dancer is fairly small in a wide, low shot and kicks fast, so the photo matters more here than on a close-up effect. Four tips from the page, with the reasoning behind each:
- Show your whole body. A photo from head to knees gives the model the arms and legs it needs for the kick-step.
- Wear casual summer clothes. A tee and loose trousers or shorts match the sunny park path better than a heavy coat.
- Use clear, even light. Heavy shadows and filters make the likeness drift in a small wide shot.
- Keep it to one person. No hat covering the eyes and no sunglasses.
If the first render drifts, retry with a sharper photo before changing anything else. A better input fixes more problems than a second attempt with the same file.
How is it different from filming the dance myself?
Filming it yourself gives you a real recording with real movement, and it is the better choice if you have three friends, a park and a free afternoon. You need people, a camera angle, a few takes and a decent edit.
The AI version needs only one photo. You get a 10-second clip with you as the dancer who bounds in, with no filming and no editing. The trade is that only one person in the clip is you. The three men in front stay as they are in the reference clip, so it is not a way to put a whole group of friends in.
It also is not a live filter and not a CapCut template. Those need you to film or cut footage yourself. On videosi the whole clip is generated from your photo.
Can I put my friends in the line too?
No. Only the dancer who bounds in is replaced, and a photo with several people in it is not supported. The three men in the line are part of the reference clip and stay as they are.
If you want a group version, the practical route is to make one clip per person and cut them together in an editor, or to film the real dance with your friends. The AI clip is best used as the funny cameo: you appear behind a line of strangers and bounce in.
Why not just write a text prompt in a general video model?
You can, and some people do, but the result tends to miss. Without a motion reference the model invents a generic dance, and the walk-in and the bounce never match the song. The dancer is small in a wide low shot, so the face drifts unless your photo is pinned as the only source.
Each try also costs credits and several minutes, and the song and its timing do not come from a text prompt at all. That is why the page gives the model a reference clip and a photo instead of asking you to describe a dance. The full prompt is not published, but it is short: who replaces whom, and keep everything else.
What can go wrong, and what happens to my credits?
The most common problem is likeness. The dancer is small in the frame and kicks fast, so the face and the feet can drift. A sharp, evenly lit photo from head to knees is the main fix.
A render can also fail, time out or be blocked. In every one of those cases the credits go back to your account automatically. The result is saved in My creations if you close the page, so you can come back to it.
Because this is an AI clip, it does not show the original dancers and is not a recording of a real dance. Do not upload photos of people who have not agreed, and do not use the video to mislead anyone.
What does one video cost in real money?
A standard 10-second Paloma Blanca AI video is 10 credits, and there is no free generation. videosi shows the exact price before you press Generate, and it changes with the quality tier you pick.
The Single pack is $8.90 for 10 credits, so one standard video costs exactly one Single pack. The Trio Pack is $17.99 for 30 credits, which covers about three standard videos if you want to retry with a better photo. Larger packs and subscriptions are listed on the pricing page.
Is it still worth posting now that the trend is fading?
The big wave ran from August to October 2025, and posts were still appearing in 2026. Our radar flagged the trend again on 17 November 2026, so we call it fading rather than new. We could not confirm the very first post, so the origin stays unconfirmed.
Fading does not mean gone. A song as well known as "Una Paloma Blanca" gets reused for years, and the walk-in joke works without the trend context. A clip of you bouncing in behind three strangers is funny on its own.
If you want it, the page is live, and the clip downloads as an MP4 without a watermark with the trend sound already on it.
Ready to try it? Open Paloma Blanca AI, upload one photo and your clip is ready in a few minutes. For a different dance, try the AI dance page.



