Back to Blog
Meme explained
5 min read

Baile Charme AI Prompt: Copy-Paste Text for Kling, Seedance, Veo and More

A ready-to-paste Baile Charme AI prompt for Kling, Seedance, Veo, Hailuo, Gemini or ChatGPT, with grandpa and duo add-ons, limits and a no-prompt option.

Baile Charme AI Prompt: Copy-Paste Text for Kling, Seedance, Veo and More

Want to make the Baile Charme video in Kling, Seedance, Veo, Hailuo, Gemini or ChatGPT? You need a prompt and one photo. Below is a prompt you can copy, what each part of it does, two add-on lines for the grandpa and duo versions, and the limits you'll run into.

What the prompt has to describe

The trend video is @seuzerj's reel from 16 October 2026: an AI grandpa, "Seu Zé de Madureira", dances charme in a packed night crowd at Pedra do Sal in Rio. A video model has never seen that reel, so the prompt has to describe it piece by piece:

  • The place. A night street party in Rio de Janeiro, at Pedra do Sal: old houses with colourful walls and blue market stalls behind the crowd.
  • The crowd and the light. A packed crowd of young people under bright white floodlights. Without this, models tend to give you an empty street or a club.
  • The moves, in order. Charme is small and smooth: side steps left and right, knees slightly bent, rolling shoulders, a cool nod. Then the glance at the dancer beside you, the cheer, the low groove and the slow turn. Listing them in order helps the model keep one continuous take.
  • The camera. A vertical 9:16 handheld phone shot from inside the crowd that eases back to a medium shot at the end.
  • Mouth closed. If you don't say it, many models make the person sing or talk.
  • No text and no celebrities. Models like to add captions and logos. And keep it to your own face, or that of an adult who agreed.

The Baile Charme AI prompt

Upload your photo as the reference image (image-to-video), then paste this. Keep it in English even if you use the app in another language; video models follow English prompts best.

Vertical 9:16 handheld phone video, one continuous shot, 15 seconds. Night street party in Rio de Janeiro (a baile charme at
Pedra do Sal): a packed crowd of young people under bright white floodlights, colourful old house walls and blue market
stalls behind. The person from the reference photo, with exactly their face, hair and outfit, stands close to the camera in
the middle of the crowd and dances charme: small smooth side steps left and right, knees slightly bent, shoulders rolling to
the beat, a cool nod, hands low in front of the body. They glance at a man in a white tank top dancing beside them, someone
throws a hand up and cheers, they groove low, then turn slowly away and back with a cool hand gesture as the camera eases back
to a medium shot. Lips closed, relaxed confident smile, no singing. Real phone-video look, slight shake, floodlight glare,
natural motion blur. No text, no captions, no logos, no celebrities.

What each part does:

  • "Vertical 9:16 … 15 seconds" sets the format. Many models cap clips at 5, 8 or 10 seconds; if yours does, change the number rather than letting it cut the turn.
  • "with exactly their face, hair and outfit" asks the model to keep you, not invent a new person.
  • "hands low in front of the body" keeps the moves small. Without it, models often turn charme into big hip-hop or funk moves.
  • "Real phone-video look, slight shake, floodlight glare" keeps it from looking like a polished music video.

Add-on lines

Paste one of these at the end of the prompt.

Grandpa version. The trend's own joke: "Ninguém botou fé no vovô no baile charme" ("Nobody believed in grandpa at the baile charme").

Make the person about 75 years old: grey hair in their own hairstyle, natural wrinkles, still clearly the same face.

Duo version. For two people, upload a photo of you both, or two photos if your model takes more than one reference.

Two people from the reference photos dance side by side, shoulder to shoulder, laughing, arm in arm, then lean in to the camera.

Honest limits of a text prompt

A prompt gets you a charme-style video. It won't get you the reel.

  • It can't copy the exact moves and camera. The model reads "side steps" and "turns slowly" and invents its own version. Timing, framing and the camera move change every try.
  • The face drifts. Image-to-video models start from your photo, but over 10 to 15 seconds of turning and moving, the face often slides toward someone else, especially in the turn away from the camera.
  • You still need the sound. Most of these models make silent video, or make up their own music. The trend runs on the reel's sound, so you add it yourself in TikTok when you post.
  • Expect several tries. Each try costs credits or time, and the crowd, light and moves come out different each time.

How Baile Charme AI is different

Baile Charme AI doesn't use a prompt you write. Each template uses a depth map of the actual reel, so the moves, timing and camera follow @seuzerj's take, and your photo takes Seu Zé's place.

  • One photo, no prompt. Upload it on the homepage, pick a model, 10 or 15 seconds and the quality.
  • The reel's sound is already in the MP4, with no watermark. A switch gives you a silent version if you'd rather add the TikTok sound yourself.
  • Versions. The grandpa version ages you to about 75. The duo version follows @zenegaooficial's duo clip from 24 September 2026 with one photo of you both, or one each.
  • Price. One video from $4.99 without an account. If your photo is rejected or a render fails, you get your money back automatically.

The full walkthrough is in how to make the Baile Charme AI video.

FAQ

Which AI video model is best for the Baile Charme prompt?

Any model that takes a reference photo (image-to-video) and makes vertical video can work: Kling, Seedance, Veo, Hailuo and others. Results vary from try to try, so expect to run the prompt a few times and keep the best one.

Can ChatGPT or Gemini make the Baile Charme video?

They can if your plan includes video generation; otherwise they can only help you edit the prompt. The same limits apply: the moves and camera won't match the reel, and you'll need to add the sound.

Should I write the prompt in my own language?

Keep the prompt in English. Video models follow English prompts most reliably, even when the rest of the app is in Portuguese, Chinese or Japanese.

Can I use a celebrity's photo?

No. Use your own photo, or one of an adult who agreed. Baile Charme AI rejects celebrity photos and photos of children.

Baile Charme AI is an independent tool, not affiliated with the trend's creators, the dancers in its clips, CapCut, or the AI model makers named here.