Bulin 47 AI Prompt: Copy-and-Paste Prompts for Seedance, Kling and Veo
Free Bulin 47 AI prompts for image-to-video models like Seedance, Kling and Veo, with tips to keep your face consistent — or skip prompting entirely.
Want to make the Bulin 47 AI video yourself in a general video model? Here are the prompts that get closest to the viral Madrid freestyle, and the settings that matter.
Shortcut: if you'd rather not experiment, the Bulin 47 AI generator edits the original clip for you — upload one photo and you're done.
The base prompt (image-to-video)
Use this with an image-to-video model and your own photo as the input image:
Vertical handheld smartphone footage at night on a busy city street outside a
restaurant entrance lit with red neon light. The person from the reference photo
is the lead performer: same face, hairstyle, skin tone and outfit. A street
creator stands beside them holding a phone and a portable bluetooth speaker
toward them, playing a hard dembow beat. The lead performer freestyles straight
to camera with confident rap hand gestures, bouncing shoulders and head nods,
mouth moving in rhythm, while a small crowd gathers behind, laughing, filming on
their phones and cheering. Natural street lighting, slight camera shake,
realistic skin, authentic viral street-interview energy, single continuous take.
Negative prompt (if your model supports one):
distorted face, different person, extra limbs, warped hands, text, watermark,
logo, subtitles, cartoon, blurry face
Variations
Football crew (the "World Cup" version) — add:
The crowd around them are football players in matching team kits and fans in
scarves, celebrating as if after a big win.
House party — add:
The setting is a packed house party with colorful string lights instead of the
street, friends dancing and hyping the performer up.
Recommended settings
| Setting | Value |
|---|---|
| Mode | Image-to-video |
| Aspect ratio | 9:16 |
| Duration | 10–15 seconds |
| Resolution | 720p |
| Camera | Handheld / slight shake |
Tips for keeping your face consistent
- Start from a sharp, front-facing, waist-up photo. The model can't invent details it never saw.
- Say "same face, hairstyle, skin tone and outfit" explicitly — it measurably reduces identity drift.
- Keep the scene description concrete (red neon entrance, portable speaker, crowd filming) rather than artistic.
- Generate several takes. AI video is stochastic; the second or third render is often the keeper.
Can ChatGPT make the Bulin 47 AI video?
ChatGPT can write or refine the prompt above, but the video itself has to be rendered by a video model. That means picking a model, paying for its plan, uploading the photo, tuning the prompt and retrying. A dedicated generator bundles all of that behind a single upload and refunds failed renders automatically.
Respect the people involved
Use your own photo or one of an adult who agreed. Don't impersonate Bulin 47, Kevin Wai or anyone else, and label the result as AI-generated when you post.