Our sample, made from two photos: the son raps, the dad holds the speaker. Both people are AI-generated.
In the Bulin 47 video a man in a pink shirt gets stopped at a doorway by a friend with a speaker, raps back, throws an arm up and dances with his shoulders. The AI version keeps all of that and changes the man. This is how to make yours with Bulin 47 AI: which photo plays which part, what the five beats ask of your photo, and what we learned from the renders we ran before putting it live.
If you want the story first (who Bulin 47 and Kevin Wai are, the dates, the footballer and Squidward versions), it is on the Bulin 47 video page. This post is only about making one.
Two parts, two photos
The original has two people on screen, and that is the thing most Bulin 47 tools skip: they swap the rapper and leave a stranger holding the speaker. Here each photo is a part.
- Photo 1 is the rapper. The person who takes Bulin's place: pink shirt, black shades, the freestyle, the laugh, the shoulders. They are in close-up for most of the 15 seconds.
- Photo 2 is the speaker friend, and it is optional. That person stands facing the rapper with the speaker held up, seen mostly from behind and over the shoulder, like Kevin in the clip. Leave it empty and a stranger holds the speaker.
The family version above is why the second photo exists. Kid on the mic, parent on the speaker; or a couple; or your dog on the mic and you holding the speaker. Pets work in either part and stay real animals, fur and all, in the shirt and shades.
What each beat needs from your photo
| Beat | What happens | What it needs |
|---|---|---|
| 0–2 s | The camera pushes through the crowd at the red door and finds the rapper | A face the AI can recognise from a few metres: not a tiny face in a wide photo |
| 2–4.5 s | The speaker comes up, the rapper turns to it and starts | Photo 2, if you use one: a clear face, since it is built from the front even though you mostly see the back |
| 4.5–9 s | Close-up over the speaker: fast rapping, hands chopping the beat | Photo 1 front-facing with the mouth visible, so the lip-sync has a mouth to work with |
| 9–11 s | Arm in the air, big laugh | Nothing extra; this is where a smiling photo pays off |
| 11–15 s | “Dame hombro”: leaning in, shoulders popping on the beat | The shoulders come from the scene, not the photo; a head-and-shoulders crop is enough |
Step 1: the rapper's photo
- No sunglasses in the photo. The video puts Bulin's black shades on the rapper. If your photo already has sunglasses, the AI has no eyes to build the face from and the result drifts. Take them off for the photo; they come back in the video.
- Front-facing, well lit, one person. The scene is dark with red light, so a photo taken in daylight or under a lamp keeps the face sharp in it.
- Head and shoulders is enough. The pink shirt is added for you. What you wear in the photo doesn't show.
- A grin helps. Beat four is a laugh. A neutral passport photo works, but a photo where the person is already smiling comes out more like them.
Step 2: the speaker friend (optional)
Same rules, and one more: a recognisable hairstyle or hat. Photo 2 is seen from behind for most of its screen time, so the back of the head carries the likeness. The dad in our sample is known by his haircut and his jacket as much as by his face. This person keeps their own outfit from the photo.
If the only photo you have of that person is dark or blurry, skip it. One good rapper with a stranger on the speaker beats two faces where one is wrong.
Step 3: make it
Open Bulin 47 AI, where the effect is already selected, add photo 1 (and photo 2 if you have one) and press Make. Standard 720p is 10 credits, 480p is 7. The 11 Bulin 47 videos made on the site between launch and 16 October 2026 all rendered at the first attempt, with a median wait of about four minutes; a render that fails is never charged.
Step 4: post it
Download the MP4. It is 15 seconds, vertical 9:16, with the original sound and the rapper's mouth on the words, so it goes straight to TikTok, Reels or Shorts with nothing to add. Captions people use:
- “Dame hombro 🔊”
- “When the speaker guy picks the wrong person”
- “Dad said he could freestyle. Dad could not.”
- “My dog has more flow than me”
Three of ours
Why it is 15 seconds and not the whole clip
Kevin's post runs 46 seconds, but the part everyone remakes starts where the camera swings from Kevin to Bulin and ends on the shoulder dance. That is the window this effect uses, with the matching 15 seconds of the original sound, so the finished video sounds like the versions already on your feed rather than a re-recorded beat.
How much it costs
10 credits at 720p, 7 at 480p; a $9.99 Single pack covers one video. Packs never expire. Details on pricing.
Good to know
This is a fan trend. RapVideo AI is independent and not affiliated with Bulin 47, Kevin Wai or their label. Use photos of yourself and people who agreed to be in the video; anyone under 18 needs a parent or guardian's consent, and no footballers, politicians or other celebrities.
FAQ
Can I make it with one photo?
Yes. One photo makes the rapper; the speaker friend is a stranger. The second photo is only for putting someone you know on the speaker.
Does it have the original sound?
Yes. The 15 seconds of the freestyle that match the shots are on the video, and the rapper's mouth follows them.
Can my kid or my dog be the rapper?
Yes to both. A child needs a parent's consent and a clear photo like anyone else. A pet can take either part and stays a real animal in the pink shirt and shades.
Make yours
Take one clear photo without sunglasses and open Bulin 47 AI. If you'd rather put yourself in a different one, the same one-tap idea works for Burning Bridges AI and STORM II AI.