Quick answer
Upload one or two photos to Starrd's What You Want Me To Do template and it recreates the viral 2018 kitchen duet behind Fredo Bang's 'Oouuh': a ukulele player strumming on the left and a singer on the right singing the "I know that's what you want me to do" hook twice, move for move from the original phone video — the chin-hand sway, the grin, the head-back "ooh, ooh, ooh", the palms-down reach and the two-hand reach to the camera — with the original audio. Two photos: photo 1 sings, photo 2 plays. One photo: you play both parts. Pets work too, standing up with the ukulele in their paws. 150 credits (about $7.50).
One or two photos in, and out comes the viral kitchen duet: one of you on ukulele, the other singing Fredo Bang's "Oouuh" hook, head back on every "ooh". Here's one made from a single photo of a golden retriever playing both parts:
What You Want Me To Do
One or two photos. One of you on ukulele, one singing Fredo Bang's 'Oouuh' hook in the kitchen. 23 seconds, vertical.
What You Get
- The kitchen. A small apartment kitchen at night, the white door behind you, filmed straight on like a phone video.
- The ukulele player, on the left, strumming the whole way through, head down, glancing up at the camera with a smile.
- The singer, on the right, singing the hook twice: the chin-hand sway, the grin, the head-back "ooh, ooh, ooh", the palms-down reach, a hand up by the head and both hands reaching toward the camera.
- The original audio, the live ukulele and vocal.
- 23 seconds, vertical 9:16, one take, ready for TikTok, Reels and Shorts.
What the Trend Actually Is
In 2018, Fredo Bang posted a freestyle with Einer Bankz, the ukulele player known for viral videos of rappers singing a cappella over his playing. The two stood in a kitchen, Einer strummed, and Fredo sang a hook: "Ooh, I know that's what you want me to do." WorldStar shared it, it went viral, and fans pushed him to finish it. It became "Oouuh", and the clip still circulates as an all-time acoustic moment.
Why it suits AI: it's a two-person format, and the funniest version is the unlikeliest duo. You and your dad. You and your dog. Two of your dog.
One Photo or Two
- Two photos: photo 1 is the singer, photo 2 plays the ukulele. Grab a friend, a partner, a sibling.
- One photo: the same subject plays both parts, side by side, with the same face and outfit. It works best on pets.
Pets Welcome
- On two legs. Both stand upright in the kitchen like people.
- Ukulele in paw. The player holds and strums it with real paws, never human hands.
- Singing with its own mouth, real teeth and tongue, head back on the high notes.
Use clear, well-lit photos where the faces are sharp. Whatever you're wearing comes with you; the kitchen, the ukulele and the performance come from the template.
The Fastest Way
What You Want Me To Do
One or two photos, the full 23-second kitchen duet with the original audio. People and pets.
One or two photos, 150 credits (about $7.50), no prompt. The template holds the kitchen, the ukulele, both performances and the audio; your photos only decide who's in the duo. It renders on Seedance 2.5 at 480p, vertical 9:16.
Build It Yourself
1. Turn the clip into a depth map. Cut the first 23 seconds (the hook twice), strip the audio, and run it through a depth-estimation model (we used Depth Anything V2). You get a greyscale video of both people's shapes and motion only, so neither original performer can leak into your video. The original ukulele player's cap still shows as a shape, so say "no cap unless it's in the photo".
2. Make one still of the duo in the kitchen. Put each subject on their side and name their role.
A photorealistic vertical phone-video still of two people side by side in a small apartment kitchen at night, filmed straight on from a few steps back. On the left, the subject from reference photo 2, holding a small wooden ukulele and strumming it, head tilted down. On the right, the subject from reference photo 1, singing with their eyes half closed, one hand raised at chest height. Cream walls, white cabinets, a closed white door behind them, warm overhead light. Keep both exact faces and their own clothes. No text, nobody else in the room.
3. Write the performance beat by beat. Watch the original at two frames a second and write both people's moves with times and expressions. Spell out that the ukulele player never sings and the singer's mouth only moves on the lyrics. Without this, the model gives you two people swaying.
4. Drive it with the depth map. Feed the still, both photos, the depth video and the beat list to a model that copies motion from a reference video (we recommend Seedance 2.5). Say the still decides everything you can see and the depth video decides only movement and timing, in full colour, never grey. Lay the audio back on at the end.
Common Mistakes
Using the raw clip. The original performers' faces come through. A depth map carries the duet without them.
Not naming who sings. Without it, both people mouth the words. Say the ukulele player's mouth stays closed.
Letting the cap leak in. The depth map keeps the original player's cap silhouette. Rule it out in the prompt.
FAQ
How many photos? One or two.
Does it work with my pet? Yes, as the singer, the ukulele player, or both.
How long is it? 23 seconds, vertical 9:16.
What does it cost? 150 credits, about $7.50.
Related Reading
- How to Make an AI DJ Video: headline the festival main stage
- How to Make an AI K-Pop Jumbotron Video: you on the concert screen
- How to Make the Drake 'Burning Bridges' AI Video: another one-take performance from one photo
- How to Make Your Pet Dance: pets on the dance floor
