Two photos. One kitchen duet. Cast the player and singer.Upload two photos and join the kitchen duet. Your first photo plays the ukulele; your second photo sings.
The kitchen duet, recast with your two people.
The format comes from Fredo Bang and Einer Bankz's acoustic kitchen performance: a ukulele player on the left and a singer on the right. AI remakes use photos to change the two performers while following the scene and gestures. It is also searched for as What You Want Me To Do AI.
Einer Bankz's original performanceThis page uses the 12-second excerpt shown in the tool, not a new song or a complete music video. The photos guide the performers' appearance; the reference guides their movement, framing and timing.
The original kitchen clip and today's photo-remake tutorials are different parts of the story.
Einer Bankz published the Oouuh acoustic performance on his YouTube channel in July 2018. The kitchen, side-by-side framing and contrasting roles make the scene easy to recognize.
Watch the original uploadSocial tutorials now show a photo-selection workflow followed by a recast kitchen duet. Those examples explain the idea; their results are not evidence of what every model or photo will produce.
View a photo-remake tutorialUse the two labeled uploads to keep the roles clear. This template requires both photos and follows the displayed excerpt rather than inventing a new performance.
No filming, instrument recording or prompt writing required.
Put the ukulele player's photo in slot 1 and the singer's photo in slot 2. Choose one clearly visible person per image.
Watch the 12-second reference, choose 480P or 720P, and check the total credits on the generate button. The role order stays the same.
When the video is ready, play it with sound. Check both identities, the instrument and the singer's mouth before downloading your MP4.
The instrument and performance come from the reference. Use your photos to show each person's appearance clearly.
The template keeps the setup focused on the kitchen duet.
Photo 1 replaces the ukulele player on the left. Photo 2 replaces the singer on the right. You do not need to describe the casting in a prompt.
The model is instructed to use each person's face, hair, body proportions and clothes. It can still introduce errors, especially around hands and overlapping objects.
The reference supplies the ukulele playing, singing gestures, camera view and timing. It is not a general text-to-video scene or a new choreography generator.
The finished clip uses the reference excerpt's audio. Uploading a photo does not clone that person's voice or create an original song.
The clip stays 12 seconds. Choose the resolution that suits your test.
480P is the default and costs 180 credits. The 720P option costs 300 credits. Both use the same two-role reference.
The generate button shows the full credit charge for your selected quality. Check your balance before starting; a new attempt is a separate generation.
Credits are the site's billing unit, not a fixed dollar amount. A higher resolution does not guarantee better identity or hand accuracy. Reference-to-video model details
Three workflows with different setup requirements. The exact CapCut template and AI model determine their available features.
Keep the duo, change the performance. Each template has its own reference and settings.
It is a photo-based remake of the kitchen duet associated with Fredo Bang and Einer Bankz's Oouuh performance. This tool replaces the two performers in a fixed reference excerpt. It does not compose a new song.
Photo 1 is the ukulele player on the left. Photo 2 is the singer on the right. Follow the role labels above the uploads; other sites may use a different order.
This version requires two separate single-person photos. It does not offer a one-person-only swap, automatic group-photo separation or a switch for one person playing both roles.
The instructions ask the model to take face, hair, body proportions and clothing from each person's photo, while keeping the reference performance. This is generative editing, not a guaranteed pixel-exact face swap, so review the whole result.
The output is a 12-second vertical 9:16 MP4. It costs 180 credits at 480P or 300 credits at 720P. The selected total is shown before generation; buying credits is separate from starting a task.
That name is commonly used for the same Oouuh kitchen-duet idea. Different tools use different cuts and upload rules. On this page, the preview shows the exact 12-second excerpt used by the template.
It includes audio from the displayed reference excerpt. Your photo is a visual reference only. This tool does not record your voice, clone it or write new music.
Not always. Faces can drift, roles can mix, instrument fingers can distort and mouth movement may not match perfectly. The player is instructed to play rather than sing, but watch both people through the whole clip before sharing.
Failed tasks return their generation credits through the site's task handling. A completed video with visual imperfections is not automatically treated as a failed task. Check History for the status and contact support if a task appears stuck.
No. This is an independent AI template, not an official artist collaboration. Use photos you are allowed to edit and follow the rights and platform rules for the source footage and music. A generated result does not grant new music or footage rights; disclose AI editing when sharing.
