About the APT AI Video (Rosé & Bruno Mars Trend)
The APT AI Video Generator turns two photos into your own version of the viral APT. dance video by Rosé and Bruno Mars. It puts you and a friend into the pink-studio APT. performance: over up to 26 seconds the two performers sing, dance, drum and clown around a drum kit and a wall of amplifiers, with quick cuts between close-ups, wide shots and a floor-level finale.
Upload two photos. Image 1 takes the singer's part (Rosé's role) and Image 2 the drummer's part (Bruno Mars's role), and each stays in that role in every shot, even when the two swap sides of the frame. Your photos supply who appears: face, gender, hair, skin tone, build and outfit. The reference clip supplies everything else: the choreography, lip movements, camera framing, cuts, the pink studio, the instruments, the timing and the song.
How to make an APT AI video with Rosé and Bruno Mars's dance
- Upload Image 1 — the singer. Add a photo of the person who should take the singer's part, Rosé's role. One person, face and hair clearly visible and well lit, wearing the outfit you want in the video. It can be a man or a woman: the role keeps your gender.
- Upload Image 2 — the drummer. Add a photo of a second, different person for the drummer's part, Bruno Mars's role. Both photos are required, and the order matters: Image 1 is always the singer and Image 2 is always the drummer.
- Choose the length and resolution. 15s covers the drum kit, the hugs and the first dance; 27s is the full clip with the floor scenes and the profile finale. 480p is faster and cheaper for checking your result; 720p is sharper for posting. The credit cost for your selection is shown next to the Generate button before anything is submitted.
- Generate, preview and download. Press Generate. Your video appears in your generations with a live progress status. When it's done, watch it in the browser and download the MP4 with the original audio.
How the template works
The template uses motion transfer. First, each photo is turned into its own identity sheet showing that person from several angles. Then, instead of pasting faces onto the original performers, the AI rebuilds both performers from those sheets and drives them with the exact motion of the reference clip, shot by shot.
Each photo is bound to one role, not to a side of the frame. The two performers change places between shots, and your people follow their roles through every cut rather than swapping when the camera changes.
What the AI keeps and what it changes
- Kept from the reference: the choreography, gestures and lip movements, the camera framing and movement, the shot cuts, the pink studio, the drum kit, amplifiers and props, the lighting, the timing and the synchronised soundtrack.
- Taken from your photos: each person's face, gender, skin tone, hairstyle, facial hair, body proportions and outfit. A man in either role stays a man, in his own clothes.
- Not carried over: the reference performers' faces, hair and clothes. Only the star-shaped novelty glasses stay, as a prop.
Tips for the best result
- Use one person per photo, and two different people for the two roles. Group photos make it unclear which face to use.
- Pick sharp, evenly lit photos with the face clearly visible and not covered by sunglasses or a hat.
- Use a photo that shows the outfit you want in the video, ideally down to the waist or full body. Missing parts are completed to match.
- Start with 15s at 480p to check your result before rendering the full clip or 720p.
- Only upload photos of yourself or of people who have agreed to appear in the video.
What to expect
Results can vary during fast dancing, when a face is partly hidden (by an arm, hair, drumsticks or the novelty glasses) and right after a cut. In those frames the model can briefly drift toward the original performer or lose detail. Clear, front-facing photos reduce this, but they can't remove it completely.
Resolution and credit cost
| Duration | 480p | 720p |
|---|---|---|
| 15s | 4,157 credits | 8,785 credits |
| 27s | 7,401 credits | 15,731 credits |
Credits are charged once when the render starts and refunded automatically if the generation fails or is blocked by moderation.
Frequently asked questions
How many photos do I need?
Exactly two, one per performer. The generator won't start until both are uploaded.
Which photo replaces whom?
Image 1 is the singer, Rosé's role, who kneels at the microphone in the opening shot. Image 2 is the drummer, Bruno Mars's role, at the drum kit. Each stays in that role in every shot.
How long is the video?
You choose: 15 seconds (the opening up to the first dance) or about 26 seconds, the full reference clip. The footage is never looped or stretched.
Is the original music kept?
Yes. The synchronised soundtrack from the reference clip is preserved, and the performance and lip movements stay in time with it.
Should I choose 480p or 720p?
480p is faster and uses fewer credits, which is handy for testing your photos. 720p is sharper and better for posting. Both keep the framing of the original.
Why does a face look off in some frames?
With fast motion, partial occlusion or right after a cut, the model can briefly drift toward the original performer or lose detail. Clear, front-facing photos of one person each help. Try 480p first.
Is the video on this page an AI result?
No. The video on this page is the source reference clip the template follows, shown as a preview. It is not an AI-generated result.
Is this an official APT. template?
No. This is an independent AI template. It is not affiliated with, endorsed by or sponsored by Rosé, Bruno Mars or anyone involved in the original video.
What happens if my generation fails?
The error is shown on your generation, and the credits for that render are returned to your balance automatically.
Discover more templates in the trends gallery or explore AI video effects.
The APT AI Video Generator is an independent AI template. It has no official affiliation with, and is not endorsed or sponsored by, Rosé, Bruno Mars or the creators of the original video. The artists' names are used only to describe the source clip. Only upload images you have the right to use.