The APT. trend puts two new people into the pink room from ROSÉ and Bruno Mars' "APT." music video: one on the drums, one on the mic. It started on October 1, 2026, with a split-screen post on X comparing the official video to an AI remake, and within days people were swapping in their friends, partners and coworkers.
This guide shows how to make your own with the APT AI video generator: who should drum and who should sing, which photos hold up in the lip close-ups, and the one kind of photo that gets a render blocked.
The original and ours
The split-screen AI remake that started the trend.
Our landscape version with two AI-generated people.
Ours follows the first 15 seconds of the official video, beat by beat, with the matching 15 seconds of the song on the file. The four beats, and who carries each one:
| Time | What happens | Who carries it |
|---|---|---|
| 0–3s | The singer kneels by the kit and starts the song into the camera | Singer |
| 3–10s | Both behind the kit, cheek to cheek, the drummer playing | Both, faces side by side |
| 10–13s | Tight close-ups on lips mouthing the words | Mostly the singer |
| 13–15s | A little dance side by side | Both |
Step 1: Decide who drums and who sings
- Photo 1 is the drummer. This person sits behind the kit.
- Photo 2 is the singer. This person opens the video alone and owns the lip close-ups.
The singer is on screen first and closest, so the stronger photo goes in slot 2. If one of you has a photo with a clear, relaxed mouth and the other is grinning with a hand near the face, the first one sings.
One photo of both? Choose One photo of both. Whoever stands on the left plays the drums, and whoever stands on the right sings.
The two photos don't need to be taken together. A selfie from today and a photo of your friend from last year work fine. The AI puts both of you in the same pink room.
Step 2: Pick photos that survive the close-ups
The general rules (clear, front-facing, even light) are in the main photo guide. For APT. specifically:
- The mouth has to be visible. No hand, mic, cup or mask near it. The lip close-ups are generated from whatever the photo shows of the mouth, and a hidden mouth has to be guessed.
- Cheek to cheek works best with similar framing. The 3–10s beat puts your faces side by side. Two photos taken at similar distances look more natural together than a close selfie next to a far-away group crop.
- Use real photos of you, not stills from the video. See below.
The photo that gets a render blocked
Two early renders on our site were stopped before they started with a copyright flag on the input image. A screenshot from the "APT." music video, or a photo of ROSÉ or Bruno Mars, can be read as copyrighted material and the video model refuses it. Use photos of yourself and your friends. Blocked renders are not charged, but you lose the few minutes it took to find out.
Very wide images can fail too. A panorama or a screenshot of a wide split-screen is outside the shape the model accepts. Crop to a normal portrait or square photo first.
Step 3: Pick landscape or vertical
The generator opens in landscape 16:9, the shape of the original video. That suits YouTube, X and the side-by-side comparisons the trend started with.
If it's only for TikTok or Reels, switch the format to vertical 9:16 before you render:
The same effect rendered vertical.
Step 4: Make it and post it
Open the APT AI video page, add your photos, keep the APT. Pink Studio effect selected and tap Make the effect. It takes a few minutes. The song is already on the MP4, so there is nothing to add in an editor.
Want the split-screen look that started the trend? Put the official video on one side and yours on the other in CapCut or any editor. Keep the official side short, since platforms can mute or remove videos with long stretches of copyrighted music.
Is there a free APT AI video?
Not here, and we'd rather say so up front. A 720p Standard video is 10 credits, the same as one song, so a $9.99 Single covers one APT. video. 480p is 7 credits. Failed and blocked renders are never charged, and packs never expire. See pricing.
Before you post
Everyone whose face is in the video should be fine with it, even if it's a joke, and anyone under 18 needs a parent or guardian's consent. The song is the original recording, so you are responsible for how you share it. RapVideo AI is independent and not affiliated with ROSÉ, Bruno Mars or their labels.
FAQ
Can I do it alone?
The APT. effect is built for two people, a drummer and a singer. Ask a friend, or use one photo of both of you.
Can the two photos have different backgrounds?
Yes. The AI places both people in the pink room, so the original backgrounds don't matter.
Why is it 15 seconds and not the full song?
It follows the opening 15 seconds of the official video, the part with the drums, the cheek-to-cheek shot, the lip close-ups and the side-by-side dance.
Why was my render blocked?
Almost always the photo: a still from the music video, a photo of the artists, or a very wide image. Swap it for your own portrait-shaped photo and try again. Blocked renders cost nothing.
Make yours
Pick two photos, put the clearer mouth on vocals and try it on the APT AI video page. Curious where the song's name comes from? Read APT. song meaning.