Quick answer
Starrd's Gangsta's Paradise template turns two photos into the viral AI choir parody of Coolio's 'Gangsta's Paradise'. Photo 1 raps the end of the verse straight into the lens, lip-synced on the beat, then the robed choir behind you sings the chorus and photo 2 steps in with the mic to sing 'gangsta's paradise', in a candle-lit cathedral under stained-glass windows. Director's Notes can move it anywhere, and the choir comes along. You each keep your own face, hair and outfit, and it works with pets. 15 seconds, vertical 9:16, Seedance 2.5, 150 credits (about $7.50). Gangsta's Paradise (Extended) is the full 30 seconds, from 'Power and the money' to the hook, 250 credits (about $12.50).
Two photos in, and out comes you and a friend performing "Gangsta's Paradise" in a candle-lit cathedral, with a robed choir behind you: one of you raps the verse into the lens, the choir takes the chorus, and the other steps in with the mic to sing the hook. Every line lip-synced, every cut on the beat. Here it is:
Gangsta's Paradise
Two photos. One of you raps the verse, the choir sings the chorus, the other sings the hook. 15 seconds, lip-synced.
What You Get
- The finale. The last lines of the verse, the choir's chorus and the "gangsta's paradise" hook, the part everyone sends around.
- Four shots, cut on the beat. A close-up of photo 1 pointing at the lens, a wide that pushes in as they rap, a quick close-up with photo 2 behind, and the big choir wide where photo 2 steps in to sing.
- Real lip sync. Photo 1 raps every line on the beat, photo 2 sings the hook, and whoever isn't performing keeps their mouth closed.
- A choir. Rows of singers in white-and-red robes behind you, singing the chorus.
- Any place you like. A candle-lit cathedral by default; Director's Notes can move it, and the choir follows.
- Your own outfits. Faces, hair and the clothes from your photos, down to the shoes.
- 15 seconds, vertical 9:16, ready for TikTok, Reels and Shorts. Want the whole verse? The 30-second extended cut is further down.
What the Video Actually Is
It's a viral AI parody with a backing choir that took off in early November 2026: a rapper performs Coolio's 1995 hit "Gangsta's Paradise" dead serious into the camera, rows of singers stand behind him, and at the chorus a second performer walks into frame and sings the hook into the mic. Within a week it had been reposted across TikTok and copied by several AI video apps in both 15-second and 30-second versions.
The joke is the contrast. The rapper gives it everything, pointing at the lens and sweeping an arm across the frame, while the choir stands perfectly still, and then everyone switches to full gospel for the hook.
Why it suits AI: you can't hire a choir for a joke, and the performance needs two people. Two photos put you and a friend in front of one, lip-synced, without learning the verse.
The 15 Seconds, Beat by Beat
The song runs at about 80 BPM, so one bar is about 3 seconds, and every cut lands on or near a downbeat.
| Time | Shot | What happens |
|---|---|---|
| 0:00 | Close-up of photo 1 | Raps "…how can they reach me", pointing straight at the lens |
| 3.1s | Wide, pushing in | Photo 1 raps "I guess they can't, I guess they won't, I guess they front" |
| 6.5s | Same shot | A big arm sweep across the lens hides the next cut |
| 8.1s | Close-up, photo 2 behind | Photo 1 finishes the verse, then goes quiet |
| 9.2s | Choir wide | The choir sings the chorus; photo 1 rises into frame, deadpan |
| 11.7s | Same shot | Photo 2 steps in with the mic |
| 12.2s | Same shot | Photo 2 sings "gangsta's paradise" to the end |
Who plays who: photo 1 raps, photo 2 sings. Put the friend who wants the bigger part first.
The 30-Second Extended Cut
The extended cut is the whole performance, opening on "Power and the money, money and the power". Photo 1 raps the full verse across eight shots: a close-up, a wide of them alone in front of the choir, a slow push-in from a medium shot, a wide two-shot where the two of you walk toward the camera and cross over, the point at the lens, the arm sweep, and a quick close-up. Then the same finale: the choir sings the chorus and photo 2 steps up to sing "gangsta's paradise".
Gangsta's Paradise (Extended)
The full 30 seconds: you rap the whole verse in front of the choir, then your friend sings the hook. Lip-synced.
Your Look, Their Choir
- Your own face, hair and outfit. You perform in what you're wearing in your photo, not in a costume.
- Your own shoes. The wide shots show you head to toe, so the full outfit comes along. If your photo stops at the waist, the rest is filled in to match.
- The choir and the place. Rows of robed singers, candles and stained glass by default, or wherever Director's Notes send you.
Full-length photos where you're standing and facing the camera work best, because the wides show your whole outfit. Clear, well-lit faces matter most for the close-ups. For a different setting, write it in Director's Notes ("on a basketball court at night") and the choir goes there too.
The Fastest Way
Gangsta's Paradise
Two photos, the 15-second finale with the choir, both of you lip-synced.
Two photos, 150 credits (about $7.50) for the 15-second finale or 250 credits (about $12.50) for the 30-second extended cut, no prompt. Both render on Seedance 2.5 at 480p, vertical 9:16.
Build It Yourself
1. Reframe it to vertical, shot by shot. The original is a 4:3 landscape clip, and a plain centre crop cuts the singer out of the two-shots. Pan and scan each solo shot to follow the rapper, and fit the full width on the shots where both performers are in frame.
2. Turn it into a depth map, and keep only the two performers. A depth map carries the moves, the camera and the cut points, but no faces, clothes or colour, so the original performers can't leak into your video. Mask out everyone in the background, because a depth map keeps any human shape it sees and your video would inherit them as extra people. The choir comes from the prompt instead.
3. Make one storyboard sheet with a panel per camera setup. Generate the close-ups, the wides and the choir wide from the two photos in one place, with a depth-map sheet of the original framings as a layout guide so each panel matches its shot.
A wide shot inside a vast candle-lit Gothic cathedral with tall stained-glass windows, rows of choir singers in white surplices over red robes standing on risers behind, hands raised, singing. In front, centred low in the frame, the person from reference image 1 stands deadpan with their mouth closed; at the left, the person from reference image 2 walks toward the camera singing into a handheld microphone, mouth wide open. Both wear exactly the clothes from their photos. Warm candlelight, soft haze, a touch of grain. No text.
4. Drive it with the depth map and the song. Put the audio inside the depth video so the model hears what it has to match, but mute the chorus there: nobody on screen sings it, and an audible chorus makes the model open someone's mouth. Feed the sheet, both photos and the depth video to a model that copies motion from a reference video (we recommend Seedance 2.5), then write what a depth map can't carry:
- A timed shot list: the framing for each shot and who is in it.
- The mouth windows: exactly when the rapper raps, when the singer sings, and that both mouths stay shut through the chorus.
- The choir: where it stands, and that it sings the chorus.
Render it and lay the full song back on at the end.
Common Mistakes
Centre-cropping the landscape clip. The singer walks in from the edge of the frame, so a centre crop loses them in the finale. Reframe per shot.
Leaving the background figures in the depth map. Every human shape in the original comes through as a person, and you get strangers standing beside you.
Leaving the chorus in the driver audio. The model hears singing and makes one of you mouth it. The choir sings the chorus; your mouths stay shut.
Letting photo 2 sing early. Without mouth windows, the singer mouths along to the rap. Say who performs when.
FAQ
How many photos? Exactly two: photo 1 raps, photo 2 sings.
Which one should I pick? Gangsta's Paradise for the finale everyone recognises; the Extended version if photo 1 wants the whole verse.
How long is it? 15 seconds, or 30 seconds for the extended cut, vertical 9:16.
What does it cost? 150 credits (about $7.50), or 250 credits (about $12.50) for the extended cut.
Related Reading
- How to Make the '58th Floor' AI Video: rap Anderson .Paak and Cordae's COLORS session with a friend
- How to Make the Beans Greens Potatoes Tomatoes AI Video: preach the whole Thanksgiving menu from the church mic
- How to Make the Clayface AI Video: rap Drake's "Solar Eclipse" to the camera in a nightclub VIP booth
- How to Make the They Say I'm 2 Much AI Video: rap ATM Danny's verse outside a donut shop at night
- Viral AI Video Trends 2026: every trend we've turned into a template this year
