How to Do the Gangsta’s Paradise AI Trend With Two Photos
Two photos, two unequal roles: one raps the verse, one takes the chorus. The browser steps, per-slot photo rules, the frame choice, and the fixes that matter.
Two photos, two jobs. In a Gangsta’s Paradise AI video the first photo takes the verse at the microphone; the second arrives around eight seconds and sings the chorus. The ranch, the riders, the hard cut and the song are fixed. Only the faces change.
That division is the whole trick, and it decides how you shoot, cast and troubleshoot. This guide walks the browser route step by step, then covers what each slot needs, the frame choice, and the failures worth fixing.
The two roles are not equal
| Slot | On screen | Job |
|---|---|---|
| 1 | 0–15 s | Opens the verse alone at the microphone, returns after the cut |
| 2 | ~8–15 s | Appears behind the rapper’s shoulder, enters from the left for the chorus |
The rapper carries the clip. The friend owns the payoff. Casting the funniest person into slot one and the reliable one into slot two usually produces a flatter video than doing it the other way around.
What you need
- Photo one: the rapper. Full-body if possible, facing the camera, sharp.
- Photo two: the friend. One person, clear face, chest-up is enough.
- A browser. The effect runs here, with no app and no location to find.
- Credits. The button quotes the price before you commit, and a failed render returns it.
Five steps in the browser
1. Open the effect and give slot one to the rapper
Gangsta’s Paradise AI video opens with the effect already selected. Slot one is on screen from the first bar to the freeze, so choose the face that should hold the frame, not the one with the best punchline.
2. Put the friend in slot two
Use a separate file with one person in it. The model places people by slot; it cannot split a group photo into two identities. The friend shows up behind the rapper’s shoulder near eight seconds, then walks in from the bottom left for the chorus.
3. Do not shorten the clip
Fifteen seconds is the format. The first nine seconds build the verse and set up the cut; the last six belong to the chorus. A five-second render stops before the friend arrives, so you pay for the setup and never reach the joke. Leave 4:3 alone unless you specifically want a vertical export.
4. Choose the resolution
480p for a test, 720p for posting, higher only if the clip will be cut into a larger edit. Each option shows its own credit cost.
5. Render, download, post
A fifteen-second render usually finishes in a few minutes. The MP4 leaves with the clip’s audio already inside it. Post it with the platform’s AI-generated label; the performance is synthetic, and the label says so.
What each photo needs
Slot one wants the whole body. The opening shot is wide enough to include boots, so a full-body photo keeps height, build and clothing consistent. A tight headshot leaves the model to invent the rest.
Slot two can be closer. The friend is seen from the chest up, first over a shoulder and then beside the rapper. A clear chest-up photo works, as long as the face is not shaded or turned away.
Both photos need even, frontal light. Backlighting, a heavy filter or a face at an angle forces the model to guess, and the guess shows up as drift later in the clip.
One face per file. Crop before uploading. Two faces in one slot means the model picks one, and it often picks wrong.
Some energy helps. A frame taken mid-sentence or mid-laugh gives the model a more useful mouth and brow than a passport stare.
Consent is not optional. Use your own photos or friends who agreed. Skip public figures and children; the joke only works if everyone in the frame is in on it.
4:3, 9:16 and the hangar mix-up
The effect rebuilds the ranch. The default 4:3 frame matches the TikTok cut; 9:16 gives you the same ranch in a vertical crop. It does not switch the scene to the white space hangar from the earlier YouTube version. If the hangar is what you want, this effect is the wrong tool.
| Frame | What you get | When to use it |
|---|---|---|
| 4:3 | Full ranch width, riders visible | The default; matches the trend |
| 9:16 | Vertical crop of the same ranch | Feed-first posts; expect less crowd |
When the result looks wrong
- The friend never shows up. The duration was shortened. Fifteen seconds is required for the entrance and the chorus.
- The friend has the wrong face. Slot two received a group photo, or both faces came from one file.
- The rapper drifts after the cut. The source photo was small, angled or filtered. Replace it with a sharp, frontal shot.
- The outfit looks borrowed. Costume pieces such as boots can survive from the template. A full-body photo reduces the mismatch.
- The crowd looks artificial. It is generated, not filmed. That is the format, not a bug.
- The vertical version loses the riders. A 9:16 crop removes the edges of the ranch. Use 4:3 if the crowd matters.
- The render fails. Credits return automatically. Retry with a clearer photo before touching other settings.
Change one variable per retry. If you swap the photo, the slot and the resolution together, you cannot tell which change helped.
One photo, two photos, and the chorus
With one photo, the rapper is yours and the singer is the model’s invention. The clip still runs, but the duet becomes a solo with a stranger. Two photos are the point: the chorus lands on a face the viewer recognises.
Browser effect or CapCut?
| Need | Route |
|---|---|
| Two chosen people | Browser effect |
| A solo clip in under a minute | CapCut template |
| The 4:3 ranch frame | Browser effect |
| No app installation | Browser effect |
| Resolution control and retries | Browser effect |
| The same TikTok sound page | CapCut template |
CapCut gets you into the trend fastest with one face. The browser effect is for the two-person version, and for anyone who wants to re-render at a different resolution.
Keep the song, caption the relationship
The audio is half the joke. The render includes the clip’s fifteen seconds; if you want the exact original mix, attach the trend’s sound page or replace the audio in CapCut. Do not substitute another track.
In the caption, name the relationship — “my brother and me”, “the coach and the point guard”, “she raps, I sing”. The roles explain the joke faster than any hashtag. Label the post as AI, keep it to fifteen seconds, and publish while the trend is still moving.
Filming the parody without AI
The real version is a two-person shoot: learn the verse, hand the chorus to a friend, frame at chest height in 4:3, and stop at fifteen seconds. Outdoors is better; snow is optional. It will not look like the ranch, but both people will have chosen to be there.
Questions people ask
Do I need two photos? Two gives you the duet. One fills the rapper’s place, but the model invents whoever sings the chorus.
Who goes in which slot? The rapper goes first and stays on screen for the full clip. The friend goes second and arrives around eight seconds.
Why does the friend appear so late? Because the source edit saves the chorus for the end. The entrance is the payoff; shortening the clip removes it.
Can I post it vertically? Yes. Set 9:16, but expect the same ranch scene with the sides cropped, not the hangar.
Can I use Coolio’s photo? No. Public figures are refused, and the song already carries the reference.
How much does a render cost? The generate button shows the exact credit cost for the chosen resolution before you confirm. Failures are refunded.
Does the file include the song? Yes, the clip’s fifteen seconds render with audio. For the original mix, use the trend’s sound page or replace the audio in CapCut.
Read next
- What is the Gangsta’s Paradise AI trend?
- Make your Gangsta’s Paradise AI video
- How much a video costs
- All AI video effects
RapReel is independent of Coolio’s estate, L.V., their labels, TikTok and CapCut. Upload only photos of people who consented to appear.