How to Make the Carpool Karaoke AI Video (Migos Car Trend)
How to make a Carpool Karaoke AI video: seat 1–4 photos, pick 9:16 or 16:9 and get a 15s clip with the original hook. Plus the Genjutsu and prompt methods.
Last updated: 2026-10-16
To make a Carpool Karaoke AI video, add one clear photo per seat (one to four riders), pick 9:16 or 16:9, and generate: about five minutes later you get a 15-second MP4 of your crew in the viral Walk It Like I Talk It car scene, with the original hook lined up. Below is that method step by step, then the two ways to do it yourself with other tools.
Method 1: this site, from photos (easiest)
- Open the generator on the homepage.
- Seat your crew. Add one photo per seat: 1 · Shotgun (lead), 2 · Driver, 3 · Back left, 4 · Back right. Fill one seat or all four; empty seats get a default passenger.
- Mark pets. Each seat has a person/pet switch. Pets sit upright and bob to the beat.
- Pick size. 9:16 for TikTok, Reels and Shorts, or 16:9 like the TV segment. HD 720p (100 credits) or 480p (50 credits).
- Tap “Make my video”. Sign in with Google and pick a pack if you have no credits ($9.90 for one HD video). Generation takes about five minutes; you can close the page and find the video in My videos.
- Download the MP4 from the result panel or My videos (kept for 14 days).
Read this first: seat order is the casting plan
In the original clip, the front passenger leads the song and gets the most close-ups, the driver reacts at the wheel, and the back seat opens the video. Put your loudest friend (or the star of the fan edit) in seat 1. Drag a photo onto another seat to swap riders before you generate.
Photo tips that make the biggest difference
- One person or pet per photo. Group photos blend faces — crop each face out first.
- Front-facing, sharp, evenly lit, head and shoulders in frame.
- No sunglasses, masks or heavy filters.
- Distinct outfits help the AI keep four riders apart.
- For pets: the whole head, facing the camera.
Method 2: Higgsfield Genjutsu (how the viral versions were made)
The F1 version that started the trend was captioned “made with Higgsfield Genjutsu”. Genjutsu swaps characters into an existing clip:
- Get a paid Higgsfield plan.
- Find and trim a clip of the car scene to use as the reference video.
- Add a character image for each rider you want to replace and describe the swap.
- Generate, check that faces didn’t blend between seats, and retry if they did.
- The clip’s own audio comes through, so the hook is already there.
It gives the most literal copy of the episode, but you need the subscription, the source clip, a written prompt and usually a few retries.
Method 3: any AI video model with a prompt
Models that take several reference images (and ideally a reference audio track) can make the scene from scratch. Our Carpool Karaoke AI prompt template gives a seat-by-seat prompt and the shot list with timings. Expect more trial and error: without the trend’s audio, the lip-sync won’t match the hook, and you’ll need to add the sound in an editor.
Troubleshooting
| Problem | Fix |
|---|---|
| Two riders look alike or swapped | Use photos with different outfits and clearer faces; one face per photo |
| A rider doesn’t look like their photo | Use a sharper, front-facing photo without sunglasses or filters |
| A pet looks off | Use a photo where the pet faces the camera with its whole head visible |
| The video failed | Your credits come back automatically — try again or swap a photo |
Posting it
- Turn on the platform’s AI-generated label on TikTok, Instagram or YouTube.
- Tag everyone in the car; group clips travel further when every rider shares them.
- The video already has the hook. You can also pick the trend’s sound in TikTok so your post shows up on the sound page.
- Fan edits of real people are fine as long as they’re clearly AI and not presented as real footage or an endorsement.
How long a Carpool Karaoke AI video takes and costs
A Carpool Karaoke AI video takes about five minutes from tapping the button to the download. One 15-second HD video is $9.90 (100 credits); 480p is 50 credits, and packs bring HD down to $4.99 a video. There is no subscription and no free trial — if a video fails, the credits come back on their own.
Quick checklist
- One clear photo per seat, faces visible
- Lead rider in seat 1 (shotgun)
- 9:16 for phones, 16:9 for YouTube
- HD for posting, 480p to try a new crew
- AI label on, crew tagged