← Back to Blog

How to Make the Carpool Karaoke AI Video (Migos Walk It Talk It)

By Summrs Team••4 min read

Lewis Hamilton Reposted an AI Video of Himself Rapping Migos

On September 24, 2026, Lewis Hamilton — seven-time Formula 1 world champion, knight of the realm, 36 million Instagram followers — put exactly two things on his Instagram Story. One was a real video of Kimi Antonelli yelling at him from a balcony at the Azerbaijan Grand Prix. The other was an AI-generated video of himself driving a car with Max Verstappen in the passenger seat, Charles Leclerc and Antonelli in the back, all four of them rapping Migos' "Walk It Talk It."

He tagged all three drivers. Verstappen and Hamilton are not friends. Hamilton and Leclerc are teammates at Ferrari and currently have tension. None of that mattered — the video was funny enough that Hamilton shared it anyway.

The source clip is the Migos Carpool Karaoke with James Corden from November 2018. Quavo in shotgun, Offset and Takeoff in the back, Corden driving. They sing Walk It Talk It, throw cash around the car, argue about who invented the dab, and do a shopping spree. The full episode has 68 million YouTube views and 1.7 million likes. It is one of the most-watched Carpool Karaoke segments ever.

An Instagram creator swapped the F1 drivers into that footage using AI. Same clip, same audio, different faces. The account has versions with Barcelona footballers and UFC fighters in the same car. Hamilton's repost is what sent it into the news cycle.

Here is how to make your own.

Try AI Photo Editing, Color Grading & Video Generation

Summrs analyzes each photo and applies professional edits automatically—color grading, object insertion, restoration, viral video generation and more. Describe what you want in plain English, and see results in seconds.

Try for Free →

How to Make It

1. Open the template — Migos Carpool Karaoke on Summrs

2. Upload four photos — one per seat. Driver, shotgun, backseat left, backseat right. Clear photos, face and upper body, even lighting. What you are wearing is what you are wearing in the video.

3. Generate and download — the AI swaps all four faces in one pass, matches the in-car lighting, and runs a lipsync to the Walk It Talk It audio. No watermark. Ready to post.

Four photos, a few minutes of processing, done.

Try AI Photo Editing, Color Grading & Video Generation

Summrs analyzes each photo and applies professional edits automatically—color grading, object insertion, restoration, viral video generation and more. Describe what you want in plain English, and see results in seconds.

Try for Free →

Why Four People in a Car Is Harder Than It Looks

The reason most AI face swap videos feature one or two people is not a creative choice — it is a technical limitation. More faces in the frame means more places for the model to fail.

The Carpool Karaoke clip makes this especially difficult. The camera cuts between multiple angles inside the car: driver side close-up, passenger side, wide shot of the backseat, rearview mirror angle. Each cut changes the lighting because sunlight hits from a different direction through the windshield. Face sizes change because the camera distances vary. Two people might be partially overlapping in the backseat.

If you run four separate face swaps and layer them, each pass processes the lighting independently. One person's skin tone drifts warm, another goes cool. Shadows that should fall the same direction across all four faces land differently because each pass has no awareness of the others. The result looks assembled rather than filmed.

The template runs all four in a single generation. One lighting context, one shadow direction, one color temperature across all four faces in every frame. That is the difference between a video that looks composited and one that looks like four people actually got in a car together.

Try AI Photo Editing, Color Grading & Video Generation

Summrs analyzes each photo and applies professional edits automatically—color grading, object insertion, restoration, viral video generation and more. Describe what you want in plain English, and see results in seconds.

Try for Free →

The Source Material

The Migos Carpool Karaoke is not a random clip. Some context on why it works as a template:

68 million YouTube views. That makes it one of the top 5 most-watched Carpool Karaoke episodes, alongside Adele (over 250M), Paul McCartney (over 100M), and BTS (over 100M). The Migos episode is the most-watched hip-hop Carpool Karaoke.

Walk It Talk It. The song they perform has 457 million YouTube views on the official video. The hook — "walk it like I talk it" — is one of those lines that transcended the song and entered everyday language. When someone hears it in the AI video, they recognize it immediately regardless of whether they follow Migos.

Four distinct personalities in a tight space. Corden driving, Quavo riding shotgun with the energy, Offset in the back being Offset, Takeoff holding it down. Each person has a role and a vibe. When you swap four friends into those seats, the roles map naturally — every friend group has the driver, the shotgun caller, the backseat energy.

The format is culturally permanent. Carpool Karaoke ran for eight years. The show ended in 2023 but the term "carpool karaoke" is embedded in how people describe singing in a car. When someone searches "carpool karaoke AI," they are looking for exactly this format.

Try AI Photo Editing, Color Grading & Video Generation

Summrs analyzes each photo and applies professional edits automatically—color grading, object insertion, restoration, viral video generation and more. Describe what you want in plain English, and see results in seconds.

Try for Free →

What People Are Making

The Hamilton F1 version is the one that broke through this week, but the format works for any group of four:

Friend groups — the most natural use case. Four friends in a car rapping Walk It Talk It is inherently funny because of the gap between who they are and what they are doing. The tight car framing makes it feel intimate, like a real moment someone filmed.

Sports teams — the F1 version proved this works. Four players from any team, four fighters from a card, four athletes from a rivalry. The car setting is neutral enough that it does not feel forced the way a studio backdrop might.

Work teams — four coworkers, four co-founders, four people from a department. It is team content that people actually watch instead of scrolling past.

Family — four family members in the car is literally carpool karaoke. The format was built for this.

The Hamilton repost showed something important: the AI version does not need to be perfect to go viral. The GPFans article literally describes Verstappen's face as "not especially convincing." Hamilton shared it anyway because the concept was funny. The idea carries the video — four F1 rivals trapped in a car rapping Migos together. Execution matters, but concept matters more.

Try AI Photo Editing, Color Grading & Video Generation

Summrs analyzes each photo and applies professional edits automatically—color grading, object insertion, restoration, viral video generation and more. Describe what you want in plain English, and see results in seconds.

Try for Free →

Related

Ready to Transform Your Workflow?

Edit photos, color grade entire shoots, and generate AI videos—all in one platform. Just describe what you want in plain English, and Summrs handles the technical work.

Try 10 Photos Free →