Quick answer
Starrd's How Long You Been Faithful? template puts you in the viral podcast moment. Upload one to three photos: photo 1 is the guest, who brags 'Sixteen years', hears 'How long you been faithful?', takes a sip through a straw and chokes on it mid-answer. Photo 2 is the host who asks 'You back to cheating, bro?' and runs laps around the studio laughing; photo 3 is the co-host who asks the question and doubles over. Any seat without a photo gets an AI-cast stranger. Everyone keeps the outfit from their photo, every line is lip-synced to the original audio, and the six shots cut on the same frames as the original. 20 seconds, widescreen 16:9, Seedance 2.5, 480p, 200 credits (about $10). People only.
The "How long you been faithful?" podcast choke, starring you and your friends. Here it is with a grandmother as the guest:
How Long You Been Faithful?
1-3 photos. You brag about 16 years, they ask how long you've been faithful, you choke on your drink. 20 seconds.
What You Get
- The setup. "You back to cheating, bro?" "That's my goal?" "...for 2019. Sixteen years."
- The question. Your co-host turns and asks it: "How long you been faithful?"
- The choke. You sip your iced coffee, start to answer, and choke on it, coughing into your fist.
- The meltdown. A wide shot of the studio: the host kicks back, falls out of the chair and runs laps; the co-host doubles over.
- A podcast studio. Round desk, mixing console, boom mics, monitors.
- Lip sync, your own outfits, the original sound, 20 seconds, widescreen.
What the Video Actually Is
The clip comes from The Brilliant Idiots, the podcast hosted by Andrew Schulz and Charlamagne tha God, with comedian Gary Owen as the guest. Asked how long he's been married, Gary says 16 years. Asked how long he's been faithful, he takes a sip and chokes before he can answer.
Reposts captioned "God literally steps in to stop him from telling lies" keep circulating; a TikTok repost from @rolexomega passed 550,000 plays.
Why it suits AI: the joke is a face getting caught out and a room losing it. Put your most suspicious friend in the guest seat.
The 20 Seconds, Shot by Shot
| Time | Shot | What happens |
|---|---|---|
| 0:00 | Photo 2, close-up | "You back to cheating, bro? This year? That's my goal." |
| 1.8s | Photo 3, side-on | Points: "That's my goal?" |
| 2.7s | Photo 1, close-up | Holding the iced coffee, smug: "...for 2019. Sixteen years." |
| 4.0s | Photo 3, side-on | Turns and asks: "How long you been faithful?" |
| 4.7s | Photo 1, close-up | Sips through the straw, starts to answer, chokes, coughs into a fist |
| 7.0s | Wide | The studio erupts: the host runs laps, the co-host doubles over, the guest coughs and laughs at the back |
Your Look, Your Roles
- Photo order sets the seats. Photo 1 is the guest, photo 2 the host, photo 3 the co-host.
- Your own outfits. Whatever each of you wears in your photo, you wear in the studio.
- Empty seats get strangers. Upload one photo and the hosts are cast for you.
Put the friend everyone suspects in photo 1: the guest gets the brag, the sip and the choke.
The Fastest Way
How Long You Been Faithful?
1-3 photos in a podcast studio, lip-synced to the original sound.
One to three photos, 200 credits (about $10), no prompt. It renders on Seedance 2.5 at 480p, widescreen 16:9, 20 seconds.
Build It Yourself
1. Crop the clip, then turn it into a depth map. The repost is a widescreen picture boxed inside a vertical video with a caption above it. Crop the picture, then run it through a depth-estimation model (Depth Anything V2 works). Smooth out the host's cap and blur the podcast logo so neither comes back.
2. Make a character sheet. The empty studio across the top and a full turnaround of each person below, with only the guest holding the iced coffee.
A photorealistic VERTICAL 9:16 CHARACTER REFERENCE SHEET in four bands: an empty podcast studio with a round desk, mixing console and boom mics on top, then a four-angle turnaround of each subject, the guest holding an iced coffee with a green straw, the studio softly out of focus behind every panel. No text.
3. Drive it with the depth map and the audio. Feed the sheet, your photos and the depth video, with the dialogue inside it, to a model that copies motion from a reference video (we recommend Seedance 2.5, because three people talk). Then write what it can't read from a depth map:
- Shot ownership: who appears in each shot, so the guest's close-ups never go to someone else.
- The mouth windows: each line on the right mouth, and nobody mouthing an off-screen voice.
- The faces: teasing, sceptical, smug, then caught out.
Lay the original audio back on at the end.
Common Mistakes
Trusting a mouth tracker on side-on faces. The co-host is in profile when asking the question; trackers read that mouth as shut. Check it by eye, or the wrong person gets the line.
Using a cheaper model with three talkers. It can hand the guest's close-up to the co-host. Seedance 2.5 kept every shot with the right person.
Letting the original host's jacket through. His outline can put a jacket on whoever sits in his seat. Tell the model nobody wears a jacket their photo doesn't show.
Everyone holding a cup. Say only the guest holds the iced coffee, or the sheet hands one to all three.
FAQ
How many photos? One to three.
How long is it? 20 seconds, widescreen 16:9.
What does it cost? 200 credits, about $10.
Pets? No, people only.
Related Reading
- How to Make the 'That's My Opinion' AI Video: another three-person argument meme
- How to Make the 'I Be Thinking Bout My Bills' AI Video: a three-seat laugh-at-a-friend skit
- How to Make the 'Deep Down in My Heart' AI Video: an interview meme for two
- Viral AI Video Trends 2026: every trend we've turned into a template this year