The I Love the Smiths AI trend puts two new faces into the elevator scene from 500 Days of Summer: one person on headphones, one person who says they love The Smiths. videosi makes it from two photos as a 15-second clip with the film sound on it. This guide goes deeper than the trend page, covering where the scene comes from, how new the trend is, how to get a good likeness and what to expect from your own clip.
It has only just started to spread, with the first dozen or so AI posts dated 7 and 8 November 2026, so the details below describe an early phase.
Key takeaways
- The scene is the headphones elevator meet-cute from the 2009 film 500 Days of Summer, cut as alternating close-ups.
- The AI version was only spotted on 6 to 8 November 2026, so it is an early, fast-moving trend.
- You need two photos: photo 1 is the person on headphones, photo 2 is the person who says the line. No prompt is written.
- The film sound is on the clip, but mouths move loosely and do not lip-sync word for word.
- A 15-second Standard 720p clip is 13 credits, and packs start at $6.90 for 10 credits.
In this guide
- What is the elevator scene the trend copies?
- How new is the AI version, and how far has it spread?
- Why does this scene work so well as an AI trend?
- Which photo goes where, and who plays which role?
- What kind of photo gives the best likeness in a close-up?
- What about the sound and the dialogue?
- What actually makes the video, and why is the prompt hidden?
- What will not work in this scene?
- Which format should you pick, and can you do it on a phone?
- How should you post it, and do you need to label it?
- How much does one clip cost, and is it worth a retry?
What is the elevator scene the trend copies?
The source is the 2009 film 500 Days of Summer. Tom stands in an office elevator with big headphones on and music leaking out. Summer steps in, hears it and says she loves The Smiths. He does not hear her, so he pulls one cup off his ear and she says it again. The scene is cut as a pair of close-ups, one face at a time, and ends with her singing a line from "There Is a Light That Never Goes Out" back at him.
The AI version keeps that structure exactly. What changes are the two faces. That is why the trend feels familiar at once: the viewer already knows the beats, so the only surprise is who is standing in the elevator.
How new is the AI version, and how far has it spread?
Very new. The film clip has been posted on TikTok for years, but the AI wave only showed up in early November 2026. Our radar flagged it on 6 November, and on 7 and 8 November we found more than a dozen posts, mostly tutorials and edits titled with the scene name. Pairings in those posts range from famous couples to ordinary friends.
We cannot confirm who posted the first AI version, and we do not quote view counts we have not checked. Treat it as an emerging trend rather than an established one: it may grow fast, or fade in a couple of weeks.
Why does this scene work so well as an AI trend?
Three reasons. First, it is short and has a clear shape: listening, a turn, a line, a headphone pulled off, the line again, a shared smile. Second, it needs only two people and one location, which suits a photo-to-video model. Third, the emotion is small and shy rather than loud, so a recognisable face and a nervous glance carry the whole clip.
It also gives people a reason to tag someone. You can cast a partner, a friend, a crush or yourself with a celebrity, and the video says what you would never say out loud.
Which photo goes where, and who plays which role?
Photo 1 is the person on headphones, the one who is listening. Photo 2 is the person who leans over and says the line. The roles are fixed by the order of the uploads, so swapping the photos swaps the roles.
Each person is taken only from their own photo, and the two are never blended. The headphones are added to the first person automatically, on the head at the start and round the neck later, so you do not need to wear any in the picture. The elevator, the cuts and the timing are built in, and you write no prompt.
What kind of photo gives the best likeness in a close-up?
Both people appear in tight close-ups, so face-shape drift shows up here before anywhere else. A front-facing, evenly lit photo with open eyes is the safest choice. Show your hair and shoulders, because hair and build carry a lot of the likeness, and the clothes in your photo are the clothes you wear in the elevator.
Avoid sunglasses, hats, strong side angles and group photos. If a photo shows two people the model may use both, so crop to one face per photo. A photo that already has headphones can also confuse the first role.
The full step by step is on the trend page.
What about the sound and the dialogue?
The finished video carries the trend audio, which is the original film sound from the elevator scene, including the line "I love The Smiths" and the lyric "To die by your side is such a heavenly way to die". It is not a single official TikTok song, it is the film audio, and videosi lays it over your clip.
One limit to know: mouths move loosely and do not lip-sync the words one by one. In a quiet, shy scene that looks natural, but do not expect a perfect match. If you prefer another sound, you can swap it in your posting app.
What actually makes the video, and why is the prompt hidden?
The clip is made with ByteDance's Seedance 2.0 or 2.5, depending on the tier you pick. The model receives your two photos plus a short motion reference of the elevator cuts, which is what keeps the alternating close-ups and the pulled-off headphone in place.
The prompt is built in and not shown, so there is nothing to copy and paste. The reason is practical: a plain text prompt tends to produce one static two-shot with no rhythm, and it often half-blends your faces with the actors from the film. A tested setup avoids both problems, and it also forbids garbled subtitles from leaking into the result.
What will not work in this scene?
The clip has exactly two people, and nobody else is added. It is conversation and eye contact only: no hug, no kiss, no touching. The elevator and the headphones are fixed, so you cannot move the scene to another place.
Likeness is the other limit. Two faces in close-up is the hardest case, and a failed try can happen. If a result looks off, the fix is nearly always a better photo rather than a different setting.
Be aware that the effect is brand new. The examples on the trend page are AI clips of the trend found online, not videosi renders, so treat them as a picture of the scene rather than a promise of the exact result.
Which format should you pick, and can you do it on a phone?
Pick 9:16 for TikTok, Reels and Shorts, or 1:1 or 16:9 for other places. The output is a 15-second MP4 with no watermark. Rendering takes several minutes, so start it and come back.
It works in a phone browser: choose both photos from your camera roll, press Generate and download the file. You only need an account to pay and download.
How should you post it, and do you need to label it?
Yes, label it as AI-generated. The clip uses real faces from your photos, so use only photos you have the right to use, and be careful with other people. A photo of a partner or a friend is the intended use; a photo of someone who has not agreed to it is not.
For reach, lean on the joke. A caption that names the pairing and the line works better than a long explanation, and the first second should show the headphones so viewers recognise the scene at once.
How much does one clip cost, and is it worth a retry?
A 15-second clip on the Standard model at 720p costs 13 credits. Credit packs start at $6.90 for 10 credits. Check the pack sizes on the page before you buy, since the cheapest pack alone does not cover a 13-credit clip.
Because each try costs credits, spend the effort before you press Generate: pick a sharp, front-facing photo for each person, crop to one face, and check which photo is first. That does more for the result than a second render.
Ready to try it? Open the I Love the Smiths AI generator, add your two photos and make the elevator scene in the browser.
