What is the I Love Smith trend?

Why it caught on

A lift is an unusual place for a scene. It is small, temporary and forces two people to stand close without saying much. The original scene uses that pressure well: one person is wearing headphones, the other hears the song leaking out, and the recognition does the work of a whole conversation.

That economy is why it travels. A viewer does not need to know the film to follow the beat: two strangers, a shared song, a small moment of agreement. The scene reads in a few seconds, which is roughly the attention a short video gets before someone scrolls.

It also asks very little of the people copying it. No choreography, no big location, no dialogue to learn. You need a lift, a second person and a phone. Compared with dance trends the barrier is low; compared with talking-head trends the scene arrives with a beginning, a middle and an end.

The name drifted as it spread. Some people search for the I Love Smith trend, others for I Love The Smiths; both lead to the same clip. The short version sounds like a title, the long version sounds like the line, and the split has probably helped the scene travel rather than hurt it.

It is also a natural fit for AI. The camera is fixed, the frame is tight and only the two faces in the frame need to change. The lift, the blocking and the timing can stay as they are. The missing pieces are the two people, which is exactly what two photos supply.

What separates it from most film quotes is that it is not a monologue. The scene needs two people to work, even though only one of them speaks. That balance is why it can be remade by friends, couples, colleagues or strangers who happen to share a lift, and why it does not feel used up after one version.

The clip also survives compression. It is mostly faces, a few colours and a small amount of movement, so it still reads on a phone screen at arm’s length. Trends that depend on choreography or fast cuts often lose their shape when they are re-uploaded; this one does not.

How people join in

The most direct route is to film it. Borrow a lift, stand where the characters stand, put a phone at roughly the same height and ask a friend to take the other role. That is four things to get right at once: the space, the light, the framing and the other person’s willingness to be on camera.

The lift is the first obstacle. Mirrors, brushed metal and ceiling lights are unforgiving, and an empty lift is harder to find than it looks. One person stepping in, a door opening at the wrong moment or a phone that will not sit at the right angle can spoil an otherwise good take.

Then comes the timing. The scene works because the two performances sit close together; the look and the reply have to land in the same beat. A fraction of a second early or late and the clip stops feeling like the scene and starts feeling like two people acting it out.

Some people edit around this: film one role, then the other, and split the screen or mask the two takes together. It can work, but it needs an editing app, a steady hand and more patience than a trend that looks effortless usually gets.

The AI route removes most of that. You upload one photo of each person and the tool places you both in the scene while the framing and timing stay from the source footage. Framing, cut and timing come from the clip rather than from your shoot. The result is not a frame-for-frame copy of the film and makes no claim about the original soundtrack; it is a fan-made recreation, 18 seconds long, in 16:9.

There is also a social cost to the filmed version. You have to ask someone to stand in a lift with you, explain the scene and hope they are comfortable being posted. For a lot of people that is the real reason they never make the clip, not the editing.

The AI version is not a replacement for a good live take. It is a way to join in when the lift, the friend or the time is missing. If you already have the people and the location, filming it can be more fun; if you do not, two photos are enough to take part.

Do it with AI

The flow is deliberately short. Sign in with Google, upload one clear photo per person, confirm that you have the right to use both and start the render. It runs in the background, so you can leave the page; the finished video waits in My videos with a download button.

The photos decide most of the result. One person per photo, facing the camera, even light, head and shoulders visible. Sunglasses, masks, heavy filters and group shots all make it worse. The face has to be clear enough to survive being placed into a tight close-up.

Consent is part of the product, not a footnote. Upload a photo of yourself, or of an adult who has given you permission. No minors, no celebrities and no public figures. Photos of anyone else are removed, and accounts that break the rule can be closed.

Privacy stays simple. Photos are deleted automatically 7 days after upload; finished videos are kept for 30 days. Every output carries an AI-generated label, and you should keep it when you share. If a video shows you without consent, the report form is on every page and reports are reviewed within 48 hours.

The price is per video, not per month. 100 credits buy one 18-second video. Packs are $8.90 for 100 credits, $21.50 for 300 and $53.90 for 1,000. Credits never expire, a failed render is refunded automatically and there is no subscription.

The result is meant to be posted, not studied. It is a fan-made recreation of one scene, not a definitive version of it, and it will not match the film frame for frame. What it gives you is the part that matters for the trend: both faces in the elevator, at the right moment.

If you are unsure whether a photo will work, start with the clearest one you have. A failed render costs nothing because the credits come back, but a good source photo saves you a second attempt.

Save the file rather than relying on the share link. The link is useful for sending the clip to someone, but it is not an archive: it stops working when the video expires, is removed or you turn sharing off.

One honest note to finish. This is an independent, fan-made tool. It is not affiliated with the film, Searchlight Pictures or The Smiths, and the output is an AI recreation rather than a copy of the scene. If that is the version of the trend you want, it is two photos away.