The street light slide is a 6-second night clip in which a tiny figure grabs a lamp post, slips out of frame and slides back in as a huge close-up. To make it, you upload one photo of yourself and let a motion reference supply the timing. This guide covers street light slide AI in more depth than the page FAQ: why it works, what to photograph, and what to check before you post.
Key takeaways
- The hook is a sudden change of scale: from a speck in a plaza to a chest-up portrait in under three seconds.
- The trend goes by several names, so search phrasing varies a lot. The origin and the sound's source are both unconfirmed.
- A good result depends mostly on the photo: one person, upper body, clear face, clothes that show at night.
- The trend sound includes strong language, so think about where you post it.
- Label the finished clip as AI-generated on the platform you publish to.
In this guide
- Why does the sudden jump in scale grab attention?
- What else is this trend called?
- What is the difference between a template and a finished-video web tool?
- How do I choose the right photo?
- How do clothes show up in a night scene?
- Why is my face small in the first seconds?
- Why does the empty plaza matter?
- What model sits behind it?
- What should I know about the sound before posting?
- How should I label the clip as AI?
- What does one try cost?
Why does the sudden jump in scale grab attention?
Most short videos hold one framing and let the subject move inside it. This one breaks that habit on purpose. For the first 0.7 seconds you see a very small person beside a bright blue-white lamp, so the eye searches the frame. Then the person swings around the pole and slips sideways along the ground, and the shot is empty.
That empty beat creates a question: where did they go? The answer arrives at about 2.6 seconds, when the same person slides in from the lower left, many times larger. The viewer gets surprise and payoff inside one clip, and a calm look into the lens in the last three seconds gives the video a clean ending that loops well.
Nothing about the idea needs a story or a caption. It is pure staging: far, gone, near. That is also why a plain photo of yourself is enough input, because the clip carries the interest by itself.
What else is this trend called?
On TikTok the same move shows up under several labels: AI slide transition, fun slide transition, streetlight transition, street lamp transition, lamp post transition and street light dance. People also add words like template, generator or CapCut when they search. They all describe the same thing: a person leaves a lamp post at night and returns as a sudden close-up.
This matters for two reasons. First, if you search only one phrase you will miss many clips. Second, the page you land on may use a different name from the one you typed. On videosi these names point to one effect at /ai-slide-transition.
As for where it started, we can only say what we checked. The earliest dated posts we found are from 21 October 2026, and they are template and tutorial clips. We did not identify an original creator, so the origin is unconfirmed. We also do not quote view counts beyond what we verified, and nearly all the reach we saw came from accounts sharing a template.
What is the difference between a template and a finished-video web tool?
A photo template inside an editing app usually takes your picture and applies motion effects such as zooms, slides and cutouts to a still image. The person never truly moves; the picture is animated. That can look fine for a quick post, and it needs an app and a few taps.
A web tool that renders a finished video works differently. It generates new frames in which the person in your photo performs the move: hand on the pole, the swing, the slide-in. You upload a photo in a browser, wait several minutes and download a video. There is nothing to install and no timeline to edit.
Neither is better in every case. If you want full control of every cut, a template and an editor suit you. If you want a ready 6-second clip with the person actually moving, a render is the closer match. videosi is independent and not affiliated with any editing app or social platform.
How do I choose the right photo?
Pick a photo with one person, a clear face turned toward the camera, and the upper body in frame. The last three seconds of the clip are a chest-up portrait, so a picture cropped tight to the face forces the model to invent shoulders and clothing.
Avoid sunglasses, hats and hair across the eyes, because the likeness drifts and the close-up shows it. Crop out friends, since the trend has exactly one person. Good light on your face in the original photo helps more than any filter.
If the photo is of someone else, they must be an adult who agreed to it. Do not use a photo of a person who has not given permission.
How do clothes show up in a night scene?
Whatever you wear in the photo is what you wear in the video. The scene is a dark plaza with one bright lamp, so a black top can vanish into the night sky and lose its shape. A lighter or brighter top reads better, and a jacket with a visible collar gives the close-up more structure.
Texture helps too. A knit or a patterned shirt gives the lamp light something to catch. Shoes matter less because the camera rarely sees them once you are close. If you have two photos that are equally sharp, choose the one with the clearer outline against a plain background.
Why is my face small in the first seconds?
That is the design. For the first 2.6 seconds you are a small figure far from the lens, and a face at that size cannot show fine detail in any video. The scale jump only works because the first shot is distant.
So judge the likeness in the close-up, from about 3.2 seconds to the end. If the far shot looks slightly different from the near one, that is a known weakness of the approach, and a sharper, front-facing source photo gives the best chance of consistency. Likeness is never guaranteed, and a failed or timed-out render returns its credits automatically.
Why does the empty plaza matter?
The one-second pause with nobody in frame is the easiest part to lose. Video models like to keep a subject on screen, and without a reference clip they tend to walk the person toward the camera. That removes both the disappearance and the jump in scale, so the clip becomes an ordinary walk.
This is why a motion reference matters for this trend. The reference fixes the shots, the timing and the fixed camera, while your photo only supplies the person. The page at street light slide AI lists all five beats with their timings if you want to see the full structure.
What model sits behind it?
videosi uses ByteDance's Seedance family, with a choice of tiers. ByteDance Seed describes Seedance 2.0 as a model that accepts text, image, audio and video inputs, so a photo and a reference clip can work together. That is the combination this effect relies on: a motion reference for the staging and a photo for the person.
We do not publish the full prompt. What you control is the photo and the aspect ratio: 9:16 for TikTok, Reels and Shorts, or 1:1 or 16:9 for other feeds.
What should I know about the sound before posting?
Most clips we checked use one TikTok sound, listed as an original sound by a user called naddy. It is built around a spoken line with strong language. The song or source behind it is unconfirmed, so we cannot tell you who owns it.
Two practical points follow. Your lips do not follow the spoken line, because the video has no lip sync; the sound is simply laid over the clip. And if you post somewhere with a family audience, or to a workplace or school account, the strong language may be a problem. Preview the clip with the sound on before you publish, and mute it or swap the audio if the setting calls for it.
How should I label the clip as AI?
Use the platform's own AI label when you upload. TikTok's community guidelines say creators must label AI-generated or significantly edited content that shows realistic-looking scenes or people, and that unlabeled content may be removed, restricted or labeled by TikTok depending on potential harm.
TikTok has also said it reads C2PA Content Credentials to label AI content uploaded from other platforms. Its newsroom post on AI transparency notes that this depends on the metadata being present, so do not count on automatic detection. Add the label yourself, and say in the caption that the clip is AI-generated if you want to be plain about it.
Other platforms have their own settings, and they change, so check the upload screen where you post.
What does one try cost?
A 6-second street light slide costs 5 credits on the Standard model at 720p, and higher tiers cost more. Credit packs start at $6.99 for 10 credits. Each render takes several minutes, and the result is saved in My creations if you close the page.
Because each attempt costs credits, put your effort into the photo first. One sharp, front-facing, upper-body picture with a clothing color that shows at night will do more for the result than repeated tries with a weak one. When you are ready, start from the street light slide page.
