The AI mini me trend takes an ordinary photo of a person and turns them into a miniature figure standing inside a full-size scene. You end up as a tiny version of yourself crossing a tabletop, sitting on a coffee cup, or wandering through a room that suddenly looks enormous. It is one of the effects on this site, and it runs on the same image-to-video pipeline as everything else here.
What the AI mini me effect actually does
You upload a photo. The site sends it through an image-to-video model and returns a short cinematic clip. For the mini me effect, the model scales the person down and keeps the surroundings life-size, so the proportions between the two stop matching the real world. A person who was standing in a normal room comes back as a small figure in that same room, and the room now reads as huge.
The clip is short by design. A five-second shot is enough for a small character to walk a few steps, turn, or look up at something far above them. Those few seconds are what make the scale trick land: your eye sees movement first, then works out that the moving thing is far too small for its setting. That delay is the whole appeal.
Nothing about the effect needs a second person or a partner photo. One clear picture of you is enough. The output is a normal video file, watermarked in the free preview, and the same clip without the watermark once you spend credits.
Why the miniature look caught on
The miniature contrast reads instantly on a phone screen. You do not need a caption or an explanation the way you do with some effects, because the scale mismatch is obvious in the first second. A big object and a tiny person in the same frame is a visual joke that works without sound, which is exactly what performs well on TikTok, Instagram Reels and YouTube Shorts.
The trend also fits a wide range of photos. You do not need a studio shot or a costume. A normal photo of a person standing up is enough to start, and that low barrier is why so many people tried it rather than just watched it. If you have one usable picture, you can be the subject.
If you want to see what the effect produces before you shoot anything, you can try the effect on this site: upload one photo to the AI mini me generator and you get a watermarked preview you can watch right away.
How to shoot a photo that shrinks well
The model can only work with what is in the picture, so the photo decides most of the result.
Stand the person up and keep them fully in frame. Head to toe, no cropping at the knees and no hand cut off at the edge. When the figure is scaled down, a cropped body has nothing to complete it, and the clip looks broken instead of small.
Keep the background simple. A plain wall, an open floor, or a single large object behind the person gives the model something stable to scale against. A busy room full of other people and furniture makes it guess which things are meant to be the environment, and the result gets muddier.
Use even light. A photo where the face is lit but the body is a dark shape gives the model little to read, and the small figure comes back as a smudge. Soft, consistent light across the whole person works much better than a harsh light from one side.
Pose so the body reads at a distance. Straight arms, a wide stance, or a move caught mid-step all survive being shrunk because the outline stays clear. A tucked-in pose turns into an unrecognizable blob once the person is tiny.
Avoid clutter behind the head and shoulders. A tree branch, a lamp, or a door frame that sits right behind a person often gets carried into the clip as something attached to them, and at small scale that looks like a growth rather than scenery.
Ideas for tiny-figure clips
Once you have a photo that works, the setting does the heavy lifting. A few directions that fit the five-second format:
A desk explorer. The person walks across a keyboard or between coffee cups. The normal objects around them become landmarks, and the everyday desk turns into a landscape.
A kitchen counter scene. A tiny figure walks past a full-size mug or looks up at a fruit bowl. The camera stays at counter height, so the scale surprise comes from the objects, not the room.
A tabletop meal. Sitting on the rim of a plate or next to a fork, a small character looks like they are dwarfed by dinner. It is a simple shot that gets the joke across with two objects and one person.
A plant pot. Standing at the base of a houseplant, the person reads as an insect-scale figure in a jungle. Green leaves above them sell the height without any editing tricks.
A shelf or bookshelf. Walking along a shelf with books as tall as buildings behind them gives the miniature clip a city feel without leaving the house.
The same idea can run in reverse. If you would rather blow a small object up to giant scale instead of shrinking a person, the make it giant effect uses the same pipeline and flips the direction.
What a clip costs
The mini me effect is not a free tool and there is no free credit to claim. What you get for free is a watermarked preview of your clip, so you can watch the result and decide whether it is worth keeping before you spend anything. Buying credits removes the watermark and unlocks the download.
One clip costs 300 credits, which is one five-second video. The entry tier is $4.99 for 300 credits, so a single clip is $4.99. If you plan to post several, the larger packs are cheaper per clip: $9.99 gives 1,500 credits for five clips, and $19.99 gives 3,300 credits for eleven clips. Those same prices are the monthly tiers, and there are yearly options as well. You can see the full breakdown on the pricing page.
Where this fits next to the other effects
The mini me effect is one of five on this site, all sharing the same image-to-video step. The zombie video generator covers the two-photo love story version, and the same engine handles the pet, Halloween makeup and giant-object effects. That means once you learn how to shoot a photo for one of them, the same habits carry over to the others.
If you already have a picture in mind, the simplest path is to try it on this site. Upload the photo, watch the watermarked preview, and only spend credits once the clip is what you wanted. There is no reason to guess whether a photo works when the preview shows you the answer for free.