What is the Eyes Projection trend?
It is a moody portrait clip rather than a dance or a meme. One person stands in front of a wall, and the whole wall is taken up by a close-up of two eyes and their brows, as if a slide projector were throwing a giant face onto it. The person looks back at the camera over their shoulder, the projected eyes look straight at the viewer, and the picture has the warm, washed-out tones and soft blur of old film, framed by black borders.
The effect works because the eyes behind the person feel alive and a little unsettling: two gazes in one frame, one human-sized and one enormous. In the original the projection is simply part of the scene. On vswp it becomes part of the swap: the eyes on the wall are taken from your photo too, so the giant gaze behind you is yours, not a stranger's.
Where does the original video come from?
The clip follows a look that photographers and video makers have long used: projecting an image onto a wall or a person and shooting the result on film, or making it look like film. Projected-eye portraits like this one circulate on TikTok and Instagram as an aesthetic, editorial-style video.
We could not confirm who first posted this exact clip or when it went viral, so we don't name a creator or link an original post. Everything described here is what you can see in the clip and its poster on the trend page.
Who and what is in it?
There is one person in the frame, and you replace her:
- The woman (replaceable). Dark, voluminous curly hair to the shoulders, small earrings, a dark top that leaves the shoulders bare. She stands in the middle of the frame, turns her head back toward the lens and holds the pose with a calm, steady look.
- The giant eyes (made from your photo). A huge close-up of two eyes and their eyebrows fills the wall from edge to edge, with long lashes and a bright catchlight in each pupil. On vswp they are made from the same photo as the face, so the projection matches you.
- The look. Sepia, almost black-and-white tones, soft grain, gentle blur at the edges and a thick black border, like a scanned piece of old film.
- The camera. A horizontal frame, close enough to show the person from the waist up with the eyes towering around her head.
How to make the Eyes Projection trend with AI
- Open the Eyes Projection trend and sign in with Telegram. There is no password.
- Give your photo (JPEG, PNG or WebP) to the woman in the frame. The same photo supplies the giant eyes on the wall.
- Decide on "Keep the original look". On, the dark top stays and only the face and hair change; off, your outfit comes from your photo.
- Press "Make the free preview" to see about 5 seconds at 480p with a small "vswp.app" mark; every new user gets one.
- Press "Create the full video". The price in Telegram Stars depends on length, quality and number of people and is shown before you start; the Stars come back if a render fails or is canceled.
- Download the MP4 at 480p or 720p with the original sound, typically ready in minutes, and share it.
By starting a video you confirm you have the right to use it and the photos; ask a friend before using theirs. New to this? Read how to put yourself in a video, or browse more clips in the trends list.
Which photo works best?
- Open eyes, looking at the camera. The eyes on the wall are the star of this clip. A photo where you look straight into the lens with both eyes open and clearly lit gives the projection your real gaze.
- No sunglasses, no fringe over the eyes. Glasses, tinted lenses or hair across the brows hide exactly what the projection needs. Use a photo with your eyes and eyebrows fully visible.
- Sharp and evenly lit. The sepia look is soft, but your face still needs detail. Avoid strong shadows, heavy filters and blurry night shots.
- Your own hair. Your hair follows your photo, so a shot where it isn't cut off or hidden under a hat works best, whether it's curly like the original or not.
- One person per photo. A group photo or a face at the edge can mix features. Crop it to just you.
More tips: which photos work best for an AI video swap.





