Window Peeper AI is the night-window joke: a slow zoom catches someone eating a burger, they notice the camera, stare back and close the curtain. Upload one photo to replace the person in the scene and make a 720p MP4 for 100 credits.
Watch the example, choose who gets caught and prepare your request. You do not need to film a room, a window or yourself eating.
Upload one sharp photo of yourself, a consenting friend, a cat or a dog. A front-facing portrait with visible eyes works well for the stare. Crop out other faces so the request has one clear subject.
The prepared Prompt asks to replace the person, not redesign the room. It starts collapsed: open Edit if you want to read it. You can keep the example unchanged rather than describing the zoom or curtain timing yourself.
Fourteen seconds and vertical 9:16 are selected initially. Duration can be adjusted from four to fifteen whole seconds, and 16:9 is also available. Resolution stays at 720p. Sign in, check visibility and confirm that your balance covers 100 credits.
Follow the request in Generation history. Once it finishes, play the whole clip with sound and check the face, burger, hands or paws, and the ending. Download the MP4 after reviewing it; Kuca does not post it to social media.
A dark building fills the opening shot, and one warm window attracts the camera. The zoom changes a distant stranger into a recognisable subject. That contrast is why the joke works better as a sequence than as a still image of someone holding food.
The person notices the lens while eating and looks straight back. There is no need for dialogue or a horror twist: the pause is the joke. A photo with clear eyes gives you a better starting point than sunglasses or a face hidden by a phone.
The curtain cuts off the camera after the stare. Review that movement in the completed video, not only the opening frame. Shorter requests leave less time for the build-up; they may compress the reaction, and the model can vary the timing or miss a beat.
The linked October 2026 posts show the same window joke with different people, including a character remix shared by @therealrodentspawn. They are examples from TikTok, not videos made with Kuca. This page lets you prepare your own identity replacement without filming anyone through a real window.
Choose a well-lit portrait where the subject is looking roughly toward the camera. The photograph does not need a burger or a dining room. Those belong to the video reference; the photo tells the model whose face, hair, clothes and appearance to use.
This form accepts one image for one subject. A group photo is not a request to add a table full of friends. Crop the person you want first, and use a photo you have permission to upload rather than a private screenshot taken from someone else.
You can try a cat or dog with a clear face and visible fur markings. The instructions ask the animal to keep its own body and paws, adapting the actions instead of attaching its head to a human body. Check the generated paws, food and curtain carefully.
Your uploaded image supplies the replacement’s appearance. You do not need a suit, football shirt or the same pose as an example. If a particular outfit matters to your joke, include enough of it in the photo to make that intention clear.
These independently posted TikTok clips show how a different subject changes the reaction. Each creator is linked below the video. They are not claimed as Kuca-generated results.
The yellow-shirt subject eats at the small table, notices the lens and gets up to close the curtain. This is the prepared reference used with your photo. Follow the gaze and the last reach toward the fabric when deciding whether your portrait gives a clear enough view of the face.
Generate videoA suited character turns the slow zoom into a familiar awkward stare. Notice how the building stays dark around the warm room and how the curtain ends the encounter. This creator’s remix is the example shown beside the tool; uploading your photo does not mean asking for the actor’s face.
Generate videoThe glasses and dark jacket give this version a different expression without changing the basic window gag. The camera still approaches from outside, then the food and reaction become visible. Compare the eyes and the closing gesture, not just the outfit, with the other examples.
Generate videoThe initial length matches the prepared fourteen-second reference. That gives the approach, eating, reaction and curtain time to unfold. The slider accepts whole seconds from four to fifteen; it does not choose a different source video when you move it.
The initial 9:16 frame suits a full-screen mobile clip. Before adding your own caption in a social editor, watch where the window and face sit. A large sticker over the middle can hide the expression that makes the joke understandable.
The 16:9 option is useful when your finished project is landscape. It asks the model for a different output frame, not an extra camera or a second room. Look at the window edges in the result rather than assuming every detail will fit exactly.
All available durations and both aspect ratios cost 100 credits per generation. Output remains 720p; there is no resolution selector. Asking for fifteen seconds cannot supply more original footage than the fourteen-second reference, and a shorter request does not reduce the credit charge.
The default instructions keep the room, burger, window and camera based on the video. Rewriting all of those at once can turn the request into a different effect. If you edit the Prompt, focus on the detail you actually want to change.
The reference already contains sound. The model will try to retain its audio with the scene, without new dialogue or a replacement song. Listen before downloading or sharing; a familiar-looking thumbnail cannot tell you whether the sound and action stay together.
Look closely when the burger touches the mouth and when the subject reaches for the curtain. Hands, paws and overlapping objects can change between frames. A recognisable first image is not enough to judge the whole video, especially with pets or a tightly cropped portrait.
This is an AI scene, not evidence that someone was secretly filmed. Use your own image or get permission from a friend, and think about the context before posting. Avoid presenting the finished clip as a real intrusion into another person’s home.
Making a video is not free. Watching the examples and preparing a photo or Prompt do not spend generation credits. A task confirmed as failed returns the charged 100 credits. Generating a new version of a completed video is another paid request, even if you keep the same photo.
Free accounts generate publicly and download with Kuca’s watermark. Paid access allows Private and downloads without Kuca’s added watermark. Your photo is stored and sent to the video provider. Private does not prevent someone with the direct file link from opening it, so do not circulate that link unnecessarily.
Watch from the distant window to the last curtain movement. Check likeness, framing and sound, then decide whether you want to share this version. The downloadable file does not grant rights to source footage, music, visible brands or another person’s likeness for advertising.
Reuse parameters brings the photo and settings back into a draft without automatically generating. Download a version you like before experimenting. If a request is still processing, check its status rather than immediately paying for another. Use the same account to return to your history.
No. Each generation costs 100 credits. Watching the examples, selecting a photo and editing the Prompt do not spend generation credits. The same price applies to every duration and aspect ratio offered here.
Yes, you can try one clear photo of a cat or dog. The instructions preserve its own face, fur and body while adapting the eating and curtain gestures. Review paws and movement before sharing the result.
No. The uploaded photo identifies the subject, while the prepared video provides the scene and actions. You do not need a burger, a lit window, a second camera or a recording of your own reaction.
Fourteen seconds is the default. You can request four to fifteen whole seconds in 9:16 or 16:9, always at 720p. The reference itself is fourteen seconds, so a longer request does not add original footage.
A task confirmed as failed returns the 100 credits charged. Check its status in Generation history before trying again. A completed result you would like to remake is a separate generation, not a failed task.
This version replaces one subject from one photo. Crop a group picture to the person you want before uploading. Adding two faces to that image does not create a second assigned role or a supported two-person scene.
The reference has sound and the model will try to preserve it. You do not upload a separate track. Listen to the finished MP4 before downloading; this tool does not add a new soundtrack afterward.
The photo is stored and sent to the provider to make the video. Free generations are public. Paid access permits Private, but anyone holding a direct file link can still open the file. Avoid uploading sensitive images.
Free accounts download with Kuca’s watermark; paid access allows downloads without that added mark. This does not remove marks already visible in the reference or give permission to use source brands or footage commercially.
You can download the MP4 and prepare your own post under the destination platform’s rules. Kuca does not publish for you. Get permission for other people’s photos and do not present the AI scene as a real recording.
