One photo · Six steps · One 14-second video
Upload one clear everyday photo of a consenting adult or one cat or dog. Choose the subject, room and quality, then generate a new 14-second 9:16 MP4 with the building approach, stare and curtain-close payoff.
Written by Window Peeper AI editorial team · Updated October 22, 2026
01 · Prepare the photo
The input is not a character sheet or background-free cutout. Use an ordinary photo with a natural setting. The image guides the subject’s recognizable face, fur, clothing and proportions; the selected room option guides the generated scene.
02 · Make the video
The controls are deliberately narrow. You choose the subject route, photo, room treatment and output quality. The preset handles the repeated story structure, so there is no prompt box or editing timeline to configure.
Pick the control that matches the single subject in your photo. Person is for one consenting adult. Animal is for one cat or dog.
Upload a clear photo with a natural background. You do not need to cut out the subject or upload a separate room image.
Leave Auto selected for a fitting room, or pick one of eight authored ideas such as Secret CEO Meeting, Grandma’s Living Room or Tiny Dollhouse Room.
Standard 480p uses 12 credits. HD 720p uses 22 credits and is the better choice when you plan to post the result.
Confirm that you are allowed to use the photo, sign in if needed and submit the setup. Do not refresh away just to restart a running job.
Watch the full result, check the subject, room, burger, stare, curtains and audio, then download the finished MP4 when it is ready to share.
Not sure which room to choose?
Use Auto for the simplest path. The eight named options are creative directions, not background-image uploads; the preview cards help you choose a tone, while generation uses the option name and text.
03 · Generation status
The site reserves 12 credits for 480p or 22 credits for 720p when the generation request is accepted.
An internal character reference is prepared from your photo. It is not shown as a separate public result.
The job moves through preparation, rendering, validation and upload. Progress can resume after navigation.
When complete, preview the private video, download the MP4 or delete the result from your account.
Known failed or blocked jobs release the reserved credits automatically. An ambiguous provider submission may enter manual review instead of being submitted again, which prevents accidental duplicate charges.
Check the quality you want and compare the current one-time packs. Purchased credits do not expire, and larger packs lower the effective cost per HD video.
Compare video packs and credits04 · Real examples
These are project-owned input-to-output pairs from the current workflow. They show the useful expectation: the source photo guides the lead subject, while the room choice changes the generated setting. They do not promise pixel-perfect identity or identical motion on every render.
Grandma’s Living Room
The curlers and patterned clothing remain recognizable in the result.
Secret CEO Meeting
The formal outfit creates a visible contrast with the absurd burger stare.
Party After Everyone Left
The Animal route keeps the cat as the single lead subject.
05 · Troubleshooting
A second render cannot repair an ineligible or unreadable photo. Start with the exact symptom, check the input and only submit again when the previous job has a clear terminal state.
Confirm the file is a supported image up to 10 MB. If a HEIC/HEIF photo cannot be converted, export it as JPG and retry.
Use a sharper front-facing image with better light and fewer obstructions. Show more of the upper body or animal body, and remove heavy filters.
Room cards communicate a creative direction, not an exact background image. Choose a more specific named option, or use Auto when the subject should lead the decision.
Leave the existing job in place and return to its saved progress. Do not create repeated submissions while the provider result is uncertain.
Review the whole 14 seconds. If the selected subject is missing or clearly replaced, the Window Peeper sequence or reference audio is absent, or the file is corrupted, request a quality review within 48 hours.
06 · Questions
Short answers to the questions that matter before you upload, spend credits or share a result.
Choose Person or Animal, upload one clear everyday photo of a consenting adult or one cat or dog, select Auto or a room idea, choose 480p or 720p, confirm permission and generate. The finished result is a 14-second vertical MP4 with the recognizable building approach, stare and curtain-close sequence.
Use a sharp, evenly lit JPG, PNG, WebP or supported HEIC/HEIF photo up to 10 MB. Show one subject clearly. For a person, include the face and upper body without glasses, hats, hands or a phone blocking the face. For a cat or dog, show the body clearly.
No. The current workflow accepts one consenting adult or one cat or dog. Do not upload children, public figures, celebrities, group photos or an image you do not have permission to use.
The output is a 14-second 9:16 MP4. It follows three beats: the camera approaches a lit window, the subject eats and stares back, and the curtains close.
Standard 480p uses 12 credits and HD 720p uses 22 credits. Credits are reserved when you submit. A known failed generation releases the reserved credits automatically.
Use the separate Window Peeper template guide to find the current CapCut listing, verify the three story beats and follow the live media-slot instructions shown in the app.
Ready to make yours?
Choose one eligible photo, let Auto pick the room or add your own punchline with a named setting, then review the full result before posting.
