Baby Face Video Maker

Upload baby or pet photos to create a face-aligned growth timelapse. The tool auto-detects eyes in each photo and locks them to the same position — making every frame line up perfectly. Runs entirely in your browser: no upload, no watermark, no account.

1 Add photos

📷
Drag & drop photos here or click to browse
JPEG / PNG / WebP • any order (drag to reorder below)

2 Settings

80%
Higher = tighter eye lock. Lower = softer stabilization that keeps more of the original framing.

3 Create timelapse

Initializing…

How it works

Each photo goes through four steps entirely inside your browser — nothing is sent to any server.

Face detection MediaPipe Face Mesh locates 468 facial landmarks in each photo, pinpointing both eye centres to sub-pixel accuracy.
Eye alignment The tool calculates the midpoint between the eyes and the inter-eye distance. Each frame is scaled and translated so those measurements land on the same canvas position.
Frame rendering The aligned image is drawn onto a square canvas at your chosen resolution. Frames are collected in memory in sequence.
Video encoding The canvas captures a live stream at your chosen FPS. MediaRecorder encodes the frames into a video file you can download and share.
Why eye alignment? The human brain notices tiny position changes between frames. By anchoring the eyes to a fixed point in every frame, the timelapse looks smooth and professional — you see only the face changing, not jitter from hand-held camera wobble or different crops.

Frequently asked questions

Does it upload my photos to a server?
No. Every step — face detection, alignment, and video encoding — runs locally in your web browser using WebAssembly and JavaScript. Your photos never leave your device. This also means the tool works offline once the page has loaded and the AI model has been downloaded (about 3 MB, cached by the browser).
What if the face isn't detected in some photos?
If MediaPipe cannot find a face in a photo (e.g. the face is too small, the angle is extreme, or it is obscured), that frame is included without alignment — it is centred on the canvas instead. A warning appears next to the thumbnail. For best results use photos where the face is clearly visible and takes up at least a quarter of the image height. Frontal or near-frontal angles work best.
What video format does it produce?
The video is encoded by your browser's built-in MediaRecorder. Most desktop browsers (Chrome, Edge) produce WebM/VP9, which plays on all major platforms and can be opened directly or converted with any video editor. Safari on Mac and iOS produces MP4/H.264. The file is labelled accordingly on download. If you need a specific format, open the downloaded file in any video editor (iMovie, DaVinci Resolve, VLC, etc.) and export to your preferred format.
Can I reorder photos before making the video?
Yes. After adding photos, drag and drop the thumbnails into the order you want — the number badge on each thumbnail updates in real time to reflect the final sequence. This is how you arrange photos from oldest to newest (or newest to oldest for a reverse timelapse).
What does the alignment strength slider do?
At 100% the tool applies the full calculated translation and scale to lock the eyes perfectly in place. At lower values the correction is partially blended with the original framing — useful when photos have very different crops and full alignment would zoom in too aggressively. 70–85% is a good starting point for most baby photo collections.
How many photos can I use?
There is no hard limit — the tool processes photos one by one in the browser. In practice, 200–300 photos at 720 × 720 works well on modern laptops. If you have thousands of photos, consider grouping them (e.g. one per month) to keep the timelapse concise and the processing time short. Processing takes roughly 0.5–2 seconds per photo depending on image size and device speed.