AetherCut™ auto-reframe
Auto-convert horizontal video to TikTok/Reels/Shorts vertical with face-and-motion-aware crop. 100% on-device.
AetherCut's auto-reframe tool repurposes a 16:9 horizontal video for vertical platforms (TikTok, Reels, Shorts) by automatically tracking the subject and re-centering the crop frame-by-frame. The face detection runs locally via TensorFlow.js, the motion analysis runs on your CPU, and the crop solver picks a smooth path that keeps the subject in frame without the crop window jittering between detections.
Every step runs in your browser. The output is the original clip composited into the target aspect ratio (9:16, 1:1, or 4:5) with the same color and quality. No frame is uploaded.
How the reframe pipeline works
Stage 1: face detection on each frame via TensorFlow.js's BlazeFace model. The model is small (~1 MB), runs in WebAssembly, and processes a 1080p frame in roughly 30ms on a 2024 laptop.
Stage 2: if no face is detected (B-roll, wide shots), fall back to motion analysis — the crop tracks whatever's moving most in the frame.
Stage 3: a smoothing solver picks a continuous crop path that keeps the subject roughly centered without snapping between detections. You can adjust the smoothness slider if you prefer tighter tracking (more jitter, less subject loss) or looser tracking (smoother, occasional subject drift).
Aspect ratios supported
9:16 vertical — for TikTok, Reels, YouTube Shorts, Snapchat. The most common reframe target.
1:1 square — for Instagram feed, LinkedIn feed posts.
4:5 portrait — for Instagram feed (the taller of the two acceptable feed ratios).
Custom — set any aspect ratio with a numeric input. Useful for less common platforms or experimental cuts.
When auto-reframe falls short
Multi-person scenes where the speaker changes — the tracker will follow the most-detected face, which may not always be the current speaker. You can manually override the tracked subject with a click at the moment the speaker changes; AetherCut respects the manual hint through the next detection cycle.
Very wide group shots cropped into 9:16 — there's a hard mathematical limit on how much horizontal information you can lose. If the speaker isn't physically near the center of the frame and there's no room to crop without losing them, the result will compromise. The fix is either a different shot or a manual zoom in the editor.
Frequently asked questions
Does auto-reframe upload my video?
No. Face detection (TensorFlow.js BlazeFace) and motion analysis both run in your browser. The crop is applied at export time, also on-device.
Can I edit the crop path after auto-reframe runs?
Yes. The crop path is exposed as a keyframe-editable track. You can grab any keyframe, drag it to override the auto-solve, and re-export.
What about One-Tap Shorts?
One-Tap Shorts (Pro) is a higher-level tool that runs auto-reframe + razor cuts + caption generation in one pipeline to turn a long video into multiple ready-to-publish shorts. Auto-reframe is the core step inside that pipeline.
Try AetherCut now — no signup required.
Open the editor and verify the no-upload claim yourself in Chrome DevTools.
Free tier · Pro $14.95/mo · $129.88/yr · Lifetime Pro one-time
Other comparisons + resources
Related searches
AetherCut answers for all of these search intents — pick the one closest to what you're looking for:
No signup. No upload. Verify privacy in DevTools in 30 seconds.