Turn a horizontal 16:9 video into a vertical 9:16 clip for Shorts, Reels or TikTok where the crop window moves with the speaker. Face detection or motion tracking finds the subject, a smooth virtual camera follows it, and you can pin the framing yourself wherever you like before exporting an MP4 in your browser.
Last updated:
Drop a horizontal video hereMP4, MOV or WebM. It is processed on your device and never uploaded.
0:00 / 0:00
Crop position over timedetected camera path keyframe
Drag the box on the video to override the automatic framing.
Advertisement
Advertisement
How to use the Auto Reframe Video
Choose a video (or press Try a sample clip). It opens locally and the analysis starts automatically.
Pick what to follow: Faces for talking-head and interview videos, Moving subject for sports, pets or products, or keep centered.
Check the framing: press Play and watch the vertical preview. Change Smoothness and Hold still zone to make the camera calmer or quicker.
Fix any moment by dragging the crop box (this pins a keyframe), and press Follow subject from here to return to automatic framing.
Press Export vertical video, wait while it plays through once, then download the MP4 or WebM.
How much of the picture is kept?
When you reframe, the crop keeps the full height of the original and cuts the width to match the new shape. For a 16:9 source the crop width is height × target aspect ratio. Zoom shrinks the crop further, which gives the virtual camera room to move up and down too.
Source
Output shape
Crop from source
Share of width kept
Export size
1920×1080 (16:9)
9:16 vertical
608×1080
31.6%
1080×1920
1920×1080 (16:9)
4:5 portrait
864×1080
45%
1080×1350
1920×1080 (16:9)
1:1 square
1080×1080
56.3%
1080×1080
3840×2160 (4K)
9:16 vertical
1215×2160
31.6%
1080×1920
1280×720 (720p)
9:16 vertical
405×720
31.6%
720×1280 recommended
A 9:16 crop of 1080p footage is only 608 pixels wide, so exporting at 1080×1920 scales it up about 1.8×. Footage shot in 4K reframes much more sharply.
Settings for common videos
Podcast or interview with one speaker: Follow Faces, Smoothness 60–80, Hold still zone 15–25%. The frame stays calm and only moves when the speaker leans away.
Two people talking: Follow Faces, then pin the crop on whoever speaks at each change, or zoom out to 1:1 or 4:5 so both fit.
Presenter walking on stage: Follow Faces, Smoothness 40–60, Hold still zone 10%.
Sports, pets, dancing: Follow Moving subject, Smoothness 30–50, Hold still zone 5–10%.
Static shot: Keep centered or pin one keyframe at the start; no tracking needed.
Keyframes
Dragging the box or pressing Pin crop here adds an orange keyframe: from that moment the crop stays exactly where you put it. Follow subject from here adds a blue keyframe that hands control back to automatic tracking. Changes between keyframes are eased over 0.4 seconds so cuts do not look abrupt. Click a keyframe in the list to jump to it, or ✕ to remove it.
Frequently asked questions
What does auto reframe mean?
Auto reframe crops a wide video to a narrower shape, such as 9:16 for phones, and moves the crop window over time so the important subject stays in the picture. Instead of a fixed center crop that cuts off a speaker who walks to the side, the frame pans to follow them, like a camera operator would.
How does the face tracking work?
The video is sampled several times per second and each frame is checked with MediaPipe's BlazeFace face detector running in your browser. The largest, most consistent face is chosen as the subject. Its positions are then smoothed with a hold-still zone and a two-way exponential filter, so the virtual camera glides instead of jumping. Without faces, motion tracking follows the area of the picture that changes most between frames.
How do I stop the crop from jittering?
Raise Smoothness so the camera reacts more slowly, and widen the Hold still zone, which lets the subject move a little inside the frame before the camera follows. For a perfectly still shot, drag the box to pin the crop at the start of that section.
Is my video uploaded?
No. The video is opened, analyzed and re-encoded by your browser on your own device. Only the face detection program and model are downloaded from a public CDN the first time; your video never leaves your computer.
Why does exporting take as long as the video?
The vertical video is recorded in real time while the original plays in the background, using your browser's MediaRecorder. A 2-minute clip takes about 2 minutes. Keep the tab visible while it exports; background tabs can be slowed down by the browser.
What output size and format do I get?
1080×1920 or 720×1280 for 9:16 (and matching sizes for 4:5 and 1:1), at about 8 or 5 Mbit/s. Chrome, Edge and Safari can save MP4 (H.264); Firefox saves WebM. If the original is lower resolution than the crop needs, the picture is scaled up and may look softer.
Should I crop or use a blurred background instead?
Auto reframe fills the whole vertical frame and keeps the subject large, which works best for one or two people. For wide scenes, slides or screen recordings where you need to see everything, the Landscape to Vertical Video tool keeps the full picture over a blurred background.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.