A video cropper alternative that decodes, crops, and encodes inside your browser without uploading the source is built around three concrete guarantees: pixel-exact coordinates, an entirely local pipeline, and a finite-duration WebM download. Unlike cloud editors that ask for an account, watermark the result, or hand you a preset-sized canvas, a coordinate-based cropper takes four integers (X, Y, width, height) and applies the same rectangle to every frame of the original. The output is the exact sub-region of the source — not a re-sized, re-compressed, or aspect-mangled version. For anyone comparing options in 2026, the meaningful differences come down to where the bytes travel, how the rectangle is defined, what container is handed back at the end, and whether audio and finite duration survive the round trip. A tool that keeps the file on disk, asks for whole-pixel math, and patches a real Matroska Duration into the output removes most of the guesswork that preset-based alternatives leave behind.
Why People Look for a Video Cropper Alternative
Most of the cropping tools that surface in a typical search have at least one of these friction points: a required sign-up, a watermark on the saved file, a preset-sized canvas instead of a free rectangle, a server upload of the full source, or an output that lacks a real container duration. Each of those friction points turns a one-minute job into a five-minute job. A tool that is meant to serve as a true video cropper alternative should remove at least the upload, the watermark, the account, and the preset — and ideally replace them with a coordinate-driven workflow that produces a clean, finite-duration WebM the user can review locally before deleting the original.
Switching decisions are rarely about features alone. They are about where the file goes, what is appended to it, and how the result behaves in the destination player. If the file leaves the device, the user is trusting the service with the full bitstream. If the result has no finite duration, it can confuse downstream tools that read container length. If the crop is locked to a preset, the user has to either accept an aspect ratio they did not want or fall back to a different tool. These three constraints — privacy, container integrity, and rectangle control — are what most alternatives are being measured on.
How This Browser Cropper Works Differently
The Video Cropper keeps the entire pipeline inside the current browser tab. The source file is read locally; its decoded dimensions are read from the browser's media stack; a Canvas element is sized to the requested crop rectangle; only the requested source rectangle is drawn into that Canvas at 30 frames per second; and a MediaRecorder writes a WebM from that Canvas stream. No part of this is uploaded, and the source is not re-encoded in the cloud. When the browser exposes a capturable audio track through media capture, the audio is appended to the Canvas stream automatically.
Where this tool visibly differs from preset-based alternatives is in how the rectangle is defined. The user enters four untrimmed whole-pixel values: X and Y offsets from the top-left corner of the source frame, plus width and height in source pixels. There is no aspect-ratio lock, no canvas resize, no drag handles that snap to 16:9 or 9:16. The same fixed rectangle is applied to every frame; the crop does not move, track a subject, or change with time. If the requested rectangle exceeds the source bounds, or if any dimension is below two pixels, or if the output area would exceed 3840 × 2160, the crop fails before any recording starts and no partial file is produced.
A second visible difference is the container finish. Browser-based recorders commonly emit a stream without a finite container duration, which can break tools that read the Matroska length. After recording, the tool inserts a real Matroska Duration value, expressed in timestamp-scale ticks, so the resulting WebM reports a real length in standards-tested players. The output is empty-checked and dimension-checked before download, which is why a one-minute source takes roughly one minute to process and produces one playable file rather than an indeterminate stream.
Crop a Video Locally in Three Steps
- Choose a browser-decodable video and wait for its source dimensions to load. Pick an MP4, WebM, MOV, M4V, or Ogg file from your device. The page reads the decoded dimensions only after the browser has finished parsing the file; you will see the source width and source height appear before the crop fields become usable. An accepted extension is not a guarantee of codec support — decoding depends on the current browser.
- Enter whole-pixel X, Y, width, and height values that stay inside every source frame. X is the right offset from the source's top-left corner, Y is the down offset, width is the horizontal extent, and height is the vertical extent. Every value must be a nonnegative whole-pixel count, width and height must each be at least 2, and X + width must be less than or equal to the source width while Y + height must be less than or equal to the source height. The output area must also stay at or below 3840 × 2160.
- Select Crop video, keep the tab open during real-time processing, then download and fully review the WebM. Processing happens at roughly the source's playback speed. A one-minute clip takes about one minute, a five-minute clip takes about five minutes, and the tab must stay open the entire time. When the WebM is offered for download, open it in your destination player, confirm the rectangle, the duration, and the audio, and only then delete the original.
Input Limits and Output Boundaries
The cropper applies the same input safety budget that the audited video pipeline already uses elsewhere. Review these limits before picking a source file, because invalid rectangles fail without producing a partial file.
| Constraint | Limit |
|---|---|
| Accepted input extensions | MP4, WebM, MOV, M4V, Ogg (codec support depends on the browser) |
| Maximum input file size | 500 MiB |
| Maximum input duration | 5 minutes (decoded duration must be above zero) |
| Maximum source side | 4096 pixels on either width or height |
| Maximum source area | 3840 × 2160 pixels |
| Maximum output area | 3840 × 2160 pixels |
| Minimum crop dimension | 2 pixels on both width and height |
| Output container | WebM with finite Matroska Duration |
| Output codec | First browser-reported WebM recorder among VP9, VP8, generic WebM |
| Output bitrate | 180 kbps to 8 Mbps, derived from cropped pixel area at 30 fps |
The bitrate is a single-pass browser setting, not a guaranteed target size, so fast motion, noise, grain, the source codec, and the browser's encoder can all shift the final size and visual quality. Keep the original until you have reviewed the full output.
Coordinate Math for a 9:16 Crop From a 1920 × 1080 Source
The cleanest way to think about cropping is to treat the source rectangle and the target rectangle as pixel math, not as a drag handle. Suppose the source is a 1920 × 1080 landscape clip and the target is a 9:16 portrait region that fits inside it without scaling. Because 9 is smaller than 16, the limiting dimension is the source height: the crop height cannot exceed 1080 pixels. The crop width then has to satisfy width / height = 9 / 16, so width = 1080 × 9 / 16 = 607.5 pixels. Widths must be whole pixels, so round 607.5 up to 608.
Centering the rectangle horizontally means X = (1920 − 608) / 2 = 656. Using Y = 0 keeps the rectangle flush with the top edge. The full rectangle is X = 656, Y = 0, width = 608, height = 1080. Verify the bounds: X + width = 656 + 608 = 1264, which is less than or equal to the source width of 1920, and Y + height = 0 + 1080 = 1080, which equals the source height. The output area is 608 × 1080 = 656,640 pixels, well under the 3840 × 2160 ceiling of 8,294,400 pixels. Enter those four values into the cropper, select Crop video, and a 9:16 portrait WebM is produced from the original landscape source without any resizing.
For other aspect targets the same approach works in reverse: pick the dimension that limits the rectangle inside the original frame, solve the ratio for the other dimension, round to a whole pixel, then position X and Y so the rectangle stays inside the source.
What This Tool Is Not Designed to Replace
A pixel-exact cropper is one tool in a small toolbox, not the whole toolbox. Several adjacent tasks look like cropping but are actually different operations, and forcing them through this tool will produce the wrong result.
- Resize. Changing the frame size of every pixel (for example, 1920 × 1080 down to 1280 × 720) is a rescale, not a crop. The output of this tool is the exact sub-region of the source, drawn at source pixel density.
- Trim. Cutting the start and end timestamps of a clip is a time-based operation. The cropper applies one rectangle to every frame of the full decoded duration; it does not shorten the clip.
- Compress only. Re-encoding a file at a lower bitrate without changing the visible frame region is handled by a re-encoder. The cropper's bitrate is derived from the cropped area and is bounded from 180 kbps through 8 Mbps.
- Redact, blur, or object removal. The tool does not track a moving subject, blur a region, or remove an element. It applies one fixed rectangle to every frame.
- Container-feature preservation. Subtitles, chapters, attachments, rotation flags, color metadata, and HDR signaling are not carried into the WebM. Audio is included only when the browser exposes a capturable track through media capture, so verify both sound and picture in the destination player.
For the jobs in the list above, the right move is to use the matching tool — resize, trim, or compress — and reserve the cropper for the exact task it was designed for: selecting a fixed rectangular sub-region of a local video and downloading a finite-duration WebM without uploading anything.
If you want a deeper walk-through of the pixel-accuracy workflow in this browser-based approach, the pixel-accuracy guide covers the same coordinate math in more detail. The Canvas drawImage behavior this tool relies on to draw the fixed source rectangle into the output Canvas is described in the WHATWG HTML specification.