How to cut a video without re-encoding

Most video tools cut a file by decoding it and encoding it again. That is slow, and every pass loses a little quality. There is a better way for the common case of splitting and trimming: copy the compressed data as it is. This guide explains how that works and what its one limitation means for your cuts.

What re-encoding costs you

A video file is compressed video. When an editor "exports" a trimmed copy, it typically decompresses every frame, then compresses them all again with its own settings. Three things happen: it takes a long time (roughly real-time or slower on a laptop, far longer for 4K); the result is a second-generation copy, so fine detail softens and dark areas get blockier; and the output depends on that tool's encoder and settings, which may not match your camera's bitrate or colour handling.

For a final edit with effects and colour work, re-encoding is unavoidable. For simply splitting a recording into pieces, it is pure loss.

What "lossless" cutting actually does

Inside an MP4 or MOV, the video is stored as a sequence of compressed packets, one per frame, next to the audio packets. A lossless cut opens the file, selects the packets between your two cut points, and writes them into a new file with fresh headers. Nothing is decoded and nothing is encoded. The output is bit-for-bit the same footage: same resolution, same frame rate, same bitrate, same audio. It is also why it takes seconds rather than minutes: the work is reading and writing, not computing.

The one catch: keyframes

Compressed video does not store every frame in full. It stores an occasional complete frame, called a keyframe (or I-frame), and describes the frames in between only as differences from the frames around them. A frame that is "the previous frame plus these changes" is meaningless without the previous frame.

So a clip that is copied rather than re-encoded can only begin at a keyframe. Phones and cameras usually place one every one to two seconds; screen recorders and streaming software often every two to ten seconds. If your cut lands between keyframes, something has to give:

• Start early at the keyframe before the cut. You keep everything you asked for plus a short lead-in. This is what ClipScissor does.
• Start late at the keyframe after the cut. You lose up to a keyframe interval of footage. Never a good default for someone cutting between spoken takes.
• Re-encode just the first few frames up to the next keyframe and copy the rest ("smart cut"). Frame-accurate, but the re-encoded section has to match the original encoder's settings precisely or players glitch at the join, which is fragile across the range of cameras people use.

The end of a clip is less constrained: a copied clip can end on almost any frame, give or take a frame where the video uses bidirectional prediction. In practice, ends land where you put them.

Getting an exact start anyway

If the lead-in matters, cut on a keyframe. In ClipScissor, drag a clip's start handle and it snaps to nearby keyframes, so you can see where the exact starts are and choose one. And since every clip is heading into an editor anyway, trimming a second of lead-in there is frame-accurate and free: the editor decodes the clip to display it, and only the frames you keep are ever re-encoded, once, at your final export.

Doing it in the browser

Packet copying does not need an encoder or heavy processing, so it runs happily inside a web browser using the standard media APIs. ClipScissor does exactly this: drop in an MP4 or MOV, press S wherever you want a cut, trim with the handles, and export. The file never leaves your computer, there is no upload wait, and a 20-minute recording exports in a few seconds. Command-line users can do the same with ffmpeg -ss … -to … -c copy, one clip at a time; ClipScissor is the same operation with a timeline, names and a ZIP.

When you do want to re-encode

Re-encode when you are changing the picture: resizing, colour grading, adding text or transitions, converting to a different codec, or when a frame-exact start is worth more than the original quality. That is the job of CapCut, Premiere, Resolve or Final Cut. Splitting is the job before that, and it should not cost you anything.

© 2026 ClipScissor. Your video never leaves your device. · Privacy policy