Skip to content
Neattrim

Lossless cuts vs re-encoded cuts: when each is right

· Editorial team

Do it now
Trim video
Cut the start/end of a video without re-encoding.

A lossless cut copies the existing video data without decoding it, so it is fast and keeps the original quality, but it can only start at a keyframe. A re-encoded cut decodes and rebuilds the video, so it can cut on any frame, at the cost of time and a small quality loss. Choose lossless unless you need frame-exact edges.

What a keyframe is

Video is not stored as a pile of full pictures. A codec saves a complete frame now and then, called a keyframe, and describes the frames in between as changes from their neighbours. Phones commonly place a keyframe every second or two, and some cameras and screen recorders leave much longer gaps.

You cannot start playing from the middle of a run of changes, because the full picture it depends on is earlier in the file. That is the whole reason lossless cutting has limits.

How a lossless cut works

The tool picks the part of the file between two points and copies those bytes into a new file. Nothing is decoded and nothing is encoded. The result is:

Very fast, often seconds for a long video, because it only copies data.

Identical in quality to the source, since the picture data is not touched.

Limited by keyframes, so the start of the new clip snaps to a keyframe near the point you chose. A typical result is a clip that starts a fraction of a second earlier than you asked, or a little later, depending on the tool.

Because the data is copied, the audio, the colour information and extra tracks usually stay the same too.

How a re-encoded cut works

The tool decodes the section you want and encodes it again, with a new keyframe exactly where you cut. The result is:

Frame exact, so the clip begins and ends on the frame you chose.

Slower, since every frame is processed, and on a phone or in a browser the time and memory grow with the length.

Slightly lossy, because encoding always discards some detail. At a good setting you will not see it, but it adds up if you re-encode the same file repeatedly.

Flexible, because you can change the format, the size or the codec in the same step.

Which one to pick

Use a lossless cut when you are trimming the start and end of a recording, when a half second of slack does not matter, when the file is big, or when you want the quality to stay exactly as it was.

Use a re-encoded cut when the cut must land on an exact frame, for example the moment a word begins, when you need to remove a short piece from the middle and join the rest, when the source has very sparse keyframes, or when you also want a smaller or different file.

If you are unsure, try the fast cut first and play the result. If the edges are wrong, repeat with the accurate setting. Keep the original until you are happy.

A common surprise

You trim a clip, play it, and the first moment shows a frozen or blank picture for a short time with sound already playing. That happens when a cut starts between keyframes and the player has no full picture yet. It is the signature of a bad lossless cut, and a sign you should use the accurate setting or choose a start point nearer a keyframe.

In this site's trim tool

The trim tool has a fast setting, which copies the data without re-encoding, and an accurate setting, which re-encodes for exact cut points. The accurate setting has a lower size limit in the browser than the fast one, because re-encoding needs more memory and time. The tool page lists the current limits.

Frequently asked questions

Is a lossless cut really lossless?

For the picture and sound data, yes: the bytes are copied. The only change is where the clip begins and ends.

Why does my lossless cut start earlier than I asked?

The tool had to start at the keyframe before your chosen point, since it cannot begin in the middle of a run of changes.

Does re-encoding ruin the quality?

At a sensible setting the difference is hard to see on a single pass. Repeated re-encoding is what degrades video visibly.

Can I cut without a keyframe at the exact point and still be lossless?

Only partly. Some tools re-encode just the short stretch up to the next keyframe and copy the rest. That gives exact edges with little loss, but support varies.

Related guides