Technology, made understandable.Our approach
Software & workThe keyframe boundary decision

Trim a video without needlessly re encoding it

Keep a video excerpt with LosslessCut on Windows 11, understand keyframe limits, and check the exported picture and sound before deciding to re-encode.

A recording has two minutes of setup, the useful demonstration, then a long goodbye. Trimming those ends should not automatically mean compressing every remaining frame again. The useful question is whether you can accept a nearby clean starting point, or need one exact frame.

This guide uses LosslessCut on Windows 11 to keep one continuous excerpt. Start with an ordinary MP4 recording that already plays correctly. Keep the original until the exported clip passes inspection; the instructions below are a workflow to follow, not a claim that your recording has been tested.

Keep the recording’s compressed pictures

FFmpeg calls the underlying operation streamcopy: compressed packets move into an output file without decoding and encoding them again. That avoids another generation of compression loss. It does not make the whole output file identical to the source: its duration, container structure and possibly metadata change.

Re-encoding becomes appropriate when you need to resize pictures, burn in text, change the video codec or otherwise alter the actual image. Simply wanting fewer minutes is a reason to try copying first. A smaller MP4 export from a conventional editor may involve compression even when its resolution stays unchanged.

Separate two delivery requests before opening an editor: “keep the demonstration at its existing quality” and “make an attachment below a size limit.” A lossless trim addresses duration. It cannot promise a particular attachment size. If the retained material remains too large, decide whether to share it as a file link or deliberately create a compressed delivery copy.

ARRI cinema camera with a large lens mounted in a handheld camera rig.
Photo: ShareGrid / Brent Barbano / Unsplash · Unsplash License. Resized for display; some views use a crop. Illustrative photograph.

Mark the part you want to keep

Get LosslessCut through the download route linked in its official getting-started guide. Make a local working folder with the source and an empty output subfolder. Choose a short trial excerpt before processing a long recording. Write down an unmistakable first event and last event, such as “hand reaches the switch” and “indicator goes dark.” These observations give you something better than a filename to check later.

  1. Open LosslessCut and press Ctrl+O to load the source. Use Space for playback. Confirm that you have the intended recording before marking it.
  2. Seek near the beginning of the useful material, pause, and press I to set the current segment’s start. Move to the intended end and press O. The segment should now have both boundaries.
  3. Check the segment list: for this task, keep one completed segment. By default segments identify material to retain. Do not use the yin-yang control to invert that meaning. A marker without an end is excluded from export.
  4. Use Working dir unset to choose your output subfolder. Keep Separate files as the export mode for this single excerpt.

These are default shortcuts. If a key behaves differently, open the shortcut reference with Shift+/ and check your keyboard layout or custom bindings. Before continuing, compare the selected span with your written first and last events. A reversed selection or accidentally omitted closing moment is easier to fix now.

Why the exported start can move backward

Many compressed recordings contain complete reference pictures, called keyframes, with other pictures depending on surrounding information. LosslessCut normally starts from the preceding keyframe. Its troubleshooting guide explains why an export can include material before the chosen start.

Hypothetically, your intended start is 00:12.500, with usable keyframes at 00:10.000 and 00:15.000. Starting at the earlier one keeps 2.5 extra seconds. Starting at the later one loses 2.5 seconds you wanted. This invented example explains the decision; it does not predict your recording’s keyframe spacing.

Zoom into the timeline to see keyframe lines. Use Alt+Left Arrow or Alt+Right Arrow to seek between keyframes, then press I at a suitable start. Leave Keyframe cut enabled for the first attempt. Exported playback is the final check.

Choose the compromise by content. Two seconds of someone walking into position may be harmless; two seconds showing a private message is unacceptable. If neither adjacent boundary preserves exactly the material you may share, use an exact-cut workflow that permits re-encoding. A fast export is no advantage if its opening reveals something you meant to remove.

Export once, then inspect the delivered file

Open Tracks to inspect included streams; keep the intended video and audio. LosslessCut’s track documentation distinguishes parallel streams from timeline segments. Multiple audio tracks may include different microphones or languages. For a first export, preserve required streams and leave optional cleanup until the basic cut works.

Keep the source container initially. Press Export or E to review export options, then Export again to confirm. Leave Smart cut off for this lossless attempt. Afterward, open the newly written file from the output folder in the player you intend to use for delivery.

  • Watch from the very beginning. Check for unwanted lead-in, black frames, frozen pictures and a missing first word.
  • Listen at a visible speech or impact event near the beginning and again near the end. Look for synchronization drift rather than checking only that sound exists.
  • Seek through the clip, then play its final seconds. Confirm the closing event is present and excluded material has not returned.
  • Check needed audio or subtitles, orientation and playback in the recipient’s intended app. Passing on your editing computer alone is insufficient.

For an illustrative size estimate, a constant combined stream rate of 20 megabits per second over 30 seconds yields about 75 decimal megabytes before container overhead: 20 × 30 ÷ 8. Real recordings often vary in bitrate. File size is therefore a rough sanity check, not proof of lossless processing or accurate boundaries.

When to retry, and when to re-encode

If the start lands one keyframe too early, try moving the start a few frames forward and export a fresh candidate. If output freezes or has the wrong duration, try a diagnostic export with only the main video track, then restore needed tracks individually. The official troubleshooting guide also suggests another container, such as MOV or Matroska. Recheck playback after every change.

A container change is a remux, not a codec conversion. It can resolve packaging problems but cannot make an unsupported codec universally playable. Nor does “lossless” promise every camera metadata field survives. Keep camera originals if capture information matters.

For an exact start, Smart cut (experimental) in export options re-encodes the portion up to the next keyframe; it does not work reliably for every file. Treat it as a separate trial. If it fails, use a conventional editor for a deliberate re-encoded delivery copy and retain the source.

Use Ctrl+Z to undo a segment edit and Ctrl+Shift+Z to redo it. Those actions do not recall a shared export or restore a deleted original. Name successive candidates clearly, compare them with your acceptance notes, and share only the checked candidate. Archive the original separately before cleaning up the working folder.

Sources, versions & scope

The steps describe the documented controls and expected checks; they are not a claim that every device or software release has been physically tested. Platform and version differences are stated with the procedure.

  1. LosslessCut: Getting started
  2. LosslessCut: Documentation — containers, segments and tracks
  3. LosslessCut: Troubleshooting, known issues and limitation
  4. FFmpeg: Streamcopy and transcoding
Share this guide