Stream Copy vs Re-encoding: What's the Difference?

When you trim a video or extract a clip, the output can arrive in one of two ways: almost instantly, or after several minutes of processing. The difference is whether the tool used stream copy or re-encoding. Understanding the distinction helps you choose the right tool and set the right expectations.

What is stream copy?

A video file is a container (like MP4 or MKV) holding encoded video and audio streams. Stream copy means copying those streams directly from the source file into a new file, without touching the encoding. The codec data is unchanged — you're essentially doing a binary copy of the relevant portions of the file.

Because no decoding or encoding happens, stream copy is extremely fast (limited only by disk speed) and produces output that is bit-for-bit identical to the source within the copied range. There is zero quality loss.

What is re-encoding?

Re-encoding means decoding the video from its compressed form back to raw pixel data, performing whatever operation you need (trim, resize, filter), and then encoding it again into the output format. This is the process that's used for compression, resizing, format conversion, adding effects, and any operation that needs to touch the actual frame content.

Re-encoding is slower because it involves two complete passes over all the data. It also introduces a generation of quality loss — even at high quality settings, each encode cycle reduces detail slightly. For most practical purposes this is imperceptible, but it's technically present.

When stream copy works

• Trimming a video to a start and end time (with keyframe precision).
• Extracting a portion of a video into a new file.
• Looping a video (concatenating copies of the same file).
• Merging videos with matching encoding parameters.
• Remuxing — changing the container format without changing the streams (e.g. MKV to MP4).

When re-encoding is unavoidable

• Changing the video codec (e.g. H.265 to H.264).
• Changing the resolution or aspect ratio.
• Adding filters, text, watermarks, or effects.
• Changing the frame rate.
• Frame-accurate cuts (stream copy can only cut at keyframe boundaries).
• Changing the audio format or sample rate.

The keyframe caveat

Stream copy's one limitation is keyframe alignment. H.264 video doesn't store every frame independently — it stores keyframes (complete frames) and intermediate frames that reference keyframes. When stream copy cuts the stream, it must start and end at a keyframe. If your requested start time falls between keyframes, the actual cut will snap to the nearest preceding keyframe, which might be 0.5–2 seconds earlier than what you specified. For most use cases — cutting down a long recording, removing a dead intro — this offset is unnoticeable.