Experiment

Why Do Two Cuts of the Same Three Seconds Report Different Durations?

The same 2.35–5.35-second toy-car interval was saved by stream copy and accurate re-encoding. The starting scenes look similar, but reported file durations differ. Inspect image quality separately from timing metadata.

Check the starting scene separately from the duration retained in a cut. Stream copy reuses encoded video, while accurate trimming re-encodes it, so the methods handle boundaries differently. Compare the moving car in both outputs alongside their reported timing metadata.

Similar beginnings can still have different timing metadata

The requested interval was three seconds

Reported duration: accurate 00:00:03.00, stream copy 00:00:03.19.

Check playback separately from image quality

Stream copy does not always begin on an obviously wrong scene. Pixel differences after re-encoding do not establish a timing offset.

Inspect the actual beginning and end of both files

The initial stills show similar car positions. This is not a claim that stream copy always visibly starts wrong: inspect actual playback alongside timing information.

Stream-copy comparison
Video details987,417 B · 00:00:03.19
Production accurate trim
Video details372,380 B · 00:00:03.00

Requested range and reported duration

Frames, audio and timestamp handling can affect reported duration. A duration difference alone does not establish a count of wrongly cut frames. Verify the needed scene in the destination player.

MethodReported durationBytes
Stream copy00:00:03.19987,417 B
Accurate00:00:03.00372,380 B

Preserving encoded pixels and controlling boundaries are different goals

Stream copy reuses encoded video data and avoids another video-encoding generation, but depends on keyframes and timestamp handling. Accurate trimming re-encodes frames, so pixel values can change.

Output bytes were not treated as a quality ranking. Inspect needed details such as the car outline and wheels at the same display size, and keep the original.

Reopen the saved result when an action or spoken boundary matters

Check that the needed action is present at the beginning and not cut off at the end. The entered numbers alone do not validate the delivered result.

Toolit currently uses accurate re-encoding. For another revision, choose a fresh range from the original rather than repeatedly re-encoding an existing trimmed copy.

Test conditions and reproduction

The same toy-car interval saved with two methods

  1. 1

    The selected interval is 2.35–5.35 seconds of an eight-second 1920×1080, 24fps generated sample. The car follows the blue line rather than crossing it.

  2. 2

    The production accurate-trim command re-encodes at H.264 CRF 23. The comparison uses a separate FFmpeg -c copy command; the current tool does not expose a stream-copy option.

  3. 3

    Both are actual local FFmpeg WASM 0.12.10 outputs with the source audio track. Durations below are file metadata read by FFmpeg, not new browser playback, speed or perceptual-quality benchmarks.

Raw reproducibility records (JSON)

The main results are explained above. These raw records contain processing conditions and file hashes. Opening a link displays JSON text, not an illustrated article.

Official documentation used for this analysis

Check timing metadata and the actual scene together

The accurate car output reports the requested three seconds; stream copy reports 3.19 seconds. Similar initial scenes do not prove the methods always match, and bytes alone do not establish quality. Inspect the final beginning, ending and audio.