Two ways to divide a video, and they answer different questions
The mode selector is the only decision that really matters here, and the right choice depends on which constraint you are working against.
Equal parts is the mode for a fixed number of pieces. You know you want the recording in four chunks and you do not much care how long each one is. The tool places boundaries at even fractions of the duration, so every part is the same length and the last one ends exactly at the end of the file.
Fixed-length segments is the mode for a hard time limit. A platform accepts 60 second clips, or you want a two hour recording in 15 minute chapters. Here the length is fixed and the count falls out of it, which means the final part is whatever is left over.
Splitting a file
- Drop an MP4, MOV, M4V or WEBM onto the box. One file at a time.
- Pick a Split mode.
- Set Number of equal parts for the first mode, or Segment length in seconds, for fixed-length mode for the second. The field belonging to the other mode is ignored.
- Run the split. The file is probed for its duration first, then each part is written in turn with a progress bar that advances part by part.
- Download the zip, or the single video file if the split produced only one part.
Why every part is a stream copy
There are two ways to cut a video. You can decode it, throw away the frames outside the range and re-encode what is left, which gives frame exact cuts and costs you an encode pass plus a generation of quality. Or you can copy the compressed data through and only rewrite the container, which is nearly instant and lossless but can only begin a part at a keyframe.
This tool takes the second route without offering the first. For splitting, that is the right trade almost every time. A split is usually a packaging operation on footage you intend to keep as it is, and paying an encode for every part in order to move a cut by a fraction of a second is a bad bargain. If you do need a frame exact boundary, cut the single range you care about with Trim Video instead, which uses the same copy strategy but lets you place both ends yourself.
For MP4, M4V and MOV outputs the tool also relocates the index to the front of each part, so a part starts playing before it has fully downloaded. WEBM does not need that treatment and does not get it.
The last part gets special treatment
Every part except the final one is given an explicit length. The last one is not, and that is deliberate rather than an oversight.
If the last part were given a fixed length, and its start had snapped backwards to an earlier keyframe as stream copies do, the fixed length would run out before the true end of the file and the tail would be clipped off. Letting the final part simply copy through to the end of the source guarantees nothing is lost, at the cost of that part sometimes running a moment longer than planned. Losing the end of a recording is a much worse failure than a slightly long final part.
One download, not a folder
Multiple parts always come back as a single zip using stored entries rather than compression, because video data is already compressed and running deflate over it would burn time to save almost nothing. Handing back one file rather than a stack of individual downloads also avoids the browser prompting repeatedly, which is what happens when a page tries to save a dozen files at once.
Related video tools
To rejoin pieces rather than break them apart, Merge Videos is the counterpart. To make the whole file smaller instead of shorter, Compress Video re-encodes an MP4 to one of four size targets. If your goal is an animation rather than a clip, MP4 to GIF converts a whole MP4 at a frame rate and width you choose, so trim it down first. Extract Video Frames pulls out stills instead. Everything else is on the video hub.