The demuxer does not read this file's container.
The container opened and the demuxer parsed no track out of it.
The container's tracks read and none of them carries video.
A video track the demuxer cannot name a codec for, so no decoder can be configured for it whatever the browser supports.
The codec is named and this browser has no decoder for it.
Frames do not arrive on a steady grid: two share a presentation timestamp,
or the gaps between them take more values than a timebase's own rounding
explains. Frame indices reconstructed as round(time * rate) name the
wrong frame on such a source.
The frame index sits after the media data, so opening over a link pays a seek to the end of the file before the first frame.
The file carries one index per fragment rather than one for the whole.
The first frame's presentation timestamp is not zero, so media time and elapsed time are different numbers on this source.
A fact about a source that conversion could be a response to.