h a l f b a k e r yBreakfast of runners-up.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
I don't know where this belongs.
Video file formats seem to have the key frames scattered throughout. If key frames are missing, much of the useful information for that section of the video is missing too, and the file is either irreparable or will be very short despite almost all the information
being there. This is bad when a file is corrupted where a file is not entirely downloaded yet, or when the file is stored on a slow medium such as an optical disc.
If the key frames were stored at the beginning of the file, it would at least be possible to reconstruct more of the file from a partial download, and to scan or search through a file quickly by storing it in RAM somewhere, either on computer or some other device, before it was completely accessible to the device handling it. It would also be easier to salvage a knackered file, as more recognisable information would still be there, and on a dodgy DVD player or DVD, where the edge of the disc seems to be more problematic, the file would still be watchable. However, i would also propose that the key frames be sorted into a more optimised order once the video file is available.
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Annotation:
|
| |
It seems obvious, even if for playback purposes, *duplicates* of the keyframes were placed at the beginning. I would be amazed to learn that none of the video formats in use work this way... cause it makes sense. [positive bunnification] |
|
| |
Don't DVD's do this ? "chapters". |
|
| |
One could store the keyframes at the beginning, and then when the video is to be played, copy it into memory/temporary file, putting them in the order they're needed. |
|
| |
Yes, [Ian], i agree that it would, and i said it would be sorted prior to playback. However, it wouldn't make a lot of difference if the entire file were in RAM at once, so for smaller files it would still make sense to leave them at the start, and in fact playback could start while it was still loading if this was done. |
|
| |
I don't know much about video files, but i'm almost sure this isn't done now because an almost complete video file is still very short even if repaired, and there clearly seems to be an issue with some DVDs when you get towards the end of a film, which i've always thought was to do with it being towards the edge of the disc. If they were on a single file, the key frames would be near the centre and the film would speed up rather than grind to a halt or get all artefacty if they were arranged that way, wouldn't it? (Sorry, genuine question that, not sarcasm). |
|
| |
I might not understand the idea
perfectly.. but for streaming, it would
massively delay the first second of
video...? |
|
| |
And for download... what makes the
beginning of the file a safer location ?
Any disk sector error at that place, and
an entire part of the movie is screwed. |
|
| |
The idea of duplicating the keyframes
may make some sense, although it
might increase the filesize a lot. |
|
| |
Actually, I see that idea as a
consequence of the frustration of
partial movie downloads ;/ Wasn't that
issue made obsolete by multithreaded
p2p downloads ? |
|
| |
It wouldn't be possible to play the video immediately, but it would be possible to fast forward through the video to see if it was worth watching, to start playing it before it was completely loaded and to repair a corrupted file more easily. |
|
| |
P2P multithreading doesn't make it obsolete. There seems to be a tendency for earlier parts of video files to arrive first because they are the first to be uploaded. Also, there are other reasons to do this than downloading, for instance problems at the edge of a disc or being able to repair a corrupted file without missing out huge chunks, which is what seems to happen at the moment. |
|
| |
The beginning of a file would be safer on an optical disc because if the whole disc only contains one file (which it doesn't but could), the early frames would be near the centre, where loading into memory would be quicker and where fingerprints and scratches are less likely. Incidentally, i harbour a deep, burning resentment towards Tomorrow's World on this issue. |
|
| |
[+] because it sounds good and I have no idea how the technology works. |
|
| |
//Tomorrow's World// Is that when they demonstrated that CDs aren't affected by scratching and Jam? |
|
| |
By the way, I think that pretty soon, the needle-scratching-over- the-record sound effect will be unrecognisable by most generations. |
|
| |
Yes it was, [Ling]. Every time i see a CD down the back of a sofa or whatever, i blame Michael Rod or whomever it was. |
|
| |
[Ian], is there any parallel to the audiophile thing with photography? What i mean is, what i see as the dubious promotion of expensive control knobs, brackets and oxygen-free cables (yes, i know the last does make a difference) but applied to photography. |
|
| |
Thanks, that's really interesting. A couple of years ago, some photos of us were printed in a women's magazine. They were taken using a digital camera which cost five grand, and the colour of my irises looked totally wrong, whereas my partner's was fine. The photographer said this was because it was Japanese and calibrated for oriental eye colour, hence Liz's eyes looked normal. Is that feasible or just an excuse? |
|
| |
What i'm saying is that i can't imagine them looking like that on traditional film, but presumably people can fiddle with that too. |
|
| |