As I understand its a subset of the MP4 container but without the limitations of supported formats (IE you can put non-mpeg stuff in mov). Are there major technical obsticals that have prevented its use? Is it not open enough?

I'm just wondering why its hardly used.

IIRC, the MPEG-4 standard is based on some QuickTime technologies. However, the MOV container is not a subset of the MPEG-4 standards as it is a proprietary container used exclusively by Apple and its QuickTime content.

Also there is an almost complete lack of implementations except of the one provided be Apple (QuickTime), since - unlike MP4 - it's not a standard of any kind.

dev0

isnt mpeg 4 based on the .mov wrapper?

from the mpeg4 standard:


and what do you mean by lack of implementation? i can export to .mov files, so it seems to have been implemented.


it seems to me that .mov is capable of doing everything the new formats can do, but is just overlooked.

is there a reason besides the fact that its not "LOL OSS"?

I think the main reason for it not being spread wider is actually the fact that you can put any A/V codec in it. This makes quite hard to implement on other OSes and specifically on hardware devices.
Also until the arrival of mpeg4 the best video codec for quicktime has been "Sorenson Video 3", which produces real bad quality. Now we got mpeg4 but why not put it in an MP4 container, which will have hardware support in the near future and is also capable of more than qt (as far as I understood the MP4 whitepaper )?

The only reason for AVI being supported on standalone devices is mpeg4. You can't just throw cinepak in and put it in your brand new DVD-combi-player...

And besides: the Win implementation of Quicktime sucks.

MPEG4 is basically MOV with a limited set of codecs that can be put inside. There are some differences, but the similarities go down to the "moov" atoms.

I'll second that. It's almost like apple doesn't care about windows users.

Can I read mov outside Quicktime player ? If not, it's not for me. I hate having to deny the advertisement asking me to update to Quicktime Pro EVERYTIME I watch a Quicktime

so theres no mplayer or VLC for windows?



is writing a .mov parser really that hard?

VideoLan (VLC) works in windows.

I installed it recently and like it. Better than Quicktime, but of course - so is everything.

There is already a directshow .mov parser, by 3ivX. Also ffdshow can handle sorrenson3. So if the file uses sorrenson for video and aac for audio, then it should playback fine in wmp, etc.

The QuickTime file format is not proprietary as such; AFAIK it is fully specified on Apple's developer pages. The implementation is proprietary, but that is an entirely different matter.


There are at least two open source implementions; MPlayer has a mov/mp4 demuxer and libavformat (part of FFmpeg) has both a muxer and a demuxer for mov, 3gpp and MPEG-4. Other than that, both the OpenQuickTime and OpenQT projects should have support for reading and writing QuickTime files (if their names are more than "clever" marketing schemes ). 3ivx have created a MOV/MP4 splitter for WiMP based on OpenQT. I hear it has some synch issues, but they should be fixed in their next release.

There are two good players supporting MOV and the derivatives for Windows: MPlayer and VLC.

With this and the comments above, I believe you're out of excuses

So they say and yet i follow their instructions exactly and it just doesn't work. I've also never heard of anyone else successfully decodeing S3 content in WMP either. Am I missing something?

I use MPC (Media Player Classic) with the legally-dubious QuicktimeAlternative codecs, that allows for a reasonable viewing experience

There's OpenQuicktime (used by 3ivx) and Quicktime4Linux, both GPL IIRC and based on the same base sourcecode. Don't have any links right now, but should be easy to find.

Menno

MOV is actually not a bad container, and i agree would have fulfilled most of the needs of today for video.

However, you have to understand that the MOV specs are created by Apple and the only way to create an official 'way' of muxing Vorbis into MOV, as well as FLAC, Wavpack4, MPC, vobsubs, compressed vobsubs, etc. , like we do with matroska now, would be to make a MOV fork, maybe based on one of the existing open libs mentioned above, because you cant expect Apple to put all this into the official MOV specs.

The point is, by doing that you may find yourself in the center of patent problems all of a sudden, as some of the stuff that can be found in MOV and MP4, like atoms, is subject to patents.

While i admit we never really looked into using MOV for our purposes when we started with MCF/matroska, i am pretty happy today we started from scratch, thus avoiding all patent and license problems.

Does this make sense to you ?

It's very extensible actually. Not sure aboput MOV, but MP4 can be completely extended and user defined codecs can be added. I think mp4creator can mux vorbis into MP4, for example.



Is not patented. Only hinting is patented, and that was only introduced in MP4 (and not necessarily present in the files. For reading you can just ignore it).

Menno

Are you sure about it?

To the best of my knowledge, only hinting is patented in MOV/MP4. And hinting is only needed in streaming.

This is the patent related to hinting:
http://patft.uspto.gov/netacgi/nph-Parser?...RS=PN/6,134,243

And Apple never demonstrated interest in charging licenses for the patent.

Edit: Bleh, Menno was faster

Anyway, regarding mov's extendability:

You can extend it all you like. There are quicktime components for adding vorbis streams in MOV - and that has never been approved by Apple.

All you need to do is create a new code for the media type atom (much like the fourCCs), and provide a quicktime component (much like a directshow filter) for your codec. In this aspect, MOV is not different from AVI.


@Menno: There is one patch for MPEG4ip that allows inclusion of Vorbis streams in the MP4 container. Of course, that is not recommended since most MP4 players expect the audio stream to be AAC, at most MP3.
http://www.maersk-moller.net/projects/mpeg4ip/

I find this hard to believe. Especially considering that there already is an Ogg Vorbis component for QuickTime - and whatever QuickTime can decode, it can put into a movie. (It may not be the optimal way the component does it, but it's possible.) What prevents you from creating a QuickTime component for viewing vobsub streams? Or just adding it to your own player and ignore QuickTime? AFAIK you can even add atoms to the file to store information, and players which do not recognise the atom would simply ignore them. At least that is how metadata is implemented in MP4 files.

Anyway, on Tuesday, 3ivx D4.5 will be released, and it should fix the bugs in D4.0.4, making MOV and MP4 in <insert Windows player here> work quite well.

EDIT: Bleh^2, Roberto and Menno were faster

Hmmm ..... i dont know why, but i dont like the idea of 'user defined' specs for muxing a special codec into a container. IMHO, there should be a central intelligence specifying how certain codecs have to be handled in the container, especially with respect to the muxing and editing app that has to deal with all this later. But maybe my lack of technical understanding is responsible for that, maybe thats not at all necessary for the MOV container, as i understand there are no so strict rules of how to mux the streams into the container, compared to matroska.



Are you sure about that ? When we launched the matroska EBML specs there were several people warning us we might infringe MP4 atoms with respect to the extensibility ? Of course, as EBML is binary XML, the only patent we fear to be confronted with is XML, and nothing else ....

There is a central intelligence, called Apple. But that doesn't prevent users and codec developers defining custom configurations for codecs.

That's the exact same as AVI. Microsoft defines fourcc codes for several codecs. And that didn't prevent people coming up with "div3", "div4", "xvid"...

Actually, it is my belief that it's actually better that not all the power is at this central intelligence's hands. Such a scenario would only lead to bureaucracy (the codec developer needs to address this central intelligence to create a spec for his codec), and could even lead to politics ("We don't like this codec developer, so we won't support his codec")



Well, you are invited to do a patent search. The only patent mentioned as belonging to Apple at the MPEGla site is the one related to hinting that I already linked here.

there's a simple trick to override this: next time before you start quicktime set your system date a few years forward, start quicktime and select 'update later', close quicktime, set system date back and you won't be bothered with that advertisement again
i don't know if this still works with QT 6.x, but it worked with 4.x and 5.x.

-andy-

This was only possible because they defined VfW and the VCM and ACM APIs, so that codec developers could define their own way of how their codecs can be used in the media framework called VfW ( Video for Windows ).

We will come there also one day, maybe together with the Gstreamer guys ( if we get a win32 port of it ) or with CoreAPI.




Well, for matroska the central intelligence is us, and we would be happy if codec developers would contact us to ask how they could best integrate their codec into our container right now .... lol ....

Again, how is this different from QuickTime? Using the QuickTime API - or your own independent implementation - you are free to put whatever data your heart desires into either an atom or a track, and then interpret that data as you see fit.

I think you will find it difficult to bring up an argument why Matroska should be used over QuickTime's file format (and the derivatives), for the simple reason that they are very much alike. With the one difference that QuickTime is an established format with several independent implementations, two of which are used commercially.

I honestly have no idea that there actually IS such as an Quicktime API, and i dont know of any codecs that have implemented it, do you ?



Argument ? Why should i see the necessity to defend ourselves for what we did ? I dont need no arguments. If people like to use our container, they do so, if they like to use MOV they can do that also. Its as simple as that.

This is the QTFF (QuickTime File Format) specification and API.
http://developer.apple.com/documentation/q...QTFF/index.html

As you can see, it's very complete and all details about how codecs are identified and handled are present there. Specifically:
http://developer.apple.com/documentation/q...hap3/index.html

And I know of several codecs that were implemented in QuickTime without Apple's "seal of approval": Vorbis, Speex, MNG, AC3, maybe 3ivX (although I'm not sure on that one)

To come back to the original topic, why istn MOV used more if its so good, has an open and well documented API, and is a widely adopted standard ?

I dont think this thread should evolve into a '... why did the matroska devs make their own container instead of using MOV ...' thread .....

"Why isn't Mac used more than PC, if it has superior hardware and the OS is more stable than PC's "standard" OS?"

"Why isn't Linux used more than Windows if it is more stable and secure?"

"Why aren't AAC and Vorbis used more than MP3, since they are better in all aspects?"

We could go on forever here...

But really, I would pinpoint it to the facts that AVI is the official container format for Windows, and Mov is the official container format for MacOS. And windows dominates the market. You can play AVI out-of-the-box on windows, but you can't play MOV without downloading QuickTime. DivX ;-), that really popularized AVI creation in windows, was only released as a VCM for the windows platform, no QuickTime component. And so on...

Maybe I was a bit too aggressive in my wording. What I meant was considering you already have created your file format, finding a reason for people to choose it over MOV will be difficult. If people are happy with MOV, Matroska won't offer them all that much new. I'd like to say the opposite is true as well, but after all, support for Matroska isn't widespread - especially not on macs.

The two formats seem very similar; the differences between EBML and atoms seem to be technicalities anyway - whether you use a UTF-8 character or a 32-bit integer to identify an element.


I believe the reason why MOV hasn't been used widely is that up to recently, there were not many other ways to create one other than using QuickTime. 3ivx D4 introduced a splitter for some Windows API - I think it's DirectShow - but it had synchronisation issues when decoding. They should be fixed with their D4.5 release on Tuesday. FFmpeg only recently added a MOV muxer to it's impressive array of supported file formats. I think the QuickTime file formats will see much wider support in the future. What makes AVI so unique is that every decent player has an implementation of it - it seems MOV is moving slowly, but steadily, in that direction.