Message: 1 Date: Tue, 4 Nov 2014 12:17:43 +0100 From: Sebastiaan ter Burg terburg@wikimedia.nl To: wikivideo-l@lists.wikimedia.org Subject: [Wikivideo-l] MOX file format Message-ID: <CAH=K=+
nsvBZB4SbPeU4cCQWGFSLErpb1qqS2FQakXB9EcCa2ng@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Hi everyone,
any thoughts on this IndieGoGo campaign? https://www.indiegogo.com/projects/mox-file-format/x/4029267
I'm a sucker of supporting this kind of projects (I also backed the openshot 2 kickstarter <
https://www.kickstarter.com/projects/421164014/openshot-video-editor-for-win...
), but I'm not sure if this initiative could fill the gap we experience in sharing high quality footage. What do you think?
-- Sebastiaan ter Burg *Projectleider Culturele Samenwerking* *Wikimedia Nederland*
From their funding page
" MOX will read and play consistently on Mac, Windows, Linux, or any other platform. This is because MOX will be an open format based on open standards"
Ha. If it was really that easy to get interopability we'd be living in the land of ogg and webm. (Arguably there may be less lock in in the pro market, but stil-open source is not interopability magic)
So the tech summary of this (afaict. Im not a video expert so correct me if im wrong) they are taking a container format aimed at pro users named mxf, which is much like tiff in that it can contain anything and hence has interopability problems as you never know whats inside. They are taking mxf making a profile of it called mox which is limitted to free codecs, and specificly codecs that a pro would want to use (lossless or high quality lossless) as an intermediary format.
The codecs are one of: Dirac, OpenEXR, DPX,PNG, and JPEG.
Dirac is an interestng choice. I suppose its chosen because it has a lossless option, but from what i understand its very slow to encode, so i wouldnt think its suitable for this usage (maybe im mistaken). The other codecs are just image codecs.
Dpx is an interesting choice given this groups goals as wikipedia describes it as "non-free SMPTE standard, 17 pages, USD 120" (although maybe that only refers to the standard. There exists free software implementations)
The audio codecs are: flac, opus and raw pcm. Flac and pcm are lossless, opus is a high quality lossy codec.
---
Im unsure what exactly the issues with sharing high quality footage are, but I assume there are three: * file size - 1gb limit on commons (and realistically >100mb is flaky) *social -some people worry about dumping source material on commons. I think this concern is overblown but one should not underestimate social problems *inconvinent formats -ogg/webm is hard to convert to. Aimed at end use not intermediary use.
This could help with the third point potentially (in the far future if it becomes an industry std, which is a big "if". Open source projects fail all the time). I think there is probably more we can do on the format front. Its a complex issue, but im pretty sure not all avenues have been exhausted, or perhaps even explored.
An approach that may lead to more immediate results is something like pro-res (if unpatented) which allegedly is decodable in ffmpeg (unpatented and decodable in ffmpeg is basically the criteria for enabling a new format on commons), and also has the benefit of existing right now
--bawolff
Hi Bawolff,
for someone that's claiming not to be an expert you sure seem to know what you're talking about. On the sharing the high quality footage: the idea is not to dump it on commons, but to set up a separate server https://docs.google.com/a/wikimedia.nl/presentation/d/1sS8WnMO6iFidEPkyNMk2gZQkuyEgRYeupI4uELS4erc/edit#slide=id.p where we can share Wikimedia related footage (at first) with each other. For example: interviews with Wikipedias, b-roll shot at events, screencaptures of instructional videos, timelines and more of that kind of stuff. Choosing one codec would be the easiest at first, but if licensing would allow it it could be more economical to share the out of camera files. In other words: no use to upload ProRess422 when I shot a video in AVCHD... The AVCHD files will be much smaller and the edit suites I know don't have a problem rendering these files to a more edit-friendly format again.
Any ideas how we could get an answer on the licensing question for hosting certain formats?
Best,
Sebastiaan
2014-11-04 17:11 GMT+01:00 bawolff bawolff+wn@gmail.com:
Message: 1 Date: Tue, 4 Nov 2014 12:17:43 +0100 From: Sebastiaan ter Burg terburg@wikimedia.nl To: wikivideo-l@lists.wikimedia.org Subject: [Wikivideo-l] MOX file format Message-ID: <CAH=K=+
nsvBZB4SbPeU4cCQWGFSLErpb1qqS2FQakXB9EcCa2ng@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Hi everyone,
any thoughts on this IndieGoGo campaign? https://www.indiegogo.com/projects/mox-file-format/x/4029267
I'm a sucker of supporting this kind of projects (I also backed the openshot 2 kickstarter <
https://www.kickstarter.com/projects/421164014/openshot-video-editor-for-win...
), but I'm not sure if this initiative could fill the gap we experience in sharing high quality footage. What do you think?
-- Sebastiaan ter Burg *Projectleider Culturele Samenwerking* *Wikimedia Nederland*
From their funding page " MOX will read and play consistently on Mac, Windows, Linux, or any other platform. This is because MOX will be an open format based on open standards"
Ha. If it was really that easy to get interopability we'd be living in the land of ogg and webm. (Arguably there may be less lock in in the pro market, but stil-open source is not interopability magic)
So the tech summary of this (afaict. Im not a video expert so correct me if im wrong) they are taking a container format aimed at pro users named mxf, which is much like tiff in that it can contain anything and hence has interopability problems as you never know whats inside. They are taking mxf making a profile of it called mox which is limitted to free codecs, and specificly codecs that a pro would want to use (lossless or high quality lossless) as an intermediary format.
The codecs are one of: Dirac, OpenEXR, DPX,PNG, and JPEG.
Dirac is an interestng choice. I suppose its chosen because it has a lossless option, but from what i understand its very slow to encode, so i wouldnt think its suitable for this usage (maybe im mistaken). The other codecs are just image codecs.
Dpx is an interesting choice given this groups goals as wikipedia describes it as "non-free SMPTE standard, 17 pages, USD 120" (although maybe that only refers to the standard. There exists free software implementations)
The audio codecs are: flac, opus and raw pcm. Flac and pcm are lossless, opus is a high quality lossy codec.
Im unsure what exactly the issues with sharing high quality footage are, but I assume there are three:
- file size - 1gb limit on commons (and realistically >100mb is flaky)
*social -some people worry about dumping source material on commons. I think this concern is overblown but one should not underestimate social problems *inconvinent formats -ogg/webm is hard to convert to. Aimed at end use not intermediary use.
This could help with the third point potentially (in the far future if it becomes an industry std, which is a big "if". Open source projects fail all the time). I think there is probably more we can do on the format front. Its a complex issue, but im pretty sure not all avenues have been exhausted, or perhaps even explored.
An approach that may lead to more immediate results is something like pro-res (if unpatented) which allegedly is decodable in ffmpeg (unpatented and decodable in ffmpeg is basically the criteria for enabling a new format on commons), and also has the benefit of existing right now
--bawolff
Wikivideo-l mailing list Wikivideo-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikivideo-l
On 11/4/14, Sebastiaan ter Burg terburg@wikimedia.nl wrote:
Hi Bawolff,
for someone that's claiming not to be an expert you sure seem to know what you're talking about. On the sharing the high quality footage: the idea is not to dump it on commons, but to set up a separate server https://docs.google.com/a/wikimedia.nl/presentation/d/1sS8WnMO6iFidEPkyNMk2gZQkuyEgRYeupI4uELS4erc/edit#slide=id.p where we can share Wikimedia related footage (at first) with each other. For example: interviews with Wikipedias, b-roll shot at events, screencaptures of instructional videos, timelines and more of that kind of stuff. Choosing one codec would be the easiest at first, but if licensing would allow it it could be more economical to share the out of camera files. In other words: no use to upload ProRess422 when I shot a video in AVCHD... The AVCHD files will be much smaller and the edit suites I know don't have a problem rendering these files to a more edit-friendly format again.
Any ideas how we could get an answer on the licensing question for hosting certain formats?
Best,
Sebastiaan
AVCHD = MP4 = H.264 (+ AC-3), so on that front we essentially have an answer, its just not a very popular answer (In this particular group anyways). The RFC was pretty conclusive, although there may be a little wiggle room about converters on say the tool labs (unclear), it seemed pretty decisive about not allowing H.264 content on Wikimedia servers in general. Perhaps there's wiggle room on unofficial servers, but than its not an official project.
In my comment about format's, I meant more that there's probably other lossless formats which we could probably use, at the expense of very large files. (Maybe anyways, it would need to be explored. I don't really know).
Keep in mind also, that converting AVCHD -> Proless/some lossless thing/etc -> AVCHD is going to cause quality loss, in a similar fashion as if someone repetitively converted JPEG -> PNG -> JPEG. Of course at this stage that's probably a minor concern. The bigger issue is getting anything up and running.
One of the problems we're facing with the renewed video initiative is a codec to share high quality footage with each other. I've emailed Apple several times to ask if there are any restrictions in sharing files - solely sharing without playback - on a public server, but they haven't responded yet. That's why this initiative caught my eye.
I'm not sure what grounds Apple would have to restrict sharing such files (Unless they had you sign an actual NDA before giving you a product that could make such a file).
Sure if they had a patent on it, they could restrict the use of products which create, convert or play the file (Do they [or other people] actually have a patent on any of the ProRes stuff? Anybody know?). But I'm not sure how they would have any ground to restrict the distribution of such files. If they somehow had a copyright claim on the resulting file, perhaps - but that sounds absurd to me. After all the file is created entirely by a mechanical process, while the mechanical process might be very creative, the result is created by a machine and thus by definition lacks creativity (Unless some "creative" constants are copied in). IANAL.
The more likely issue, is that if the file format is patented, we wouldn't be able to use due to internal political reasons, rather than any external restrictions.
Hay said:
Then there's the problem of hardware. Think about OGG/WebM: used very little on mobile because there's no support for hardware acceleration (and hence: better battery life).
So, no. I don't think this initiative could fill the 'open source high quality video gap'.
I don't think that's really where this particular group is targeting. A format like mox would have impractically large file sizes for mobile (or any sort of end viewing), and hardware acceleration really doesn't matter very much when almost no compression is being used. Work in the high quality open source video codec space for final product video is more centered around VP9 and Dalla AFAIK.
--bawolff
Hmm, is Motion-JPEG no longer a thing, or does it not scale to the quality or bitrates needed nowadays? Used to be pretty standard back in the days I was fiddling with video editing 10-15 years ago (MJPEG would be packaged usually in .avi or .mov depending on the platform's preferred video container).
In general I'd recommend:
a) Seriously consider a non-US-based hosting platform for original files to stay outside the extra-scary US patent regime. (IANAL etc)
b) Store original-format files whenever possible for archival purposes.
c) Provide conversions from anything ffmpeg/avconv-supported to a lossless or near-lossless common intermediate format, whatever it may be.
d) Allow originals to be used as transcode/editing sources for server-side processing if feasible, otherwise use common intermediate format.
e) Make lossy transcodes at lower resolutions available for on-web playback (potentially including playback of clips in an on-web preview & editor tool, with the high-res source or intermediate-format versions used server-side to produce high-quality output from the EDL)
f) Provide "one-touch publishing" from the video upload & editing suite to Commons, sending a high-quality WebM transcode of the full-quality original file or edited output over to Commons. (Probably sideloading via URL.)
-- brion
On Tue, Nov 4, 2014 at 1:56 PM, bawolff bawolff+wn@gmail.com wrote:
On 11/4/14, Sebastiaan ter Burg terburg@wikimedia.nl wrote:
Hi Bawolff,
for someone that's claiming not to be an expert you sure seem to know
what
you're talking about. On the sharing the high quality footage: the idea
is
not to dump it on commons, but to set up a separate server <
https://docs.google.com/a/wikimedia.nl/presentation/d/1sS8WnMO6iFidEPkyNMk2g...
where we can share Wikimedia related footage (at first) with each other. For example: interviews with Wikipedias, b-roll shot at events,
screencaptures
of instructional videos, timelines and more of that kind of stuff.
Choosing
one codec would be the easiest at first, but if licensing would allow it
it
could be more economical to share the out of camera files. In other
words:
no use to upload ProRess422 when I shot a video in AVCHD... The AVCHD
files
will be much smaller and the edit suites I know don't have a problem rendering these files to a more edit-friendly format again.
Any ideas how we could get an answer on the licensing question for
hosting
certain formats?
Best,
Sebastiaan
AVCHD = MP4 = H.264 (+ AC-3), so on that front we essentially have an answer, its just not a very popular answer (In this particular group anyways). The RFC was pretty conclusive, although there may be a little wiggle room about converters on say the tool labs (unclear), it seemed pretty decisive about not allowing H.264 content on Wikimedia servers in general. Perhaps there's wiggle room on unofficial servers, but than its not an official project.
In my comment about format's, I meant more that there's probably other lossless formats which we could probably use, at the expense of very large files. (Maybe anyways, it would need to be explored. I don't really know).
Keep in mind also, that converting AVCHD -> Proless/some lossless thing/etc -> AVCHD is going to cause quality loss, in a similar fashion as if someone repetitively converted JPEG -> PNG -> JPEG. Of course at this stage that's probably a minor concern. The bigger issue is getting anything up and running.
One of the problems we're facing with the renewed video initiative is a codec to share high quality footage with each other. I've emailed Apple several times to ask if there are any restrictions in sharing files - solely sharing without playback - on a public server, but they haven't responded yet. That's why this initiative caught my eye.
I'm not sure what grounds Apple would have to restrict sharing such files (Unless they had you sign an actual NDA before giving you a product that could make such a file).
Sure if they had a patent on it, they could restrict the use of products which create, convert or play the file (Do they [or other people] actually have a patent on any of the ProRes stuff? Anybody know?). But I'm not sure how they would have any ground to restrict the distribution of such files. If they somehow had a copyright claim on the resulting file, perhaps - but that sounds absurd to me. After all the file is created entirely by a mechanical process, while the mechanical process might be very creative, the result is created by a machine and thus by definition lacks creativity (Unless some "creative" constants are copied in). IANAL.
The more likely issue, is that if the file format is patented, we wouldn't be able to use due to internal political reasons, rather than any external restrictions.
Hay said:
Then there's the problem of hardware. Think about OGG/WebM: used very little on mobile because there's no support for hardware acceleration (and hence: better battery life).
So, no. I don't think this initiative could fill the 'open source high quality video gap'.
I don't think that's really where this particular group is targeting. A format like mox would have impractically large file sizes for mobile (or any sort of end viewing), and hardware acceleration really doesn't matter very much when almost no compression is being used. Work in the high quality open source video codec space for final product video is more centered around VP9 and Dalla AFAIK.
--bawolff
Wikivideo-l mailing list Wikivideo-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikivideo-l
wikivideo-l@lists.wikimedia.org