Slashdot Mirror

H.264 is miles ahead of Theora, *especially* at low bit rates

I agree with this statement.

The original page, comparing Theora to YouTube videos, was not trying to trick people into thinking Theora is better than H.264. A guy at Google famously claimed: "If [youtube] were to switch to theora and maintain even a semblance of the current youtube quality it would take up most available bandwidth across the Internet." The Theora guy compared the quality available per bit with Theora, with the actual quality of actual videos from YouTube; and he concluded that Theora was able to meet or beat the quality of the example videos he pulled from YouTube. At the bottom, in the conclusions section, he noted that YouTube uses H.263, and just a subset at that, for many videos, making it easier for Theora to match YouTube; and even for H.264, YouTube isn't making full use of the standard. He speculated that perhaps YouTube was making a tradeoff, allowing the files to be bigger to make them easier to seek within or some such. Here's the link again; go read what he actually wrote.

I'm sure there are idiotic, deranged Theora fanboys out there who claimed it is better in all ways than H.264, but I am not one. If you make effective use of H.264 you get the highest quality per bit possible with current video technology, full stop. The only advantage of Theora is that it is patent-free and BSD-licensed. That is a very large advantage for some purposes.

But now we have VP8, which is much better than Theora, while still freely available for use. It's not as good as H.264, but it seems to be better than everything else, and good enough for practical use. It will have its place.

H.264 isn't going away. But short of a successful patent challenge, WebM isn't going away either. Its advantages in freedom will make it the top choice for many purposes, even though it can't match the ultimate quality per bit of H.264.

steveha

I remember the comparisons of Theora Vs "H.264" and the only one even remotely favorable for Theora was actually underhandedly using H.263 for most of the comparisons. If you need a citation beyond the one I just gave, notice how this page never mentions the fact that that is H.263, but instead propagandizes that it is H.264 in their comparison images (directly copied from the first citation, which does at least honestly note that it is H.263)

Sorry. H.264 is miles ahead of Theora, *especially* at low bit rates, as you can see here, take note of the 486Kbit still images halfway down. On the left is 1Mbit Theora, on the right is 486Kbit Theora, and in the middle are the best images... 486Kbit H.264

Maybe I'm ignorant. Maybe I'm just dumb. But I can't, for the life of me, understand why it is so difficult for Apple and Microsoft just to ship libtheora with their Operating Systems and stop bothering everbody. Seriously, it's licensed BSD for god-sakes. And as far as hardware decoding goes, most computers (By computers, I mean anything with a microprocessor in it. Your iPhone, your Desktop, your Laptop, and your Calculator.) have the processing power to handle Theora in software, if they are capable of using Flash already. Furthermore, if it is really that big of a deal, that a company needs a hardware decoder, the company that wants it the worst will pay to finish up the decoder project. This would mean that people would have to upgrade their devices, but can you really tell me that people who own iPhones don't upgrade them regularly to keep up with Apple's latest shiny new toy? Lastly, their is the argument of quality, but that has already proven to be crap. See here Theora may not be the fastest, but it's not really any worse or better than anything else we can use in terms of quality. As such, the only really difference is that H264 is closed and patented, and Theora isn't. And if we're are completely honest with ourselves, submarine patents are a non-issue. Remember, the patent system is so screwed up, that there is a patent on a stick. (Trust me, google it. It's there, in all it's patent glory.) So although patent trolls could cause problems, it's not like that's anything new and it would be dealt with, just like with the hardware decoding.

So, in conclusion, I don't know why people keep arguing. I didn't know it was that hard for a multi-billion dollar company to link in one more library. It's not like any other codec will do any better. Theora just has the advantage of being open by nature, and not encumbered by patents. So I wish we could just move on, and use Theora.

(In the interest of disclosure, I am very pro theora and am a long time Linux user. As such, any solution involving patents is revolting, and flash is disgusting in terms of support and distribution.)

Unfortunately I don't have a way to verify but I have been told that Safari does not allow 3rd-party codecs to be used (otherwise it should be trivial to add WebM support to it?).

Things may have changed since I last heard, but yes, that's the implication -- it should be trivial to provide a WebM plugin for Safari, which would provide exactly as much WebM as Microsoft is promising IE will support.

Also what you say just isn't true: Safari does _not_ use native platform codec support it uses _Apple's_ (witness Safari on Windows), meaning fixing any issues with it is still fully within Apple's control.

I don't think anyone ever really claimed Safari on Windows was a good thing. But since QuickTime has a plugin system, Apple is limited to fixing vulnerabilities in the codecs they ship.

Also, it is the native platform on OS X. If you want to play video, that's what you use. So if I were to write a browser, I'd do exactly what Safari does on OS X, and I'd probably do something similar to what IE does on Windows, and I'd plug into something like GStreamer on Linux.

Of course the big "gotcha" with the BSD/GPL codecs is that MPEG-LA has over 2000 patents that pretty much cover everything one has to do to get video to go from a file on a medium to a picture on a screen

Or so they would want everyone to believe. They might have over 2000 patents stuffed into the standards but that is altogether different.

It would have been three hours long and covered half as much stuff it they took the time to explain every detail. Checkout the wiki for pointers to tons of background material.

"follow-ups to "flesh out" the information would be an extremely helpful next step to break down the various issues under discussion"

As you wish: A Digital Media Primer For Geeks — Wiki edition.

Obviously there is a lot that still need to be done but the folks at Xiph.Org are on it.

Better than an ebook--there's a wiki page with a full transcript and helpful screenshots: http://wiki.xiph.org/A_Digital_Media_Primer_For_Geeks_(episode_1)

Available as OGG and WebM. Is Xiph working for Google now?

Xiph.Org has been pushing for unencumbered codecs for over a decade and contributed to the creation of webm.

It might be more fair to say "WebM? Is Google working for Xiph now?"

Yes. Yes they are, and it is so sweet. Don't be evil sill means a little something sometimes.

Available as OGG and WebM. Is Xiph working for Google now?

Xiph.Org has been pushing for unencumbered codecs for over a decade and contributed to the creation of webm.

It might be more fair to say "WebM? Is Google working for Xiph now?"

Yes. Yes they are, and it is so sweet. Don't be evil sill means a little something sometimes.

Google, by contrast... "Big Brother"? Have you read 1984?

Yes I have. Obviously the current situation isn't like the brutal dictatorship in the book, but the information gathering is getting there. Not a camera in your home, but spying on all the sites you visit.

they only gather information from people who willingly donate said information, or from information already in public spaces.

"willingly" would be opt-in, instead of having to opt-out. How many web sites use Google Analytics? Google also owns DoubleClick.

by accident

There's nothing accidental about collecting all this information (ignoring the wireless case), which once collected, can be abused. If they get a National Security Letter they will have to comply with it. If a rogue employee decides to misuse the data, it's done. If Google decides to misuse the data, it's done. All this is possible because they collect the data.

By having, say, youtube.com/html5 not work at all.

It's completely Google's fault for requiring H.264. They could always fall back to another format. HTML5 does not require H.264. If you want to talk about FUD, then H.264 is where it's at, straight from Google: "If [youtube] were to switch to theora and maintain even a semblance of the current youtube quality it would take up most available bandwidth across the Internet." http://people.xiph.org/~greg/video/ytcompare/comparison.html

YouTube won't re-encode it for the high quality version.

Nothing is stopping them. At the worst, they can offer a lower-quality version, they way they do now with Flash Video. Requiring a patent-laden format to view video content on the web is evil. It's GIF all over again.

Second, only Theora might be open.

More FUD. It's been around for years, and there's no evidence that it isn't. There's no need to use scare words.

Huh, a music CD is a sequence of digitally encoded sectors, one after the other, with error detection and correction. Ripping it to disk files, computer performance should be TOTALLY IRRELEVANT on the face of it. Maybe if you used a half decent ripping tool ... CDDA Paranoia . Of course, that tool only runs on a real operating system ... linux.

Here's Xiph's support: http://www.xiph.org/press/2010/webm/ With Xiph's support, and now ffmpeg folk working on it, WebM's looking very good.

There are a few misconceptions here. WebM isn't Matroska, it's a Matroska subset with a few (important) differences. See http://www.webmproject.org/code/specs/container/ for details.

Which of the differences listed at http://www.webmproject.org/code/specs/container/ do you claim is important?

Ogg has an index which (similarly to Matroska) is optional.

The page from xiph.org does not contain the word "index", and the page I found with Google states: "It is at best incomplete and at worst completely broken."

Actually the current version of Ogg supports an optional index

I was confused. Apparently XiphWiki says Ogg Index is highly not final.

This could be worked around by putting this data in the header somewhere.

AVI has such an index. Matroska (wrapper used by WebM) has an index. Ogg does not, but unless you're on a satellite link, four HTTP seeks won't kill you.

There are a few misconceptions here. WebM isn't Matroska, it's a Matroska subset with a few (important) differences. See http://www.webmproject.org/code/specs/container/ for details. Ogg has an index which (similarly to Matroska) is optional. See:

http://www.xiph.org/ogg/doc/skeleton.html
http://pearce.org.nz/video/indexed-seek-demo.html

ffmpeg2theora includes a Skeleton track with index by default since version 0.27:

http://v2v.cc/~j/ffmpeg2theora/

it requires the video to find the first keyframe, and then play from there.

Ever noticed a delay in changing the channel on digital cable or satellite TV, or jumpiness when rewinding or fast-forwarding a DVD? A lot of that is waiting for the next keyframe. There is a tradeoff between bitrate and seek granuarity because more seekable encodes have to use more keyframes. In fact, a few codecs have an option to have a frame incorporate motion relative to multiple keyframes. Such a "B-frame" can use ordinary keyframes (I-frames), semi-keyframes predicted from prior keyframes (P-frames), future frames (the P-frame or I-frame after a B-frame), and even a long-term reference frame (a golden frame). If a scene uses a golden frame across multiple shots, good luck jumping into the middle of that scene.

Using ranges to try to stream is a massive, massive hack.

Or in other words, a "steaming pile of hack". But MKV already has cueing data, and Xiph is working on an Ogg index for more efficient seeking. The <video> element would reference video data URIs at multiple levels of detail, and a player would download each video's index.

HTTP lets you seek to a byte range, but how does that map to a location within the file?

Find the known timestamps before and after the desired seek point, interpolate where you would need to seek if the part between known timestamps had a constant bit rate, and seek to that part of the file. The last article about Ogg vs. MKV presented a test result that it takes on average 3.5 iterations of this algorithm to get to the right part.

This could be worked around by putting this data in the header somewhere.

AVI has such an index. Matroska (wrapper used by WebM) has an index. Ogg does not, but unless you're on a satellite link, four HTTP seeks won't kill you.

> requires proof that VP8 uses processes and techniques that are substantially similar to
> those claimed

Not even that. To be infringing, you have to be subject to _all_ the claims of the patent. "substantially similar" is not enough if there is one particular claim that doesn't apply to you.

A common way of working around a patent is to pick a particular claim from the patent and make sure that your work is not covered by that claim.

http://lists.xiph.org/pipermail/theora/2010-April/003769.html has a pretty good summary of the state of the patent situation for H.264. Especially note the paragraph starting "It doesn't have to be this way".

First Usenet article referencing Shoutcast was on December 31, 1998. It stated that Shoutcast had been released earlier that same day.

Icecast's first public release happened a few weeks later, on January 18, 1999, according to the tarball, which lines up neatly with the same date on the Freshmeat page.

Wow, seems like not so many people know:

http://dir.xiph.org/index.php

Icecast has been around a long, long time. I've been running a stream off it since 1999 (!!).

Check it out if you like Bay Area punk: http://gilman.duckpond.net/

What is your response to the assertion that multimedia isn't the patent minefield that everyone wants to (make us) believe?

There are now different positions on WebM. In my blog post I actually mention all of them, including (in an update) Carlo Daffara's analysis, which is rather optimistic about WebM not infringing MPEG LA's patents. I never said that it does, but I admit that MPEG LA's statement that it may put together a WebM patent pool is nothing that I would advise anyone to discount as a possible scenario.

What I do say is that Google will have the biggest benefit if this works out, hence I believe it would be reasonable for Google to provide not just vague assurances but a detailed analysis of the patent situation and some form of indemnification. One key difference between H.264 and WebM is that H.264 is already used in countless commercial products, so if anyone could assert any rights against it, it normally should have happened, while VP8 has so far had much less market penetration but if it became (like some propose) part of HTML 5, then this could change overnight and result in patent enforcement.

What is your response to the assertion that multimedia isn't the patent minefield that everyone wants to (make us) believe?

Also worth pointing out that the majority of HTML5 browsers already supports both Vorbis and Theora. I think Safari is the ONLY HTML5-using browser that doesn't have support for Vorbis out of the box. I believe Chrome ALSO supports vorbis and theora (not just aac/h.264), and between Firefox and Chrome that's got to be at least, what, roughly 60-80% of HTML5-supporting browser traffic?

Since Apple® Quicktime® can actually be taught to understand Vorbis, Theora, and all of the other Xiph codecs system-wide (i.e. not just inside the web browser) with the installation of a single component (XiphQT), I think support for Vorbis is already better than people give it credit for...

Vorbis has been used by a number of major game publishers for game audio, precisely because it has no license fees. I think WoW would be a big enough target for the patent trolls, how about you?

That's nothing. Take a look at the very end of http://people.xiph.org/~xiphmont/demo/theora/demo9.html where he says the release after that will be named... Eyjafjallajökull.

So it's much the same way that Safari supports Ogg Theora/Vorbis video and audio - once you've installed the free XiphQT QuickTime® component...

(XiphQT's not that bad - it makes the entire QuickTime®-using system able to understand Ogg file formats and the Xiph codecs, not just the browser. Even so, it's still one additional component that needs to be installed.)

No thanks, I'll take an objective analysis over a rushed out piece of FUD from non-credible ffmpeg/x264 developers any day.

Theora is just gonna end up another Vorbis, a teeny tiny niche nobody but a few FOSS geeks use, just like how everyone plays MP3s even though they are patent encumbered.

Maybe you should read this list of games using Vorbis:

http://wiki.xiph.org/Games_that_use_Vorbis

Millions of people use Vorbis, whether they know it or not. I'm guessing you're one of ones who don't know.

Opera currently supports only Ogg Theora. And it's apparently built-in, as you'd expect for "only support", rather than using the OS. That's a shame.. they are trying to dictate user choice. They do mention that Firefox and Chrome also support Theora... Chrome offering the choice, though I don't know hoe much choice.

The Opera folks also point out that (on the Mac anyway, probably the PC too), Safari is actually doing the right thing. Despite all of Steve Jobs yaking about H.264. Safari uses the Quicktime subsystem to decode the tag. So, if you install the Quicktime CODEC for Ogg Theora, it'll play in Safari. Go here: http://www.xiph.org/quicktime/

A telling thing is Opera's informative introduction to HTML5 video:
http://dev.opera.com/articles/view/everything-you-need-to-know-about-html5-video-and-audio/

You may notice that this has actually made playing video more complex than it was when if was Flash or "just embed a file".

Ars has a good article summarizing a comparison study between Theora and h.264. Basically, Theora produces much lower quality videos with larger filesizes and higher CPU utilization when compared to h.264 videos with identical bitrates.

It's a poor article. The Streaming Learning Center study is deeply flawed. See the comments accompanying it. The method Jan Ozer used to encode his videos was woeful.

Let's use a far more practical standard of measurement. Let's compare how H.264 is used by YouTube today to deliver web video:

http://people.xiph.org/~maikmerten/youtube/

Perhaps Ogg Theora has something more going for it than just an open licence.

People keep making claims like yours and the Theora developers keep SPECIFICALLY addressing those claims, and yet you APPHOLES keep making those claims.

REFERENCE

What provides protection for Theora is: (in no particular order)
(1) Active avoidance of the well known and/or aggressively enforced codec IPR.
(2) The preference for older techniques and with a strong prior
art-history needed for (1) also provides some protection against
unknown patents.
(3) Theora and or VP3 have been shipped by a multitude of
deep-pocketed entities (IBM; RedHat; Google; Apple[1]; Mozilla; etc)
who would make much better litigation targets than you likely would.
At least some of these have done their own reviews and decided to go
forward.
(4) The dynamics of patent enforcement discourage long-shot
prosecution for patents that royalties are being collected on from
other uses due to the risk of claim invalidation.

Xiph's Gregory Maxwell wrote a couple of responses on the Theora mailing list:

http://lists.xiph.org/pipermail/theora/2010-April/003766.html
http://lists.xiph.org/pipermail/theora/2010-April/003769.html

Xiph's Gregory Maxwell wrote a couple of responses on the Theora mailing list:

http://lists.xiph.org/pipermail/theora/2010-April/003766.html
http://lists.xiph.org/pipermail/theora/2010-April/003769.html

The biggest pusher of open standards on the web? That must be why they support Theora in Safari for the HTML5 video tag.

Works fine here. You just need the proper codec. Granted, not on an iPhone, but still.

Here's the perspective of Greg Maxwell from Xiph on Steve Jobs' claims:

http://lists.xiph.org/pipermail/theora/2010-April/003769.html

safari only supports h.264 in the html5 video tag

Safari supports whatever codec happens to be installed. By default Apple installs H.264 and not Theora (which is still available separately).

And, yes, I know defaults are powerful things.

H.264 is not supported by Firefox so you do need to encode videos again in Ogg Theora if you want to support HTML5 video in Firefox.

Looking at it the other way around, Safari doesn't support the open codec so you need to use H.264 to support Safari (and IE9, when that comes out).

Gee, I wonder why I can watch this on my Mac with Safari without a problem then. Must be doing something wrong. Ahh, the miracles of Quicktime with the right codec http://xiph.org/quicktime/

The only significant hardware vendor that refuses to support Ogg is Apple. Even that's not necessarily true if you're a fan of the Rockbox firmware. Either way, all the Apple guys probably stopped reading this thread two or three levels back.

Ok, I'll keep this short. There are two marginally popular DAPs that support Ogg Vorbis - Cowan and iRiver. See http://wiki.xiph.org/index.php/PortablePlayers

It's not just Apple that doesn't support Ogg Vorbis. It's basically everyone. But assume for a moment you are correct. You make it sound as if Apple makes up an insignificant percentage of the DAP market. Apple's iPod makes up 90% of the market for hard-drive based DAPs and 70% of the overall market. Including other Apple products, that number increases to at least 75%. Ignoring 75% of the market is actually a big deal. See http://www.appleinsider.com/articles/06/05/24/ipod_how_big_can_it_get.html AND http://battleangel.org/2006/08/26/dap-market-shares/

Furthermore, the Ogg Vorbis supporting players all fall into the last category which appears to make up 6% of the total market. Do you really think that being excluded from nearly 95% of the DAP market is inconsequential?

You keep citing to statistics on file-sharing sites, for the formats that leeches would prefer to have provided to them for free. Number one, I frankly don't understand the motivation behind catering to those who would look a gift horse in the mouth. However, if you want to share a CD with the torrent community, or your girlfriend, or whoever, then fine... rip to MP3. I don't see what any of that has to do with the format that you choose to rip for your own personal purposes.

The reason is obvious. Nobody wants to their CD collection multiple times. Ripping to MP3 ensures you can share with anyone. Ripping to FLAC or another lossless format ensures you can transcode to any format. Ripping to Ogg Vorbis ensures you have a format that virtually nobody wants.

Here's a simpler comparison for you. Theora compared to YouTube H.264. The Theora video is smaller and looks better to me:

http://people.xiph.org/~maikmerten/youtube/

Ah, the gotcha is in the source:

http://svn.xiph.org/trunk/vorbis-tools/ogginfo/ogginfo2.c

Ogginfo's source includes information on how to process the metadata for various codecs.

So, the grandparent's complaint is still valid. Ogginfo appears to require recompilation for every stream that they want to support inside an ogg container.

That's not Google's view, though. That's just a single guy, who's working independently from Google, but they finance his work. Their official stance with respect to backing H.264 for HTML5 didn't change - they're still opposed to standardizing on Theora.

Oh, I remember this guy. See: http://people.xiph.org/~greg/video/ytcompare/comparison.html

I doubt this person speaks for all of Google too, considering other departments are doing the opposite as shown before.

Another online-video comparison:

http://people.xiph.org/~maikmerten/youtube/

Here's a better comparison: http://people.xiph.org/~greg/video/ytcompare/comparison.html

In order to play Flash video, or Silverlight video, browsers need a plugin.

That's less of a disadvantage if PC makers install the plug-in on new PCs. This appears to be the case at least for Flash Player.

Theora/HTML5 video can play in Firefox, Opera, Google Chrome (without any plugin) and IE (this browser alone requires a plugin).

Doesn't Safari need the XiphQT plug-in too?

>>> http://people.xiph.org/~greg/video/ytcompare/comparison.html

I don't see any difference between the MPEG4 codecs ((H.264+AAC) and the OGG codecs (Theora+Vorbis). The two images look identical. It's too bad the author did not provide larger images.

Chris DiBona of the Google open source group claimed that "If [youtube] were to switch to theora and maintain even a semblance of the current youtube quality it would take up most available bandwidth across the Internet."

This was shown to be false.

Mr DiBona then mysteriously vanished without trace.

Could he please manifest and either (a) support his claims or (b) concede his error?

Thanks ever so much.

Its a crossroad, because supporting the patended h264 is betraying theora adoption in html5 video tag (witch was the first proposal), witch is betraying open source, whitch is the essense of firefox. And also opera supports ogg/theora in html5. On the other hand theora's main problem is not performance, is that it allows direct content downloading, witch will be never accepted by most companies like youtube. I guess its time for google to proove the not evil moto, because if youtube will support theora the case is closed, and it will be a victory for open standarts.

Ogg Theora is complete market failure that nobody actually uses for anything. End of discussion, period.

You can't end a discussion with false assertions, Theora isn't in widespread use. But it is used in Firefox, Opera, Chrome, Wikipedia, Dailymotion, and a small number (for now) of comercial games. Thus the "nobody actually uses for anything" part is demonstratively wrong and you can go back under the bridge.

>>I don't recall an H.264 fee when I bought my computer. Neither will end users. You will pay where you are getting your content. That me be increased fees or more ads at Youtube or wherever. >>The simple truth is that H.264 IS better in quality than Ogg, which is closer in compression ratios to MPEG-2. That is crap. Look at this: http://people.xiph.org/~greg/video/ytcompare/comparison.html Then go download Opera and Chrome. Opera uses Theora and Chrome uses H.265. Go find some sites that use h.264 like Youtube. It looks horrible. Now go find some sites that use Theora they look awesome.