Live data from Hacker News

No, you can't do that with H.264

bemasc.net

41–50 of 79 posts

Re: No, you can't do that with H.264

#41
post #9
post #5

Earlier quoted context omitted.

In the sort of world that we've occupied until now there is nobody who is going to sue you for making a cat video using a particular codec. But I notice that over time enforcement of licenses is becoming more strict, and perhaps eventually license enforcement will be automatic. Using the H.264 standard for video on the internet seems like a dumb decision to me. It's not rocket science to choose some format which does…

Unfortunately, Theora is just simply inferior in every technical aspect to h264. The decision comes down to quality vs price. Choosing quality for a higher price is definitely not a "dumb" decision, surely you too often buy a bit more expensive product that promises a better quality. On the other hand, choosing a cheaper, lower quality product is understandable as well if quality is not that important. Vorbis, howeve…

Vorbis, however is really in the league of mp3/aac/wmv, so in the case of audio, picking Vorbis is really a no-brainer.

Maybe on desktops, but not on portables. For whatever reason (bad implementations?), Ogg Vorbis takes more processing to decode than mp3. Numerous portable players have shorter battery lives when playing Ogg.

http://www.hydrogenaudio.org/forums/lofiversion/index.php/t5...

Re: No, you can't do that with H.264

#42

Earlier quoted context omitted.

Sigh. I've recently been doing a lot of Theora encoding as I experiment with HTML5. I agree that Theora is immediately noticeable as worse quality than h264, and it's not a subtle difference. Even when the situation should have tricked me into thinking it was not Theora - for example, I was testing on Mac Safari and it should have been playing h264 but accepted the Theora because Perian was installed - I instantly no…

Are you sure you're using Theora 1.1 (aka Thusnelda)? Version 1.1 might be consider "inefficient" but 1.0 was simply "broken" for many uses. You mention Macs and I'm not sure if there's been an official 1.1 release of the XiphQT Quicktime plugin yet, which would affect you if you're using Mac tools to generate the video. Also, does Perian play Theora now? I'm guessing you meant the XiphQT plugin which enables Safari…

Ah, you are probably right about the playback. I have both installed - if you say it's XiphQT that enables Safari to use Theora, you're probably right.

As for the encoding, I was using ffmpeg2theora 0.25, which according to the release notes bundles libtheora 1.1. Whether or not I'm using it optimally, though, I can't say. I was loosely following the instructions for batch encoding here:

http://diveintohtml5.org/video.html

.. but no matter what I do, the .ogv comes out looking like a poorly encoded DivX.

Anyway I'll keep trying, thanks for the correction on the QT plugin.

Re: No, you can't do that with H.264

#43
post #41
post #9

Earlier quoted context omitted.

Unfortunately, Theora is just simply inferior in every technical aspect to h264. The decision comes down to quality vs price. Choosing quality for a higher price is definitely not a "dumb" decision, surely you too often buy a bit more expensive product that promises a better quality. On the other hand, choosing a cheaper, lower quality product is understandable as well if quality is not that important. Vorbis, howeve…

Vorbis, however is really in the league of mp3/aac/wmv, so in the case of audio, picking Vorbis is really a no-brainer. Maybe on desktops, but not on portables. For whatever reason (bad implementations?), Ogg Vorbis takes more processing to decode than mp3. Numerous portable players have shorter battery lives when playing Ogg. http://www.hydrogenaudio.org/forums/lofiversion/index.php/t5...

Wild guess: The libvorbis, the "official" codec is using floating point numbers, hence the decreased performance on mobile devices that don't have native floating point support. There are integer-based codecs for MP3 (such as libmad) so it's not an issue with MP3.

I suppose the solution would be writing an integer-based codec for Vorbis as well.

Re: No, you can't do that with H.264

#44
post #41

Earlier quoted context omitted.

Vorbis, however is really in the league of mp3/aac/wmv, so in the case of audio, picking Vorbis is really a no-brainer. Maybe on desktops, but not on portables. For whatever reason (bad implementations?), Ogg Vorbis takes more processing to decode than mp3. Numerous portable players have shorter battery lives when playing Ogg. http://www.hydrogenaudio.org/forums/lofiversion/index.php/t5...

Wild guess: The libvorbis, the "official" codec is using floating point numbers, hence the decreased performance on mobile devices that don't have native floating point support. There are integer-based codecs for MP3 (such as libmad) so it's not an issue with MP3. I suppose the solution would be writing an integer-based codec for Vorbis as well.

There is an integer implimentation called tremor. I believe it is more to do with the fact that there is often a dedicated mp3 decoder in these portable devices, and it has to burn the cpu for decoding vorbis.

Re: No, you can't do that with H.264

#45
post #41

Earlier quoted context omitted.

Vorbis, however is really in the league of mp3/aac/wmv, so in the case of audio, picking Vorbis is really a no-brainer. Maybe on desktops, but not on portables. For whatever reason (bad implementations?), Ogg Vorbis takes more processing to decode than mp3. Numerous portable players have shorter battery lives when playing Ogg. http://www.hydrogenaudio.org/forums/lofiversion/index.php/t5...

Wild guess: The libvorbis, the "official" codec is using floating point numbers, hence the decreased performance on mobile devices that don't have native floating point support. There are integer-based codecs for MP3 (such as libmad) so it's not an issue with MP3. I suppose the solution would be writing an integer-based codec for Vorbis as well.

There is integer-only decoder for Vorbis, called Tremor.

What is missing is decoding by DSPs built in mobile devices. You need to bug TI, Broadcom and other DSP makers for that.

Re: No, you can't do that with H.264

#46
post #9
post #5

Earlier quoted context omitted.

In the sort of world that we've occupied until now there is nobody who is going to sue you for making a cat video using a particular codec. But I notice that over time enforcement of licenses is becoming more strict, and perhaps eventually license enforcement will be automatic. Using the H.264 standard for video on the internet seems like a dumb decision to me. It's not rocket science to choose some format which does…

Unfortunately, Theora is just simply inferior in every technical aspect to h264. The decision comes down to quality vs price. Choosing quality for a higher price is definitely not a "dumb" decision, surely you too often buy a bit more expensive product that promises a better quality. On the other hand, choosing a cheaper, lower quality product is understandable as well if quality is not that important. Vorbis, howeve…

The current version of the official Theora encoder does allow you to achieve perceived visual quality comparable to that of the current h264 encoders at not that much higher bitrate in case of, at least, the typical "Internet Video" content.

Yes, to reach certain quality targets using Theora might mean increased bandwidth cost, but it will also be the only major cost left, for a moment leaving aside the question of adoption. That seem to be a much more predictable cost model than what MPEG LA offers. It might, as well, turn out to be a more affordable model for smaller companies not reaching the MPEG license fee caps.

Paying more for a better product? I'd chose the, possibly higher, price of pushing Theora adoption and Theora-encoded bits now to get a promise of a better "quality", in more aspects than just the visual appearance, down the road.

Re: No, you can't do that with H.264

#47
post #13
post #10

The problem with these write-ups about how bad H.264 is (or will be in 2011) is that it will not reach the right people in time; the masses. The people who clicked through to an article with H.264 in the title already know about the ins and outs. The companies that implement and support H.264 also know all about the license, and they have most likely gotten some pretty sweet deals for getting this thing as widely dep…

Worse yet, we do not even know how the license will look When skimming the comments on that page, seems there is something new, (as of yesterday): http://www.mpegla.com/Lists/MPEG%20LA%20News%20List/Attachme... [pdf]

For the PDF avoidant.

MPEG LA’s AVC License Will Continue Not to Charge Royalties for Internet Video that is Free to End Users

(DENVER, CO, US – 2 February 2010) – MPEG LA announced today that its AVC Patent Portfolio License will continue not to charge royalties for Internet Video that is free to end users (known as Internet Broadcast AVC Video) during the next License term from January 1, 2011 to December 31, 2016. Products and services other than Internet Broadcast AVC Video continue to be royalty-bearing, and royalties to apply during the next term will be announced before the end of 2010. MPEG LA's AVC Patent Portfolio License provides access to essential patent rights for the AVC/H.264 (MPEG-4 Part 10) digital video coding standard. In addition to Internet Broadcast AVC Video, MPEG LA’s AVC Patent Portfolio License provides coverage for devices that decode and encode AVC video, AVC video sold to end users for a fee on a title or subscription basis and free television video services. AVC video is used in set-top boxes, media player and other personal computer software, mobile devices including telephones and mobile television receivers, Blu-ray DiscTM players and recorders, Blu-ray video optical discs, game machines, personal media player devices and still and video cameras.

For more information about MPEG LA’s AVC License or to request a copy of the License, please visit http://www.mpegla.com/main/programs/AVC/Pages/Intro.aspx

Re: No, you can't do that with H.264

#48
post #41
post #9

Earlier quoted context omitted.

Unfortunately, Theora is just simply inferior in every technical aspect to h264. The decision comes down to quality vs price. Choosing quality for a higher price is definitely not a "dumb" decision, surely you too often buy a bit more expensive product that promises a better quality. On the other hand, choosing a cheaper, lower quality product is understandable as well if quality is not that important. Vorbis, howeve…

Vorbis, however is really in the league of mp3/aac/wmv, so in the case of audio, picking Vorbis is really a no-brainer. Maybe on desktops, but not on portables. For whatever reason (bad implementations?), Ogg Vorbis takes more processing to decode than mp3. Numerous portable players have shorter battery lives when playing Ogg. http://www.hydrogenaudio.org/forums/lofiversion/index.php/t5...

I'm not sure your link supports your assertion. The negative reports seem 3-6 years old and perhaps based on 3rd party software vs the software they came with.

A more recent set of figures shows that implementation can have an impact, but that if you've got control of the software you'll probably do better with Vorbis. Note the test is on matching bitrates, but Vorbis is more efficient per bit.

http://tuomas.kulve.fi/blog/2009/11/07/n900-battery-duration...

It also shows that FLAC has longer battery runtimes which I'm guessing is because flash based players no longer need to spin up disks more for the larger filesizes.

Re: No, you can't do that with H.264

#49

Earlier quoted context omitted.

Are you sure you're using Theora 1.1 (aka Thusnelda)? Version 1.1 might be consider "inefficient" but 1.0 was simply "broken" for many uses. You mention Macs and I'm not sure if there's been an official 1.1 release of the XiphQT Quicktime plugin yet, which would affect you if you're using Mac tools to generate the video. Also, does Perian play Theora now? I'm guessing you meant the XiphQT plugin which enables Safari…

Ah, you are probably right about the playback. I have both installed - if you say it's XiphQT that enables Safari to use Theora, you're probably right. As for the encoding, I was using ffmpeg2theora 0.25, which according to the release notes bundles libtheora 1.1. Whether or not I'm using it optimally, though, I can't say. I was loosely following the instructions for batch encoding here: http://diveintohtml5.org/vide…

I was actually hopeful that Perian had started supporting it, I'd guess their installed base would be much higher than XiphQT which would mean, e.g., Wikipedia content just working for many Safari users.

But I checked the site, they only do vorbis in mkv.

Re: No, you can't do that with H.264

#50
I'm making a service for indie filmmakers to self distribute films and this has been a big pickle.

I was making an encoder for them to download using ffmpeg but then realized I may have wasted my time because of the licensing of h264.

So the problem became: do I take the cost and encode their video or do I let them send the video prencoded for me. I had assumed that if they had encoding software that they have already paid for a license. Anyways, i've added support for ogg to the service (with warnings about it's usability).

Does anyone know if sorenson squeeze has this licensing issue?

Post reply on HN