Talk:Advanced Video Coding/Archive 2

Apple iPod?
The Apple iPod and iTouch/iPhone all support a version of the H.264 codec called "low-complexity" baseline. Even though this is not an approved profile, it would seem documenting this is pretty important given the ubiquitos nature of the Apple products. —Preceding unsigned comment added by 131.107.0.101 (talk) 18:04, 11 September 2008 (UTC)

Reply: However, this is a wiki about the standard of h.264. The standard does not define a low-complexity baseline profile. It may still be worth mentioning. ALSO there is now the Constrained Baseline Profile (CBP) listed as profile, which is as well not a part of the specifications. AFAIK. If someone could point me to the ITU-T h.264 document and page where the CBP is described? —Preceding unsigned comment added by Oliver.stampfli (talk • contribs) 02:17, 9 December 2008 (UTC)


 * Regarding the "Constrained Baseline Profile", please see JVT-AC204. —Mulligatawny (talk) 20:37, 9 December 2008 (UTC)

See M4V for details. ToxicWasteGrounds (talk) 17:39, 18 January 2010 (UTC)

CopyVio?
http://mediacoder.sourceforge.net/wiki/index.php/H.264 seems very similar —Preceding unsigned comment added by 98.207.199.32 (talk) 17:51, 1 April 2008 (UTC)

QuickTime H.264 capabilities are out of date
As of QuickTime 7.2 and later, additional H.264 features such as 8x8dct and other aspects of High Profile are now supported. I've added a couple of these myself, but a review would be good - there may be other H.264 features QuickTime now supports. - 209.91.141.250 (brad) 7 February 2008 EDIT: Now it's horrendously outdated; Quicktime is version 7.4. I did another update on Quicktime capabilities, but we need an expert here. June 28, 2009

Much too technical
There is such a thing as too much information. I'm sure videophiles would be thoroughly interested in reading about logarithmic step size control and macroblock-adaptive frame-field coding, but the average wikipedia user would leave this article feeling like they've learnt nothing more about H264. This article needs to be more like the WMV article, so that the user's most common questions (i.e. what is the quality like, how do I use it, where can I play it) can be answered. This article doesn't even have anything on subjective quality. —Preceding unsigned comment added by 58.107.140.52 (talk) 03:55, 18 December 2007 (UTC)


 * Seconded. I added a  tag to the article.  Could an knowledgeable individual please generalize it a bit for the mass audience? -Clueless (talk) 02:49, 15 April 2008 (UTC)


 * Wrong. It is a great article.  You guys just need to learn more about this area to be able to understand it.-anon  —Preceding unsigned comment added by 71.131.8.26 (talk) 19:56, 18 July 2008 (UTC)


 * I agree with anon. I think it's a darn good article, and I fear what an "improvement" might look like. —Pawnbroker (talk) 18:27, 19 August 2008 (UTC)


 * "You guys just need to learn more about this area to be able to understand it." Isn't that what Wikipedia is for? Isn't your argument a little circular? This article is far too technical for a mass audience. —Preceding unsigned comment added by 198.147.10.56 (talk) 20:55, 9 October 2009 (UTC)
 * The article really needs to stripped and summarized.  Mahanga Talk 19:18, 21 January 2010 (UTC)


 * The technical details have been very important for me to learn, so I do not think it should be rewritten for less technical people. However, I do think this article could use additional sections (e.g. subjective quality) near the beginning that less technical readers would find useful. In the end, wikipedia is supposed to be for everyone, both non-technical and technical, so articles should serve as many people as possible. —Preceding unsigned comment added by 192.160.117.139 (talk) 01:40, 29 November 2010 (UTC)

ProSieben
ProSieben stopped broadcasting HDTV this year in february. —Preceding unsigned comment added by 82.83.90.225 (talk) 11:51, 24 February 2008 (UTC)

Hardware Requirement?
If your computer doesn't have at least a Pentium 4, the video will not play, or play right. Why is this not mentioned? —Preceding unsigned comment added by 75.185.14.219 (talk) 23:33, 12 March 2008 (UTC)
 * Because such a statement is completely incorrect, given that H.264 video can range in bitrate and resolution nearly arbitrarily. In terms of CPU requirement for playback, H.264 is probably not even 50% more than that of MPEG-4 ASP at the same bitrate. — Da rk •S hik ari [T] 02:23, 14 March 2008 (UTC)

codec's don't work that way, as u could easily decode it on any decent GPU regardless of your cpu. i've seen someone decode it on a Ati AGP GPU with a 733mhz Pentium3 @1080P~25mbps. video decoding can be completely done without a fast CPU, as long as there is some form of co-processor. Markthemac (talk) 06:38, 19 January 2009 (UTC)


 * It really depends on the video file and the way it was encoded. I've played H.264 files from a 040 Color Classic to a Mac Pro, and variety of computers in between. The 040 Color Classic isn't going to play a file that's great by any shot (think like 80 x 60 encoded at 56 kbps video/8 kbps audio or so). Still, I know a good many people who think they know how to do video on a computer unable to open MP4 files, let alone MP4/M4V files encoded with the great H.264. Sierraoffline444 (talk) 21:36, 5 May 2010 (UTC)

I think something more should be mentioned about hardware requirements. My laptop has a Pentium 3 CPU and it can not decode H.264. Fewer and fewer Youtube videos is playing, the majority is now only playing sound without motion pictures. The video card has 8 MB of dedicated video RAM, I don't know if it has some GPU on it. This laptop does not meet the hardware requirements and I would like to know what those requirements are. —Preceding unsigned comment added by Urbanus Secundus (talk • contribs) 00:14, 29 June 2010 (UTC)

Availability of the codec
The article doesn't answer the question where the codec is actually used. In fact, I think it is deliberately avoided as an issue, because the public isn't supposed to be able to get it from free sources, outside formal markets. It seems that the only consumer source is the commercial Quicktime Pro, but that isn't mentioned either. All other users are big corporations who have obviously bought hefty rights to use it. Someone fix this. EDIT: I found that the open source community's free Avidemux is able to save x264 files, which is somehow different from H.264, but still the same quality. Teemu Ruskeepää (talk) 07:35, 6 July 2008 (UTC)


 * Actually there's a lot of information in the article about where the codec is used. For example, see the "Applications" section of the article.  There used to be more such information in the article, but people complained that there was too much of it, so a new article was created that's listed in the "See also" section: H.264/MPEG-4 AVC Products and Implementations.  The remark about Quicktime Pro being the only consumer use of the standard is completely wrong.  I believe the remark about x264 being substantially something different than H.264 is also wrong -- as far as I know, x264 is an implementation of H.264. The topic of how the standard is used by consumers isn't avoided - it's just that the purpose of the article is to provide information about the standard itself, not to list a bunch of implementations, to discuss the role of big corporations in modern society, or to promote some kind of open source agenda.  —Pawnbroker (talk) 18:16, 19 August 2008 (UTC)

Chart on decoder
There is a chart on encoders, but not one on decoders. Can somebody start or add one. Decoders are of more interest to most people watching content. —Preceding unsigned comment added by 71.131.8.26 (talk) 23:07, 18 July 2008 (UTC)
 * All the variety is in the the encoders, the only place where decoders could differ would be decoding performance or platform support, not much scope there for a chart - Xedaf (talk) 21:43, 27 August 2008 (UTC)
 * That's actually not correct. Each "Profile" of the standard corresponds to a separate defined set of decoding features, as well as being a set of defined syntax that an encoder can use.  See the chart of features supported for each Profile. In practice, there is also somewhat more variation in the decoding capabilities of various products than what is reflected in the profile/level structure that is specified in the standard.  I suspect that the chart of encoder product capabilities is actually (at least partly) a chart of decoder capabilities. Pawnbroker (talk) 17:07, 2 September 2008 (UTC)

Infobox
There should be an info box about this codec to summarize it, and it should contain information about licensing and patents, so that this information as well as other things are available with a quick glance (the point of an infobox). Yfrwlf (talk) 20:07, 23 July 2008 (UTC)

Further ther is no mention of frame accuracy or use in editorial pipelines. I believe it is not frame accurate as it is a streaming codec. 09:15, 13 October 2009 —Preceding unsigned comment added by Bbeard78 (talk • contribs)

Penetration or Market Share
It would be great if this page, and others about codecs, either linked to or contained some information about how widely used the codec is. It would help in trying to figure out what codec to use to preserve a video. Granted, there are other factors involved, but one way to avoid digirot (by which I mean loss of information due to unavailability of software to decode it) is to select a widely-used codec. Szetlan (talk) 02:50, 11 December 2008 (UTC)

Software encoder feature comparison - multithreads ?
I suggest adding a new row into the table "Software encoder feature comparison". That is whether the codec can run in multiple threads. In todays world (2009 januar) most consumer CPUs (desktop PCs) are multicore. --Xerces8 (talk) 22:02, 26 January 2009 (UTC)


 * x264 seems to support at least 2 threads for encoding. --Xerces8 (talk) 14:52, 6 October 2010 (UTC)

Streaming license
Is it true that they will be able to cash in from streaming websites starting 2011? I read this article but am none the wiser. --87.162.37.135 (talk) 16:02, 27 April 2009 (UTC)


 * I'm not sure what's special about 2011, but the patent holders of the technology used in H.264 encoders and decoders can "cash in" any time they want if the software (or hardware) hasn't been licensed by them. Just like Visto "cashed in" (to the tune of 3.6 million) by suing the makers of BlackBerry devices (RIM). Now whether a patent infringement claim is actionable and whether any action will ever be taken are two different things. I'd wager that if the holders of the H.264 patents ever believe that their bottom line is being harmed more than they would gain from litigation over unlicensed use of their technology, we would see litigation pretty quickly. I don't think there are any specific patents that cover simply hosting / distributing H.264 content though, so websites should be safe (unless they are encoding uploaded videos to H.264 using unlicensed software). INAL, just my personal understanding. 64.234.67.2 (talk) 11:38, 24 May 2009 (UTC)


 * They've decided to charge fees starting from 2016. Initially, the cut date was the end of 2010. See http://yro.slashdot.org/story/10/02/03/1528242/MPEG-LA-Extends-H264-Royalty-Free-Period?art_pos=1 and http://www.mpegla.com/Lists/MPEG%20LA%20News%20List/Attachments/226/n-10-02-02.pdf (pdf file) Mariushm (talk) 16:29, 3 February 2010 (UTC)

RapiHD
RapiHD doesn't seem to be a software-only encoder. It should be listed under hardware assisted encoders. - xpclient  Talk 06:23, 2 August 2009 (UTC)

Update to the "Levels" section of the main page
Regarding the "Levels" section, it doesn't inform anyone on how to figure the maximum reference frames of an arbitrary resolution in conjunction with a chosen BP/MP/HP Level. I'd like to propose the following formula and include a reference to the h.264 standard for Chart A-1 which is necessary to calculate the equation. While technical, it isn't explained in the most technical detail, only the parts that are necessary to do the math. I often see the question asked "How many Reference Frames Can I use?" and there hasn't been a definitive answer; at least not on Wikipedia. Formula to determine Maximum Reference Frames: A more practical, and fool-proof way to interpret this would be: Example1: Level 4.1, 720x480 video: Example 2: Level 3, 720x480 video: Below is a table of appropriate maxDPB values:

References: [h.264 Spec] Ttmcmurry (talk) 19:14, 27 October 2009 (UTC)


 * I went ahead and committed this to the main document today at work. Ttmcmurry (talk) 01:08, 29 October 2009 (UTC)

Nowhere in the article is the concept of "Levels" defined. —Preceding unsigned comment added by 94.25.144.112 (talk) 16:45, 5 February 2010 (UTC)


 * I just added a definition in the "levels" section of the page. —Pangolin (talk) 04:57, 10 May 2010 (UTC)

Excel Forumula In order to make the forumla easier to digest and especially because this section doesn't explain what a macroblock is or how to determine how many macroblocks there are in a given picwidth/picheight dimension, why does it keep getting changed? Further, why is Excel's MIN function introduced? It isn't necessary: Excel's MIN returns the lowest number in a series of numbers, however if you look at the formula, there is only one number returned; it is not an array nor a matrix. In fact, the formula is looking for the sixteen lowest numbers. FLOOR returns the correct result regardless of a MIN function. The goal I had in mind was to make the formula practical and ready-to-use as-is. Consider these Excel forumulas: =MIN(FLOOR((8100/((800/16) * (448/16))), 1), 16) gives a result of 5 =MIN(8100/((800/16) * (448/16))) gives a result of 5.785714286 =FLOOR(8100/((800/16) * (448/16)),1) gives a result of 5 =(8100/((800/16) * (448/16))) gives a result of 5.785714286 As you can see, the forumla does not need a MIN function to return a correct result (5), again, because the MIN function is designed to show the lowest result of a sequence of numbers (array or matrix). For that reason I'm proposing to change it back to a more user-readable format that does the math properly. The alternate proposed is to explain how to determine a macroblock; though I fathom most won't read the spec and for the sake of simplicity we should think for anyone who wants to copy and paste the forumula.

Ttmcmurry (talk) 04:15, 10 June 2010 (UTC)


 * Above all, if we provide a formula, the most important thing is to make sure that the formula is correct. Here are some issues to consider:
 * The width and height of a coded frame in units of macroblocks must have integer values. Any provided formula must ensure that. For example, to produce a decoded frame size of 24×24 luma samples, the coded frame size must be 32×32 (4 macroblocks) or larger. Some of the suggested formulas do not reflect that. Instead those formulas are implicitly assuming that such a frame could contain only 3 macroblocks rather than 4 macroblocks (or more, per item 3 below).
 * When frame_mbs_only_flag is 0, the height of a coded frame in units of macroblocks must not only be an integer, it must also be an even number. If we try to compute the number of macroblocks in height by always dividing by 16, we would be neglecting that.
 * When non-zero frame cropping parameters are supplied (i.e., non-zero values of frame_crop_left_offset, frame_crop_right_offset, frame_crop_top_offset, or frame_crop_right_offset), a coded frame may contain more macroblocks than the minimum number that is needed to span the output width and height.
 * According to the formula in the standard, the DPB capacity requirement can never exceed 16 frames. Thus, when the number of macroblocks in a coded frame is small, it is necessary to apply a min function to ensure that the final output of the formula cannot exceed 16. The statement that "the formula is looking for the sixteen lowest numbers" seems incorrect. The function MIN(x, 16) will produce the value x when x is less than 16, and will produce the value 16 when x is greater than or equal to 16.
 * Because of these issues, I think there really isn't a simple (correct) formula of the form that you are seeking. Supplying a formula that sometimes provides an incorrect answer would be worse than not providing a formula at all.
 * —FalseAlarm (talk) 19:00, 10 June 2010 (UTC)

Constrained baseline profile
The text refering to constrained baseline profile "widely used" is incorrect. That profile was introduced as part of the March 2009 version of the H.264 standard, which would be way too recent for widespread adoption considering H.264 has been part of many video conferencing systems since about 2005. -- 64.254.110.34 (talk) 23:22, 5 January 2010 (UTC)


 * I removed that statement, although it may not have been completely incorrect. Although the formal profile definition for the "constrained baseline profile" within this standard is relatively recent, its definition consists of a subset of the tools features supported in the existing "baseline profile". In fact all "constrained baseline profile" bitstreams are also "baseline profile" bitstreams.  This is actually discussed in the standard (in a note in the section defining the "constrained baseline profile"). Both profiles use the same profile identification code number (profile_idc = 66).  The usage of the tool subset that is now defined as "constrained baseline profile" actually did exist before that subset was officially recognized as a profile of the standard. I clarified this in the article. -Mulligatawny (talk) 02:30, 6 January 2010 (UTC)

Putting the geekspeak at the bottom
What do you all think about putting all the tech stuff at the bottom and move patents, success and applications to the top? Seems more logical. I came here to read about the patent and had to scroll through a bunch of tech stuff. --Wikifier (talk) 23:15, 9 February 2010 (UTC)

So when do the patents end?
We know that in 2018 that MP3's Patents all end, so what about H.264? 24.240.67.71 (talk) 14:55, 13 April 2010 (UTC)

Lists.whatwg.org is currently linked for the proposition that some patents "may not" expire until 2028. Many of these putative 2028-expiring patents actually expire around 2018 suggesting that the author of the linked page has no idea of how to calculate patent terms and/or has made no real effort to do so. In any case, the linked page is nonsense. —Preceding unsigned comment added by 72.197.204.126 (talk) 03:35, 22 January 2011 (UTC)
 * You are right, some of the patent expirations are calculated incorrectly, since I did not take into account patent continuations or foreign patent priority data (not realizing at the time the importance of this data). See User:Jrincayc/Patent_utils  I haven't fixed this yet since I have not had the time to figure out if it is even possible to figure this information out via automatically parsing the patent data available from the US PTO's automatically gettable data such as http://patft1.uspto.gov/netacgi/nph-Parser?patentnumber=6,289,308  Those pages seem to be missing the term 35 U.S.C. 154(b) patent adjustment information.  Assistance with this would be appreciated. For example when and why does  http://patft1.uspto.gov/netacgi/nph-Parser?patentnumber=7,388,526 expire? Jrincayc (talk) 16:15, 13 February 2011 (UTC)
 * Removed "The last US MPEG LA patents for H.264 may not expire until 2028. " until I get a chance to fix it. Jrincayc (talk) 05:19, 19 February 2011 (UTC)

I updated the patent utilities User:Jrincayc/Patent_utils so now they assume that any related patent adjusts the 20 year term. I still am not handling foreign patent information. I will need to dredge through the law to figure out how that affects the term. Does 2025 sound right based on the new table? Jrincayc (talk) 03:59, 24 February 2011 (UTC) Incorporates more data. Some data however is not in the USPTO's html pages (such as patent adjustment info). Looks like there will be patents till at least 2023. Jrincayc (talk) 03:39, 3 March 2011 (UTC) Program updated again. Still has problems, but now tries harder to get correct related foreign and US patent dates, and tries to get the term extension as well. Jrincayc (talk) 23:20, 24 April 2011 (UTC)

Looks like it is Patent7826532 http://www.google.com/patents/about?id=2onYAAAAEBAJ that expires in 2027 that tails the pack. Shall I re-add something like that to the article? Jrincayc (talk) 02:50, 25 April 2011 (UTC)

Patent licensing
Shouldn't it be mentioned that Apple and Microsoft are among the companies licensing H.264 through MPEG LA? (see AVC/H.264 Licensors). It helps to see some of the reasons for promoting H.264 as Flash's successor. 92.251.247.226 (talk) 08:04, 3 May 2010 (UTC)

Is Commercial Use Permitted by the Patent License?
Since patents are mentioned in this article, isn't it true that all OEM licenses bought by camera manufacturers are limited to "personal use only"? (This is in the first paragraph of the SUMMARY OF AVC/H.264 LICENSE TERMS where it says, "sublicenses for encoder and decoder manufacturers granting the right to manufacture and sell and a limited right only for personal consumer use by or between end users") IANAL so I can't tell for sure, but if this is true then the use of cameras with this encoder would prohibit commercial work. If my speculation is true, then I think it should be mentioned in the article. R39525 (talk) 19:35, 13 May 2010 (UTC)

Free codec h.264 get fraudelous handling from some licence company asking money
I was already used into the h.264 by philips vesta and not changes from that time, they used it into the vesta webcam and was already freeware into 1999 when i buy this webcam, it cames with the program Ulead mediastudio pro 5.2 without any licences. So it a faboleus story that the h.264, is a new format and somebody earn money on it because saying he invented or developed it. Later they making probely an streaming audio codec by this avc. So they dont develope anything. They use something for free, change some code probely not into the video codec but in the audio stream and where going making money for it by putting a lincence fee on it. The company who do that have is not the owner or inventor from this codec. —Preceding unsigned comment added by 87.212.148.18 (talk) 08:14, 5 May 2010 (UTC)


 * Could you re-write/re-author/re-compose this entry, in English perhaps? Sierraoffline444 (talk) 21:38, 5 May 2010 (UTC)

Does H.264 have anything to do with DivX?
From the description, H.264 sounds a lot like the DivX codec, but this article makes absolutely no mention of DivX. Should someone clarify the relationship/similarities between DivX and H.264? —Preceding unsigned comment added by 212.104.143.20 (talk) 14:36, 24 May 2010 (UTC)
 * H.264 is a coding standard. A specification. A format. DivX is a brand name of software products made by the DivX, Inc. company, such as the DivX Codec, which is a software product that encodes and decodes video in the MPEG-4 ASP and H.264 formats. So you would be comparing apples and oranges. H.264 is not a codec (from the definition: "A codec is a device or computer program"), H.264 is format. And the DivX Codec is not a video format, it is a software product, a software implementation. There are many other MPEG-4 ASP and H.264 codecs, too (i.e. software or hardware products that encode/decode MPEG-4 ASP or H.264 video), made by various people and companies. The DivX Codec is just one of them (this also implies that they are compatible, because they all use the same video format). See for example the comparison of video codecs article. &mdash;J. M. (talk) 16:55, 24 May 2010 (UTC)
 * However, there is an implementation of H.264 in DivX known as DivX Plus HD which is usually used for HD video and the corresponding audio codec is 5.1 surround AAC. WhiplashInferno (talk) 07:04, 19 August 2010 (UTC)

x264 10 bit support
x264 now supports 9bit and 10bit color depth but it's preliminary work - the input is still just 8 bit and there are no assembly functions for these depths making everything very slow.

I'm not sure that it's right to say x264 right now has 10 bit capability as it's obvious it's still a work in progress. —Preceding unsigned comment added by 89.137.96.153 (talk) 15:16, 6 July 2010 (UTC)


 * High bit-depth input is now available. So that's one these issues gone. J Darnley (talk) 01:05, 3 October 2010 (UTC)


 * You source please? Fleet Command (talk) 05:47, 3 October 2010 (UTC)


 * The source code of course! http://git.videolan.org/?p=x264.git;a=commit;h=284b3149b88f08c6ca324de05a92882c37fb1a44 J Darnley (talk) 15:08, 5 October 2010 (UTC)

Corresponding audio codec?
Which audio codecs typically get used along side H.264? Also, which container formats? SharkD  Talk  01:59, 19 August 2010 (UTC)
 * In the MP4 file format, AAC is usually the main audio format whereas with other containers (M2TS, MKV), Dolby Digital is a main format, though in M2TS, because it is the Blu-ray Disc container format, it can be used with any audio codec supported by BD. WhiplashInferno (talk) 07:01, 19 August 2010 (UTC)

Explain Intra profile
I was lost when I read about Intra profile, explain it or link to another page explaining it. It it related? http://en.wikipedia.org/wiki/AVC-Intra — Preceding unsigned comment added by OlivierToupin (talk • contribs) 17:03, 13 December 2010

Profiles
Missing from the profiles section is the relevance of the idea in real terms. Do content creators target a profile in which case that content will play back optimally on systems that also target that profile? What does a Profile, or anything else, mean when its definition omits how people engage with it? Rtdrury (talk) 23:47, 10 February 2011 (UTC)
 * That is a valid objection to the article. I think an expert needs to look into this. Fleet Command (talk) 22:02, 11 February 2011 (UTC)
 * I don't understand the question. —Mulligatawny (talk) 21:23, 15 April 2011 (UTC)

VCL
Just pointing out that the initialism "VCL" is used in the article, without any definition of it. I believe it is "Video Coding Layer", but is not defined or described anywhere. + m t  04:14, 15 April 2011 (UTC)
 * Fixed. It was only in one place (a table heading). —Mulligatawny (talk) 21:21, 15 April 2011 (UTC)