Column
2004/07/09 Renewal of 20:29
Does it begin to move by "H. the 264/AVC improvement version"?
The situation of a next-generation optical disc (1/3)
H.264/AVC was recognized by the ad BANSUDO codec (high compression codec) by DVD Forum which promotes HD DVD. The Blu-ray Disc camp made required [ MPEG-2 ] is also advancing evaluation to HD image with the improvement version "H. 264/AVC FRExt" of H.264/AVC. The next-generation optical disc relation which was the motion under the water surface until now began to become active suddenly by H.264/AVC FRExt.
Although H.264/AVC and VC-9 (codec for broadcast/distribution media carried out based on the video compression algorithm of Windows Media Video 9) were recognized as an ad BANSUDO codec (high compression codec) at the board of governors of DVD Forum on the other day, on the other hand, the Blu-ray Disc (BD) camp was also advancing evaluation of an ad BANSUDO codec uniquely.
It is "H. 264/AVC FRExt" that BD camp is evaluating and inquiring. It is the codec which improved H.264/AVC of the schedule recognized in the MPEG committee soon. "FRExt" is the abbreviation for Fidelity Range Extention, and the width of the animation for which H.264/AVC is adapted by this can be expanded. The merit of H.264/AVC specifically optimized by the image of SD resolution extends so that HD image may also be harnessed.
H. Extension which makes 264/AVC the optimal codec also for HD image
BD camp has said that not H.264/AVC but MPEG-2 are required, in order to harness the fine detail information on HD image until now. Refer to the coverage report of PHL. . Although the reason was [ H.264/AVC ] very excellent in the quality of image of SD image, it was that the detail of the domain where the space frequency which is the feature is high is lost about HD image.
Same examination/evaluation result has the circumstances by which it was shown by DVD Forum, consequently not only H.264/AVC but VC-9 were added to the ad BANSUDO codec of HD DVD. As for the compressive ability of HD image, by the test of those days, VC-9 were higher. Those days, each movie studio company of Hollywood says that it was carrying out to "Why quality of image is worse than MPEG-2 although it is the new codec which collected the latest technology" to H.264/AVC at the mouth.
The technology in which FRExt can wipe away "the unfavorable criticism of H.264/AVC" in such a HD image is incorporated.
Being extended by FRExt is the 8x8 transformer form and the quantization procession (Quantization matrix) which were introduced into a color sampling format, the number of sampling bits, and it by MPEG-2. Although HHI and SandVideo (Broadcom purchases) had proposed the proposal [ except a quantization procession ] at the end of last year, it proposed that Matsushita Electric Industrial added a quantization procession to this, and it was decided mostly that a new profile called a FRExt profile joins H.264/AVC. Now, what added 1 color sampling format by the addition is made into final specification, and it is formally decided in the meeting of JVT (Joint Video Team) performed closely that it will be H.264/AVC FRExt by which temporary recognition is carried out.
H. 264/AVC FRExt combines the high definition-ized tool proved [ profile / Main / of H.264/AVC ] by MPEG-2 so to speak. The same of dividing into the grid of 8x8 and performing transformer form is said of MPEG-2 and the method are the same (MPEG4 4x4), and using a quantization procession. As for these two techniques, VC-9 are used.
In addition, in H.264/AVC FRExt, since decoder specification changes, the decoder for Main profiles of H.264/AVC cannot be used, if it remains as it is. However, since it is the same, fundamental algorithm is said [ that it can respond to H.264/AVC FRExt only by making a design change of the about 10% of the decoder tip currently developed for H.264/AVC, and ]. In order that processing cost may hardly go up, there is no influence on performance.
H. FRExt which erased the demerit of 264/AVC
In fact, the day which demonstrates the sample of HD image compressed by H.264/AVC FRExt to the studio persons involved in Hollywood, and since I was staying in Los Angeles by chance, the writer had got the opportunity to confirm the ability by his eyes (beginning of May).
The movie "Erin Brockovich" known for there being much film grain (particle) is 1 A scene, and having been used for comparison is three kinds of the image (image for evaluation called DCI-A) by CG without grain, and a general film image (the image for evaluation and film grain which are called DCI-B are also mixed with a degree in the middle).
A codec is the reference codec of H.264/AVC besides H.264/AVC FRExt, and MPEG-2 (24Mbps(es) of a D-VHS emulation). The bit rate of an ad BANSUDO codec is four kinds such as 8/12/16/20Mbps. The image after original and compression is divided in the central point, and it evaluates by comparing by the technique of making one of the two a mirror image, and compounding him.
It is quite a KITSUI image that Erin Brockovich has quite much grain and has the scene where the printing type of books is reflected on the way etc. at a compression codec. The particle size of grain goes up by 8Mbps(es) truly, and, on the whole, it becomes NOIJI. Although it is considerably improved if it is 12Mbps(es), the difference from original is a little worrisome. However, when the bit rate was raised to 16Mbps(es), it became an image more beautiful than the D-VHS emulation of MPEG-2/24Mbps rather. The scene where the textures of a film and space frequency are high is also satisfactory at all.
If it becomes DCI-A without grain from a vine vine, a problem will not be felt at all by 8Mbps. There is also no impression to which edge faded and the difference with original will hardly be noticed also by 8Mbps(es).
Although DCI-B had the scene which paper snowstorm dances to a blue sky in the beginning, and sense of incongruity was felt a little for the whole including the bouquet of the bride who appears on the way when it was 8Mbps, it is almost satisfactory if set to 12Mbps. Although the distinction with original will become difficult if the bit rate goes up to 16Mbps(es), it is an impression better than MPEG-2/24Mbps also at 12Mbps.
MPEG-2/24Mbps is an impression only whose noise decreases H.264/AVC FRExt leaving fine textures and a detail although it is the tendency to compare originally and for noise of a screen to increase generally. 8Mbps(es) are enough if it is the animation material made from computer graphics. If it was a general movie, it was as a result of [ satisfying at 12Mbps(es) ] compression.
H.264/AVC FRExt is adopted as BD.
It seems that it, for the moment, does not consider BD camp re-compressing an image by the ad BANSUDO codec, and recording on videotape to BD-RE. The ad BANSUDO codec is not contained in the logic format specification for which BD-RE is used on it although one-layer 25 G bytes / two-layer 50 G bytes of BD-RE 1.1 was decided. Although BD-RE 1.2 of 2X writing is due to be announced also immediately next year, there is no schedule which puts in an ad BANSUDO codec at the time.
However, BD-ROM A former report Although it came out and I reported that ROM media were BD-RO, a certain ad BANSUDO codec is contained in writing it as BD-ROM here, since it is ROM by another data. To the present schedule, the logic format specification which contained the ad BANSUDO codec in the BDA general meeting in which selection of the ad BANSUDO codec adopted by BD-ROM also by the end of July is performed by the manager member, and will be performed in September, this year seems to be announced. Within the year, full specification including Java correspondence, network correspondence, etc. is determined.
In addition to H.264/AVC FRExt, VC-9 of Microsoft as well as HD DVD are likely to come up for discussion at the ad BANSUDO codec contained in the logic format specification of BD-ROM. H. Since 264/AVC FRExt and VC-9 have the technical feature alike very well, they also have a skeptical opinion about mounting both.
Since the profile for compressing in TARESU image sauce into VC-9 does not exist (at present), there is also a problem that the specification about materials other than a movie is indefinite. However, a way called adoption may also have both by political conclusion as approach against Microsoft of HD DVD slippage.
Anyway, the ad BANSUDO codec which can realize the quality of image MPEG-2 / more than 24Mbps equivalent by 8 - 12Mbps will be carried also in BD-ROM. When H.264/AVC FRExt is used temporarily, it is possible to stuff 3 hour + audio 3 language +SD bonus image 6 hours into the one-layer disk of BD-ROM for HD image of 12Mbps(es). Or television series can also perform selection called a part for 18 hour in SD quality.
With the viewpoint of the videotape recording of high-definition television broadcasting, between the present, since MPEG-2 are likely to be used, although a two-layer disk is probably needed, as for package media, almost all goods may become enough [ an one layer disk ]. This will work advantageously in respect of disk manufacture cost.
However, the appearance of BD recorder which can respond to BD-ROM reproduction is postponed rather than the schedule said at the beginning.
HD DVD is also H.264/AVC FRExt.
At the beginning, the ad BANSUDO codec was due to be contained in the logic format of BD-ROM. However, evaluation of H.264/AVC FRExt is regarded as adoption of an ad BANSUDO codec having progressed steeply, in response to the fact that it was good. In addition, there is specification change about copyright management technology etc., and it is concluded before that the scenario "the hardware corresponding to [ determine ROM specification in June - July, and ] ROM reproduction with the year-end new product" about which Mr. Nishitani of Sony was speaking was changed.
It is because when it comes to ad BANSUDO codec adoption the decoder tip must be carried, and there is change of copyright management technology etc. and it cannot be adopted as a very year-end product. It is better for the hardware in which BD-ROM reproduction is possible to consider next year or subsequent ones (but a BD-ROM software package will be put on the market next year only after an end).
On the other hand, what does HD DVD after already opting for adoption of H.264/AVC become?
Mr. Yamada of Toshiba "adopting H.264/AVC and VC-9 as the ad BANSUDO codec of HD DVD by DVD Forum determined. However, perfect specification decision which profile to use is not made. Although it had determined to adopt MPEG-2 before completion of MPEG-2 also at the time of present [ DVD ], it is the same as it. Therefore, what is necessary is just to make H.264/AVC FRExt into the profile used by HD DVD, if they say that H.264/AVC FRExt is good. It was said that it carried out slowly, applying [ of such fine specification ] it within the year."
Although Toshiba was developing the decoder tip of H.264/AVC with SandVideo, SandVideo proposed ideas, such as 8x8 transformer form which became the origin of H.264/AVC FRExt, from the first. VC-9 use the same technology -- having -- てお -- "-- H. -- the change is as bold as" which is not (Mr. Yamada) to the schedule of HD DVD just because 264/AVC FRExt came out
Anyway, it is becoming deterministic that H.264/AVC FRExt is used for HD DVD and BD-ROM. Various new motions are also occurring in parallel to this. Although only the motion under the water surface was the subject of the intense next-generation optical disc relation the past several months, from now on, it will apply in autumn, and will begin to move to the further severe earthquake. I want to see an opportunity and to clarify little by little about subjects other than an ad BANSUDO codec.
