Super Class (220): development of digital video coding
HTTPS: // www.expreview.com/73428.html many complex
Everyone waiting, this is the third series of multimedia file formats classroom article, have already spoken of the container and audio encoding, and now we want to see the most complex video coding, people always find ways to improve video coding efficiency, to provide the best possible picture quality in a small volume, to meet the demand for video transmission, storage. And the first two articles in a different container introduction of audio encoding, the video is encoded with a clearer development context, and not many more than the kind of unified audio coding easier to rationalize the current internationally accepted standard video coding basically by MPEG (moving picture Experts Group) and ITU-T (International telecommunication Union telecommunication standardization sector) and other organizations led development, and there are some sporadic coding, they may occupy the mainstream position within a short period of time, but eventually gave way to internationally accepted standards.
Prevailing on major international coding standard ITU-T organization H.26x series of video coding and MPEG coding standards organizations to develop part, one thing should be noted that the same standard in a different organization there may be called a different name, the most typical example is the AVC (advanced video coding), you may be more familiar with its other name --H.264, AVC MPEG organization is in the standard give it a name, MPEG organization is affiliated with the international organization for standardization (ISO) and the international Electrotechnical Commission (IEC), so the ISO standard, its official name is "MPEG-4 Part 10, Advanced Video Coding". In this case more common coding H.26x series, will hereinafter spout.
In this context the main, basically encompasses nearly half a century since, the development of video coding technology, we will mainly along this main H.26x and MPEG context, as our readers sort out a simple development of video coding course.
Why do we need the video compression encoding?
It is simply to reduce the size of the video capacity is occupied.
Digital video is essentially a continuous frame images, although the size of an image is not large, but must have at least 24 images (generally) per second, which would add up to occupy a very large space, we do not have and more places to store raw data, then only one way to go, compress it. The video encoding process is this compression process, but with the same audio, video files inside look at the algorithm is essentially no redundant information in a conventional data compression, so it is necessary to develop algorithms for video compression, the actual to remove redundant information exists, thereby reducing the amount of data which, to achieve the purpose of reducing the occupied capacity. Therefore, the current video encoding basically lossy, after the encoded video means that there will be a loss in picture quality.
Video coding development before the age of Blu-ray
Let us first along international standards, in chronological order to look at video encoding is how step by step "modernization".
In the era of analog television and in film, we see the contents of all analog signals to restore them. But with the continuous improvement of people's needs, and the computer network to flourish, we need a new, digital code capable of carrying video content to support video content delivery on the Internet, or stored in a digital store device.
When the late seventies and early eighties, people have developed a compression algorithm multimedia content for many new images, this time to develop digital video encoding conditions are ripe, and the first to develop the actual coding, which is later played a leading role in the field of digital video coding video coding Experts Group ( V IDEO C oding E Xperts G roup), was the name they still ITU- "international telegraph and telephone Consultative Committee (CCITT)" of T (International telecommunication Union telecommunication standardization sector) organized the following expert group. This code is named H.120, which was born in 1984, is a biased in favor of early coding experimental nature, mainly based on differential PCM coding is used to save TV content, but it does not have large-scale practical application.
H.261: the introduction of a variety of characteristics, laid the foundation of modern video coding
After a few years in the development of complete H.120, VCEG they did not study in the above video encoding stop. At this time, many multinational companies have used the network video conferencing needs, and in Internet bandwidth is not sufficient era, people need a new video coding to achieve smooth and high-quality real-time video communications, H.261 is adopted.
H.261 and first digital video coding standard H.120 is not directly inherited, it can be said to be a completely fresh start coding. In the algorithm for image compression, H.261 uses our now more familiar with the discrete cosine transform (DCT) algorithm, which play a major role in the subsequent JPEG encoding. But more than that, it introduces a set of features for the video, laid the foundation for modern video encoding, the main ones macroblock (Macroblock) and motion compensation (Motion Compensation) macro blocks.
Inter-predicted macroblock based motion compensation
We know that the video is a combination of an image consisting of one, the next second video generally will contain 24,25,30,60 or more images, they play out in a certain interval of time, based on Persistence of vision principle to form a smooth, will be moving screen. Between several successive frames, there is actually a large number of duplicate images, such as the following example:
Green billiards table in a white sports top
A ball movement direction and distance to describe the image change
If based on traditional ideas of each frame of image compression to do so, the entire video is obviously still a lot of redundancy after compression. So how to do it? H.261 standard introduces thinking macroblock, the whole picture will cut into many small pieces, and then the introduction of inter prediction based on motion compensation - mostly pictures do not move, then we will not move part follow block before compression results, plus the distance the movable part with the direction of movement of such a vector is not described can save a large amount of storage space it?
DCT algorithm
The 8x8 block of pixels into a
DCT algorithm originated in the 1970s to the late 1980s, researchers began to be used for image compression. This algorithm may convert the image from the spatial domain to the frequency domain, then quantized to do - to reduce the sensitivity of the human eye is less high frequency information, retained most of the low frequency information, thereby reducing the volume of the image. And finally with efficient data coding the data after the processing is further compressed, as used herein, the Zig-Zag scanning and variable length encoding.
Note: there are a lot of high-frequency portion of the image details, while the low frequency part of a larger profile and other information the coverage there.
Luminance channel do DCT transform image, the color can be seen above the continuous portion is flat, while the bottom has many details
In the H.261 and H.261 video coding based framework of after, DCT compression algorithm is mainly targeted at key frames, called keyframes, motion compensation is used as a baseline reference. Analogy, like Flash animation keyframes, as it defines a starting point, follow a few frames are based on this key frame out of the calculation. Because it only intra-frame compression, and not other frames, also known as Intra-frame (intra-coded frame), referred to as an I frame.
Summary: the creation of hybrid coding framework, landmark
The design goal is H.261 coding the bit rate of the video in the range of 64 ~ 2048kbps, for real-time applications such as video telephony. It first establishes a framework for intra prediction coding and inter prediction used simultaneously, each of the frame itself there in eliminating redundancy, the redundant information is eliminated between the frame and the frame, thus greatly reducing the bit rate, It became the actual availability of a very high video coding. Almost all video encoding and its coding framework also affected after, especially H.26x and MPEG family.
It should be noted, H.261 specifies only how to decode only the encoder to produce the final video stream can be successfully decoded to all H.261 decoder. As for how the coding in front of you, with a specific algorithm does not matter how different, almost all of the video encoding after this point applies.
MPEG-1 Part 2: introducing the concept of a frame type, become standard VCD
Almost at the same time H.261 developed, in 1988, ISO International Organization for Standardization and IEC established two MPEG (Moving Picture Experts Group, M oving P icture E Xperts G roup) to develop internationally standardized audio and video compression encoding. They completed in November 1992 to develop a set of standard MPEG-1, wherein the second portion of a new standardized video compression encoding, which deeply affected H.261, inheritance and development of the block, motion compensation, DCT algorithm thinking, and made its own improvements, such as the introduction of a new bi-predictive frame, sub-pixel precision motion compensation and other new technologies.
Introducing bidirectional predictive frame (B-frame)
After the introduction of H.261 inter prediction algorithm based on motion compensation video frame actually has been divided into two categories, and one is complete, called keyframes (Intra-frame), it is a complete still image , it can be directly decoded. Another frame by motion compensation algorithm is calculated over the key frame.
MPEG-1 Part 2 introduces the concept of the frame type, referred to as the original key frame "the I frame", the inter-frame prediction based on calculated as P frames. H.261 has two outer frame types, it introduces a new frame: bidirectional predictive frame, also called B-frames.
Originally only the first P frame can be prediction, that is, only one can be obtained based on the previous calculations. Bidirectional prediction, by definition, it can be used as a front of the own reference, it can also be used for prediction behind that frame. Since the reference of additional information, B-frame itself can contain less amount of information, which would naturally even higher compression ratio can only do the unidirectional prediction frame P ratio. However, the introduction of B-frames brings a new problem, namely the difficulty increased Codec.
Introducing a sequence of frames ( G roup O F P ictures)
Frame sequence number is the combined image frames in sequential order, referred to as GOP. A head of a GOP is an I-frame, there will only be an I frame, it contains a reference reference image information of the GOP, followed by a number of P frames, B frames, they are all based on the I frame beginning after calculated.
The above describes a complete image on the GOP, one can see the I and P frames are separated by three B-frames. In practical applications, B-frame is indeed the largest number of frame types.
Sub-pixel accuracy motion compensated
Introduced in H.261 inter prediction accuracy of pixel-level, many motion block aim is inaccurate, this has been effectively improved in MPEG-1. They introduced a motion compensation sub-pixel level, can be described half-pixel level motion pixel block.
Summary: Successful successor
MPEG-1 successfully inherited the technical framework of H.261, and its effective supplement, so as to achieve a good compression ratio. In that it is generally still use VHS videotape era, MPEG-1 have been able to provide picture quality similar to VHS tape with a rate of 1 ~ 2Mbps. It also makes it VCD been selected as the standard in the world, especially popular in our country more than ten years.
However, MPEG-1 mainly for low bit rate applications, but in fact it behaves at high bit rate is not bad, then, MPEG soon launch its upgraded version, which is MPEG-2.
MPEG-2 Part 2 / H.262: DVD and (pre) digital television standard
Launched in 1994 in the standardization of MPEG-2 video encoding a novel, it was admitted in 1995 as ITU-T H.262, here we simply call it the MPEG-2. With respect to the 1993 launch of the MPEG-1, it does not much change, mainly been improved for DVD applications and digital era.
Interlaced
Interlaced scanning is also not on the outdated concept today, in the early nineties, this scanning method effectively reduces the data bandwidth required for video transmission. Usually we see the video screen mostly progressive scan (Progressive scan), for example video vertical resolution is 1080 pixels, the vertical resolution of each frame of the screen is 1080 pixels.
And interlaced scanning, every other line by definition is a sweep, it will split each frame into two fields, each field to retain half of the original information frames. This scanning method to ensure the fluency of the screen while reducing the demand for transmission bandwidth, television broadcasting system has been adopted by all countries to use. MPEG-2 in the development of fully taking into account the needs of the digital television system, added support for interlaced.
And SD for high rate, high-definition
From the 1990s, digital television system gradually began to spread, it brings greater transmission bandwidth. Meanwhile, DVD standard is about to dust settles, it offers several times larger capacity than CD, capable of carrying more clear picture. Thus, MPEG-2 to enhance their target bit rate range, actually multiplied pride from 1 ~ 2Mbps MPEG-1 times to 6 ~ 10Mbps, even in the era of high-definition, it is possible to use a transmission rate of about 20Mbps HD picture.
Summary: Once the most common video coding
MPEG-2, although not a member of too many new features, in terms of the actual compression ratio is not much improvement, but since it was selected as the standard encoding DVD-Video, digital TV, DV and so on a range of applications, successfully became prevailing worldwide video encoding format, to date, it is still extensively used in digital television systems.
H.263: a good partner FLV and 3GP
The original H.261 and MPEG-1 is biased in favor of low bit rate applications, with the rapid development of Internet and communication technologies, the demand for online video is improving, the pursuit of higher quality video at low bit rates become the new target, while makers as a major communications industry standard, ITU-T H.261 launched a direct successor --H.263 in 1995.
H.263 has multiple versions, introduced in 1995 in the first edition, it is mainly the introduction of sub-pixel precision motion compensation in MPEG-1 began to apply the same support to 1/2 pixel accuracy. In addition, it improves the DCT algorithm, adding a new motion vector median prediction, in H.261 coding efficiency improvement compared to the more obvious.
It should be noted that the above characteristics are only part of its foundation, only need to implement these new things even support H.263, but it also gives a number of additional, feature enhancements for the compression ratio, for example, in MPEG-1 in the new B-frame, to become the additional H.263 PB-frames.
H.263 is a coding is escalating, there are two H.263 + and H.263 ++ upgraded version of the official after the first edition. In H.263 +, which focuses enhance compression ratio, relative to the first edition of 15 to 25 percent of the overall lift. At the same time in the 2001 revision, it also introduces the concept of "Profile", the various characteristics into several levels, full support for a certain level of support that is characteristic of this Profile, for example, part of the basis of the first edition of H.263 it is the "Baseline" Profile.
H.263 get in the Internet and communications industries are widely used, it was active in a variety of video sites above, and Flash Player propped up one day Internet online video together, and in the communications industry, adopted by 3GPP organization become a standard video encode multiple communication standards, such as MMS-- is MMS.
Further it is organized with reference to MPEG, MPEG-4 Part 2 as a basis.
MPEG-4 Part 2: many features achieve many
After MPEG-2, MPEG organization has a new goal - to develop a higher compression ratio coding framework, while retaining support for low complexity. In 1998, MPEG-4 standard was born, where the second part of the definition of a new set of visual coding system, yes, it is not intended only for video applications, but visual (Visual) in a broad sense, it is also known as MPEG-4 "Visual".
In fact, its core design and H.263 convergence, but contains more about the enhanced coding efficiency. It defines a plurality of different complexity Profile <br>, from the most basic to the highly complex Simple Profile Simple Studio Profile, do not support B frames former, the latter or even to support 4K resolution and 12-bit, 4: 4: 4 sampling encoding.
Although the MPEG-4 Visual coding is an ambitious, but it was cold and to be criticized by the industry. It is a compression ratio compared to MPEG-2 is not a major upgrade, but due to licensing and patent costs, many manufacturers choose to implement a compatible MP EG-4 Visual coding, rather than the direct use of standards, which the classic DivX and Xvid have two brothers use as basis for Microsoft's Windows Media Video, a little upgrade to WMV9.
Other coding
Time has come to the twenty-first century, high-definition video and high-definition television began to spread, new applications resulting in higher demand, forcing the industry began to study new and more efficient video encoding, we know AVC coming soon, but in the presentation before it, we take a look at a few other more video coding applications.
MJPEG
JPEG surely we are all familiar with, are inextricably linked between the MJPEG with JPEG. The video is not a one of you? That each frame is compressed with JPEG, and then combined not on line yet? Yes, MJPEG is a combination of JPEG images, each frame contains a complete image information, and as such, its compression rate is not high, but simple to implement features so many digital camera manufacturers use it as a camera's video encoding in fact it has been fairly widely used.
RealMedia
For the people, RealMedia is definitely a word with feelings. RM series coding their home over a decade ago in the domestic network had considerable coverage. In fact, it's basically achieve the same period of reference from the international standards, such clarity and very high compression ratio, pressure over the same period of rv40 DivX is a reference to the H.264 formed.
RM biggest problem is not broad range of support, play RM require plug-ins in the browser, the rise of video sites based on Flash Player also use it gradually becomes narrow and eventually under the impact of the genuine H.264, RM slow slowly disappeared.
WMV
Microsoft has its own dream living room, in addition to the Xbox, they want the PC into the living room, of course, nothing to do with Windows Media Video. Microsoft based on MPEG-4 Part 2 to create a new set of coding, at first they were called Microsoft MPEG-4 or Microsoft ISO MPEG-4, but soon, it will be included in the Microsoft Windows Media family first version is WMV7.
Microsoft's next on the basis of the above WMV7 continue to join their own thing, so that it can adapt to the higher-resolution video, and finally, they added new Profile in WMV9, produced a new VC-1 encoding.
Blu-ray era standards battle
After the popularity of DVD, the era of high-definition video soon arrived. It was soon discovered that even double-DVD, its capacity for 1080p video, it is completely not enough. Soon, large companies developed two new Blu-ray disc with a laser beam, one is led by the DVD Forum developed HD DVD, led by Sony and the other is a fresh start of Blu-ray. Two kinds of disc format war that we refrain here want to say is, along with the emergence of new video coding standard with a new disc format --VC-1 and H.264.
AVC / H.264: a master dominate the arena
HD DVD and Blu-ray standard in a total of three video encoding support, one is the old MPEG-2, and the second is Microsoft's main push of the VC-1, the last is the new AVC. Although its name is very simple, in fact, it is a great background, MPEG and ITU-T are the two organizations jointly launched a new generation of international standards, there is the MPEG specification as MPEG-4 Part 10 Advanced Video Codec, in ITU-R where it is standardized as H.264.
H.264 for the name, I think we should all be familiar. But is this now we are able to come into contact with every day video encoding formats, has over a decade ago triggered a crisis of software solutions, the next time many mainstream CPU triumph over the horse, but also makes ANI three are joined in their products assisted decoding hardware acceleration unit, but it does not matter with our theme, for the time being refrain from that. Talk about, or severe between H.264 here, what with what it means to be able to make the leap in quality above coding, which monopolize the market more than ten years have not shown signs of recession.
Conclude the following main points: more flexible macroblock division method, a larger number of reference frames, more advanced prediction and intra-frame compression ratio higher data compression algorithm.
More flexible method of dividing a macroblock
Previous standards, the method dividing a macroblock is fixed to 16x16 pixels of a macroblock. But in the new era, this extensive classification method is not flexible enough, so H.264 while allowing 16x8,8x16,8x8,8x4,4x8 and 4x4 these higher fineness divided manner. While the H.264 motion compensation sub-pixel accuracy from the description of the half-pixel precision down to the extent of 1/4. As a result, in the inter prediction new encoding with higher accuracy, but the actual amount of data and it does not add much to improve the compression ratio.
Larger number of reference frames
In the previous standard, each P or B frame may reference a limited number of frames and the number is too small, H.264 stroke to limit the extent of 16 to relax, and in most application scenarios, each frame may reference number of frames have at least 4 to 5, while in the previous standard, P-frames only with reference to one, B is 2 frames. This feature can improve the picture quality of most scenes, or decrease the volume.
More advanced intra-frame compression
Each macroblock prediction mode information contained
For the I-frame, H.264 also introduces a new compression method. In general, for a given pixel in the image, and it is close to the color of adjacent pixels is small gap, so we can use this feature to further reduce the size of the individual images, how to use it? The H.264 pixel color variation within a single macro block specification became a formula, encoding just write here which formula applied on the line. Of course, here I expressed relatively simple, complete intra-prediction is very complex, H.264 macroblock 4x4 specifies the 9 prediction modes for 16x16 macroblock luminance plane predetermined four available modes. Greatly reducing the amount of data a single frame of the image, while maintaining high image quality.
Adds the predicted differential image can restore the original image information
CABAC
In the final phase encoded data when lossless compression, H.264 encoding in addition to supporting H.261 in VLC existed, but added two new lossless data compression coding, one is an upgraded version of VLC --CAVLC the other is a higher degree of complexity of CABAC (reference hereinbefore adaptive binary arithmetic coding, C ontext-based a daptive B inary a rithmetic C oding).
CABAC entropy coding is also a major principle is replaced with less frequency data length coding away appear, and with a short high frequency encoding data replacement occurs, but it introduces more statistical optimization, and dynamic adaptability. Although more calculated in decoding, it is possible to save more than the CAVLC data amount, usually 10%.
Summary: enormous change is a huge success
In addition to the points described above, H.264 as well as a lot of new features, with the MPEG-4 Visual difference is that these new features effectively help H.264 has made significant progress in saving capacity. Let me give you an example of a strong contrast, DVD Video standard video, using the MPEG-2 encoding, bit rate of about 9Mbps, but its resolution is only 720x480, but in some scenarios we can clearly see damage resulting lossy compression; the same rate, in H.264 above, the carrier may have from 1080p video, and has a good quality.
In addition there is a major upgrade on the outer coding efficiency, H.264 for network transmission characteristics of coding organization is optimized to make it more resistant to packet loss and interference. Under various means, it has become the rule of the past decade the field of video encoding, and can say that it has become the de facto standard in HTML 5, and now you can hardly see a device does not support H.264 encoding, from the phone to the camera, from streaming video to Blu-ray Disc, it's a wide range of applications, high performance, even in the moment of the new code has emerged, it still has a strong vitality and irreplaceable, it is foreseeable that, H.264 will for some time continued to rule the field of video coding.
VC-1: Failed challenger
After entering the HD era, Microsoft also follow the trend, has been upgraded to WMV9, adding new features for high-definition video, making it capable of 1080p HD video level, new code is the VC-1. Compared with H.264, VC-1 overall complexity is lower, and therefore on the software solution for CPU more friendly. In fact, VC-1 SMPTE also through the International Organization for Standardization.
VC-1 has a certain relationship with the bundled HD DVD, Blu-ray at the beginning of the war has also been a certain amount of promotion in this way. However, with the defeat of HD DVD camp, VC-1 also will disappear, very difficult to see.
UHD streaming media era, the rise of new coding
H.264 is very powerful, but it took a little less in the ultra-clean era. With the leaps and bounds upgrade video resolution, H.264 showed signs of fatigue, it is in dealing with 4K video has no way to provide a good compression ratio. Clearly, people need a new code to inherit its location, and its immediate successor --HEVC, after years of research, and finally in 2013 was passed.
HEVC / H.265 / MPEG-H Part 2: Video coding throne successor
The HEVC, stands for High Efficiency Video Coding ( H IgH E fficiency V IDEO C oding), similarly, it is also by the MPEG and ITU-T international standard developed jointly encoded. Is included in the MPEG-H specification, for the second part (Part 2), ITU-T in there, which is a new member of the family H.26x for H.265.
HEVC mainly for high-definition and ultra-definition resolution video and development, compared to the previous generation AVC, it has better image quality at low bit rate, while in the face of high-resolution video, also provide ultra high compression ratio help 4K video stuffing Blu-ray Disc.
Instead of coding the macroblock tree units
HEVC introduces new coding tree unit ( C oding T REE the U- nits) concept, a substituted off years macroblock concept present in video encoding, which monolith area much larger, reaching 64x64, but still retains the can variable size and split feature, the smallest unit of 16x16. Coding tree contains a single coding unit small, they may be presented in the form of a quadtree, and quickly can determine whether the unit which can be subdivided, inside the minimum coding unit may be divided into 8x8 size, level of granularity is still It is very high.
Single coding unit may continue to be cut, sorted, may be a prediction unit (Prediction Units), which may indicate the form of the prediction unit is intra prediction or inter-picture prediction or even no change can be skipped unit; can also be a conversion unit (Transform units), which can be converted or quantized DCT.
Introducing HEVC coding tree so that both units with a large area of the cell to improve the coding efficiency also can be refined when needed, to retain finer detail. The so-called rough place on the rough, this place is fine fine, HEVC in its help to make more efficient stream.
More efficient DCT
Since a large area of the maximum block, then the DCT algorithms need to keep the job, will expand the HEVC DCT algorithm maximum size of 32x32 to the point, the image changes relatively flat portion, it has a higher compression ratio.
33 kinds of the intra prediction direction
Remember to write the above H.264 4x4 macroblock introduced nine intra prediction direction it? HEVC directly to this figure raised to 33 species in the repression of a still image not only to achieve greater efficiency, but also to achieve a higher accuracy, which is a great tool which successfully reached the market still image coding. Although the coding difficulty is high, but as long as there is no hardware encoder so many problems.
Summary: efficient coding, but trapped in the high patent fees
Compared to AVC, HEVC coding efficiency at high resolution but also a very large increase, for instance, the same period of 4K video, using H.264 encoding size may be large than HEVC out of a double. This huge rate of progress also makes Blu-ray use it as a standard code directly, launched UHD BD, while it improved compression above the single-frame image is also better than it has the ability to JPEG, so we see the mobile terminal more and more devices to select it as the default video, photo output coding.
But compared to AVC, HEVC promotion a lot slower, it is a high degree of difficulty than the H.264 codec too much, but this is by the brightest hardware encoder and software optimization gradually dissolve out, the current common basically device supports hardware HEVC codec; the second is the HEVC high patent costs, it is not a free encoding format, although it is no problem for personal use, but you want it compatible manufacturers, this pen high patent fees enough for them to be deterred, particularly advocating a free and open Internet market. Thus, we see that many manufacturers have chosen free and open VPx series encoding, as well as series of successors --AV1.
VPx series with the AV1: to free as a selling point
VPx series coding actually has a long history. It is the predecessor company On2 Technologies TrueMotion series of video encoding, in the development of TrueMotion VP8 encoding, the company acquired by Google. In Google's intervention, VP8 from the original proprietary technology into an open technology, open source under the BSD license.
From a technical point of view, VP8 technology used is similar to that of H.264. Although we have seen in the publicity, VP8 have a better than H.264 compression efficiency, but in practical applications, because it has some flaws in the design, performance is not as H.264, although it eventually entered the Web standard, but have not seen people use it, it is extracted from the WebP technology by its intra-frame compression welcomed.
VP8 performance is not ideal, Google will soon launch its successor --VP9. This time, they are referred to HEVC, the design goal is also efficient coding at high resolution. VP9 in some of the design is influenced by the HEVC, such as superblock same maximum of 64x64 (Super Block). The final result is reached VP9 than VP8 offers up to 50% efficiency improvement. And it looks like it can HEVC par, but it also encountered similar problems and VP8, promotion is not open. VP9 actual scope of application is also limited to Google's own Youtube, I can only say is a lack of practical application scenarios.
But soon, some manufacturers recognize the evils of HEVC high patent fees, they decided to create an open alliance to promote open and free media coding standards. This alliance is the Open Media Alliance ( A lliance for O PEN M EDIA), there is a founding member of the Amazon, Cisco, Google, Intel, Microsoft, Mozilla , and those of us familiar with Netflix large companies, followed by the addition of apples there, ARM, Samsung, NVIDIA, AMD is also familiar with these companies.
Google is also developing in their contribution to the VP10 out as the basis for a new coding Union, and soon, coded named AV1 born. In Facebook's tests, which were stronger than VP9 and H.264 on 34%, 46.2%, this looks really reach the level of the HEVC.
In these two years, AV1 indeed starting to get the attention vendors, such as Netflix recently been identified as the main coding to be used AV1, while Intel also introduced a free open source SVT-AV1 encoders, make full use of their own AVX - 512 instruction set . But this alliance is still quite loose, for example, one of the members of the League of Apple, currently AV1 is simply indifferent to its devices all turned HEVC.
But Netflix decided to use as the main coding AV1 from this attitude of view, AV1 free, open nature still has considerable appeal. But in terms of hardware is the lack of support for it, not only for AV1 do not end PC hardware decoding, more number of mobile devices does not fit, just recently there is a propaganda was the first to join AV1 hardware decoding SoC was released. Compared to hardware support for a more complete HEVC, it would be a threshold on the way of promotion of AV1.
Future encoding: VVC
Currently MPEG and VCEG has begun to study HEVC's successor, and now we know that the message is that it is temporarily named Versatile Video Coding (versatile video coding), and will be H.266. It is oriented to the future of video encoding, will support from 4K to 16K resolution video compression, and support for 360 ° video, its goal is to enhance the effectiveness of HEVC encoding on the basis of double.
New features are added in the future it may be: more complex coding section; larger, more detailed block division; global frame of reference; more intra-prediction mode (there are currently 65 kinds) ...... in complexity above, compared to HEVC, VVC will direct a higher dimension. But faced with the current international standards for the AV1 open standards on behalf of challenges, hard to say they will cancel some features, which will be officially released it's time to advance.
Summary: Times
Display with a resolution of more and more television, more and more network bandwidth, device for multimedia content processing power is growing, video coding has been changing with the times and continue to progress forward, but its frame from H .261 beginning, never had major changes, but each new coding are to join the new, more efficient algorithms take advantage of the growth of semiconducting properties in this established framework. Compared to progress is not obvious audio coding, video coding in new bandwidth capacity provided above saving effect is much more significant, and even update the code on quality performance is also an advantage. In the near future, 10-bit color depth and HDR will be popular, replacing lost in the last century is still a fundamental standard SDR content, bring us more exciting visual experience. Such as HEVC encoding actually prepared for it, in the future, their scenario will exceed even video field, with regard to the new Apple device, for example, it has become a real HEVC standard encoding format, spoken in the field of image and video in.
In addition, according to the latest report, the current accounts for the bulk of Internet traffic is video traffic, along with streaming media continue in-depth daily life for a greater flow of video transmission only, while the overall bandwidth of the Internet is not unlimited upgrade, for content providers, the traffic costs are also considerable portion of the cost, better efficiency compression encoding naturally by their favor. In fact, coding escalating this thing is a win-win, users and content providers can profit from it.
Because of the time and personal capacity constraints, this article also has many shortcomings, but I still want an overview of these codes to let more people know the correct coding knowledge, if we can play a role in better ideas, so that more people interested in coding, start your own research, it is the best thing.