AMWA Releases MXF Commercial Delivery Specification
The Advanced Media Workflow Association (AMWA) has released a new MXF Commercial Delivery specification, AS-12. The constrained version of MXF has been developed to enable more efficient handling of commercials through the many transactional and media processing operations between conception and air.
As broadcasters look to serve commercials to long tail delivery platforms as well as their primary channels, controlling costs is all-important. Many versions may exist of the same commercial, adding confusion to the traffic operations. Versions may be sourced from different distribution routes, arriving with different wrappers and codecs, as well as different aspect ratios.
MXF Commercial Delivery aims to solve two problems: unique identification, and defining a master spot for the creation of long-tail versions.
MXF Commercial Delivery unambiguously identifies the spot through the Ad-ID unique identifier carried in a “digital slate”. Current practice to identify commercials is by the visual slate preceding the commercial. Although a human operator can read this by playing the commercial, it does not lend itself to use by automated systems.
With MXF Commercial Delivery the advertisement identification metadata is carried as a Descriptive Metadata track and serves as a digital slate. The digital slate can be used to reconcile the video and audio components of the commercial with the traffic instruction thus preventing expensive mistakes. The Ad-ID unique identifier ensures that what the advertiser ordered gets to air.
MXF Commercial Delivery, AS-12, is an addition to AS-03, MXF for Delivery. AS-03 defines MXF files optimized for program delivery, and intended for direct playout via a video server. Used together, the specifications allow the agency to supply broadcasters with a master commercial, along with information like closed captions and AFD that the broadcaster can use to create the lower resolution versions appropriate to their long-tail delivery platforms.
The AS-12 metadata or digital slate can be created early in the production process to uniquely identify the commercial. AS-12 carries fields to identify the advertiser, the agency, brand and product, as well as the title. Through the use of the guaranteed unique Ad-ID, the rekeying of identifiers that typically happens today is avoided. House codes used by agencies or broadcasters are replaced with the Ad-ID, avoiding many of the issue of misidentified commercials that have been commonplace.
The Commercial Delivery specification is sponsored by AMWA principal member Ad-ID, a joint venture of the American Association of Advertising Agencies (4A's) and Association of National Advertisers (ANA), and establishes a baseline for improved Operations, Administration, and measurement of advertising assets across the myriad of current and emerging delivery platforms, which when fully deployed, will result in substantial financial gains and improvements in productivity that will flow back to all participants within the supply chain.
Source: Advanced Media Workflow Association
MPEG Transport Stream
The MPEG-2 standard is defined by ISO/IEC 13818 as "the generic coding of moving pictures and associated audio information." It combines lossy video compression and lossy audio compression to comply with bandwidth requirements. The basic structure of all MPEG compression systems is asymmetric because the encoder is always more sophisticated than the decoder.
MPEG encoders are always algorithmic. The better ones are also adaptive, using a feedback path. MPEG decoders are not adaptive and perform a fixed function. This works well for applications like broadcasting, where the number of expensive complex encoders is few and the number of simple inexpensive decoders is enormous.
The MPEG standard provides little information about how encoder processes and operation. Rather, MPEG-2 specifies how a decoder interprets metadata in a bit stream. The metadata tells the decoder the rate the video was encoded, defines the audio coding, and identifies channels and other vital stream information.
A decoder that successfully deciphers MPEG streams is called compliant. The beauty of MPEG is that it allows different encoder designs to evolve simultaneously. Generic low-cost and proprietary high-performance encoders and encoding schemes all work because they are all designed to communicate with the compliant decoder base.
Stream Structures
An MPEG-2 stream can be either an Elementary Stream (ES), a Packetized Elementary Stream (PES) or a Transport Stream (TS). The ES and PES begin with and are stored as files. Individual ESs are essentially endless because the length of an ES is as long as the program itself.
Starting with analog video and audio content, individual ESs are created by applying MPEG-2 compression algorithms to the source content in the MPEG-2 encoder. This process is typically called ingest. The encoder creates an individual compressed ES for each audio and video stream. An optimally functioning encoder will appear transparent when decoded in a set-top box and displayed on a professional video monitor.
A good ES depends on several factors, beginning with the quality of the original source material, and the care used in monitoring and controlling audio and video variables when material is ingested. The better the baseband signal, the better the quality of the digital file. Also influencing ES quality is the encoded stream bit rate and how well the encoder applies its MPEG-2 compression algorithms within the allowable bit rate.
MPEG-2 has two main compression components: intraframe spatial compression and interframe motion compression. Encoders use a variety of techniques, some proprietary, to maintain the maximum allowed bit rate while at the same time allocating bits to both compression components. This balancing act can sometimes be unsuccessful. It is a tradeoff between allocating bits for detail in a single frame and bits to represent frame to frame motion changes.
Researchers are still investigating what constitutes a good picture. Presently, there is no direct correlation between the data in the ES and subjective picture quality. For now, the best way of checking encoding quality is with the human eye, after decoding.
The Packetized ES
Each ES is broken into variable-length packets. The result is a PES containing a header and payload bytes. The header includes information about the encoding process required by the MPEG decoder to decompress the ES.
Each individual ES results in an individual PES. At this point, audio and video information still resides in separate PESs. The PES is primarily a logical construct and is not actually intended to be used for interchange, transport and interoperability. The PES also serves as a common conversion point between TSs and PSs.
Both the TS and PS are formed by packetizing PES files. During the formation of the TS, additional packets, containing tables needed to demultiplex the TS, are inserted. These tables are collectively called PSI and will be addressed in detail later.
Some packets contain timing information for their associated program, called the program clock reference (PCR). The PCR is inserted into one of the optional header fields of the TS packet. Recovery of the PCR allows the decoder to synchronize its clock to the rate of the original encoder clock.
Null packets, containing a dummy payload, may also be inserted to fill the intervals between information-bearing packets.
TS packets are fixed in length at 188 bytes with a minimum 4-byte header and a maximum 184-byte payload. Key fields in the minimum 4-byte header are the sync byte and the Packet ID (PID). The sync byte's function is indicated by its name. It is a long digital word used for defining the beginning of a TS packet.
The PID
The PID is a unique address identifier. Every video and audio stream as well as each PSI table needs a unique PID. The PID value is provisioned in the MPEG multiplexing equipment. Certain PID values are reserved or specified by organizations such as the Digital Video Broadcasting Group (DVB) and the Advanced Television Systems Committee (ATSC) for electronic program guides.
In order to reconstruct a program from all its video, audio and table components, it is necessary to ensure that the PID assignment is done correctly and that there is consistency between PSI table contents and the associated video and audio streams. This is one of the more critical points in a MPEG-2 stream.
There are four other important fields in the TS header. One is the continuity counter. It is a 4-bit field that repeatedly increments zero through 15 for each PID. It’s used to determine if packets are lost or repeated PCR. Second is the discontinuity indicator. It indicates a time base (PCR) and continuity counter discontinuity, which allows the decoder to handle such discontinuities. Third is the random access indicator. It indicates that the next PES packet in the PID stream contains a video-sequence header or the first byte of an audio frame. Fourth is the splice countdown. It indicates the number packets of the same PID number to the splice point when a new PES packet begins.
PSI
During the formation of the TS, additional packets, containing tables needed to demultiplex the TS, are inserted. These tables are collectively called PSI. PSI is part of the TS. PSI is a set of tables required for demultiplexing and sorting out which PIDs belong to which programs.
To identify which audio and video PIDs contain the content of a particular program, a Program Map Table (PMT) must be decoded. Each program requires its own PMT with a unique PID value.
In order to determine which PID contains the desired program's PMT, the Program Allocation Table (PAT) must be decoded. The PAT is the master PSI table with PID value always equal to zero (PID = 0). If the PAT cannot be found and decoded in the TS, then no programs can be found, decompressed, or viewed.
For a set-top box or ATSC tuner to successfully perform the program recovery and decompression process, the PSI tables must be sent periodically and with a fast enough repetition rate that it doesn’t delay channel-surfing viewers. Thus, checking the PSI tables for correct syntax and repetition rate is a vital part of MPEG testing.
Testing PSI involves verifying the accuracy and consistency of PSI contents. As programs change or multiplexer provisioning is modified, some problems may occur. One problem would be unreferenced PID. Packets with a PID value are present in the TS but are not referenced in any table.
If there are no packets with the PID value referred to in a PSI table present in the TS, the problem could be a missing PID.
Another useful PSI test is a check of program content. Just because there are no unreferenced or missing PIDs indicated does not mean that the viewer is receiving the correct program. There may also be a mismatch of the audio content from one program being delivered with the video content from another program. Because MPEG allows multiple audio channels for multiple languages, an air-check can ensure that viewers are receiving the correct language.
It is possible to use a set-top box and television to do the air check, but a better way would be to use an MPEG test set that incorporates all the PSI table checks plus a built-in decompressor with picture and audio display. This would allow you to correlate PSI contents and actual program content as well as allow a quick visual and aural check of ES.
By Ned Soseman, Broadcast Engineering
Production & Exchange Formats for 3DTV Programmes
The purpose of this EBU recommendation is to give technical aid to broadcasters who intend to use current (or future) 2D HDTV infrastructures to produce 3DTV programmes.
Super Hi-Vision at 120fps for Olympics
NHK is developing an 8K image sensor for its Super Hi-Vision Ultra HD system capable of 120 frames per second, which it plans to unveil in Tokyo on 23 April and should be ready for use at the London Olympics.
The new 33-megapixel (7680x4320 pixel) CMOS sensor will use an advanced two-stage (4-bit then 8-bit) cyclic analogue-to-digital converter to deliver a 12-bit image at the higher frame rate (twice that of the current SHV cameras). This architecture also allows it to reduce power consumption, with the ADC drawing 800mW (out of a total drive power of 2.5W). To deal with the high number of pixels involved, the sensor outputs alternating rows of pixels to ADCs on either side, each of which has 48 parallel outputs.
The sensors will probably still get rather warm at these frame rates, so heat management will be a significant issue. Each pixel measures 2.8 x 2.8 microns, and the 26.5 x 21.2mm chip (which is about the width of a Super35mm sensor, but taller) will use a 0.18-micron manufacturing process. The chip is being developed with Shizuoka University, and engineers revealed details of the sensor at the recent IEEE International Solid-State Circuit Conference in San Francisco.
Two SHV cameras will be used to capture parts of the 2012 Olympics, with transmission to three large screens around the UK (plus one in the International Broadcast Centre) and three in Japan. Having the 120fps sensor will reduce motion blur and allow for much better slo-mo replay.
However, as the resolution increases (and 8K is 16 times the resolution of HD), motion defects become much more noticeable. “300 frames per second might just be acceptable, but 600fps would be better,” said colour scientist and camera consultant, Alan Roberts. At 300fps, the material would also be easily compatible with both 50 and 60Hz display.
Using Long GoP compression, which typically combines a group of pictures in half a second, would lead to a GoP of 150 or 300 frames, but this wouldn’t lead to huge bandwidth requirements. “The motion between frames is very small, so the compression is much easier,” he explained, which means the increase in bit rate needed to convey high framerate material is minimal. “The problem is in the shooting and the editing, because of the monstrous data files you have to deal with,” he added.
By David Fox, TVB Europe
The 3D Production Guide
3net, the 24/7 3D network and 3D television production studio, along with joint venture partners Discovery, Sony and IMAX, announced the release of the most complete guide to 3D television production ever assembled.
Featuring stereoscopic expertise from the top producers and technical advisors of the company and its corporate ownership, The 3D Production Guide has now been made freely available to the public via multiple websites.
The 50-page illustrated manual includes detailed information garnered from the combined 50 years of experience in the area of 3D from those who contributed to its creation. The guide outlines in detail all of the facets involved in creating top-quality 3D content for television, from initial workflow planning, to production, post production, stereographic correction and final delivery.
The guide was authored by Bert Collins, Josh Derby, Bruce Dobrin, Don Eklund, Buzz Hays, Jim Houston, George Joblove and Spencer Stephens, with Bert Collins and Josh Derby serving as editors. It will be constantly updated and amended as the dynamics of 3D television production continue to evolve.
Source: 3net
EyeIO: Netflix’s Secret Weapon Against Bandwidth Caps?
Palo Alto, Calif.–based video encoding startup EyeIO left stealth mode on Wednesday with the announcement that it has licensed its technology to one of the biggest players in the online video space. Netflix is using eyeIO’s encoding technology to cut down on the bandwidth of its streams, allowing the company to deliver HD video without busting subscribers’ bandwidth caps or overwhelming networks in emerging markets.
EyeIO has been operating stealthily since the end of 2010, and it was able to win Netflix as a customer last summer. Netflix hasn’t said where and in which capacity it is exactly using the technology it has been licensing from eyeIO, but the company’s VP of Product Development, Greg Peters, said in a press release that eyeIO is “an important part of the technology [Netflix uses] to improve video quality and overcome bandwidth challenges presented by Internet infrastructure.”
Standard-definition Netflix streams can consume up to 2.2 Mbps of bandwidth. Netflix’s 720p HD videos come in at roughly 3.8 Mbps, and 1080p videos go up to 4.8 Mbps. EyeIO CEO Rodolfo Vargas told me during a phone conversation on Tuesday that his company’s encoding technology can achieve better-looking results than most established encoders with 20 percent bandwidth savings and that eyeIO can still deliver similar quality to other encoders with up to 50 percent bandwidth savings. Content in 720p could be streamed using 1.8 Mbps, he explained. The company does this by optimizing the encoding process, which means that the results are regular, albeit smaller, H.264 files that can be played by end users without any need for additional plug-ins.
EyeIO was founded by online video technology veterans; Vargas used to be the senior program manager for video at Microsoft, and one of his co-founders, Robert Hagerty, used to be the chairman and CEO of teleconferencing provider Polycom. The company is privately funded and currently has fewer than 10 full-time employees but is looking to expand over the coming months.
By Janko Roettgers, GigaOM
Large-Sensor Camcorders and DSLRs
When Canon announced its EOS C300 camcorder, which has a 4K2K sensor but does not record 4K2K video, the company also announced it was developing a DSLR that will record 4K2K video. Although it may seem a bit strange that a camcorder designed to employ high-quality cinema lenses is limited to full HD video recording yet a still camera will be able to record 4K2K video, it's not strange given the history of DSLRs.
When large CCD and CMOS chips replaced an SLR's 35mm film, the next logical step was to place an LCD on the digital camera so one could review shots in the field. The next logical step was a “live view” mode that allowed one to view what was being recorded. It was only a small step to compress live view images and record them as video.
Large-sensor digital camcorders have evolved from DSLRs. It is primarily a marketing decision whether to release digital motion picture technology in a still camera package, a camcorder package or both. However, one clear advantage of a camcorder package is space for a mic jack (even XLRs), a headphone jack and manual audio controls.
When a potential buyer who is in the process of learning about 4K2K production and post production encounters the same technology in two different packages, it may prove confusing.
When a videographer shoots with a still camera, he or she will find expected camcorder functions missing. For example, every professional camcorder has some form of ND filtration; DSLRs do not.
A primary differentiator of DSLRs and traditional camcorders is their optical system. This is true for current HD and future 4K2K products.
Frame Size
While video cameras have frame sizes that relate directly to sensor size, such as 2/3in, DSLR frame size relates to 35mm film — in particular, 35mm still film. When shooting 35mm slide or negative film, each 36mm × 24mm image is placed with perforations above and below the frame.
DSLRs with 36mm × 24mm sensors are called full-frame cameras. The Canon EOS-1D X, announced for March 2012, employs an 18-megapixel 28.7mm × 19.1mm sensor. Canon calls it an APS-H sensor.
There are small variations in APS frame size: Canon APS-C (22.2mm × 14.8mm), and Nikon/Sony-C (23.4mm × 15.6mm). Both full-frame and APS sensors, when taking photos, have a 3:2 (1.50:1) aspect ratio. Panasonic uses a slightly smaller sensor for its AF100 camcorder and GH2 still camera called Micro Four Thirds (M43), which has a 1.33:1 aspect ratio and a frame size of 17.3mm × 13mm.
Sensors smaller than a full-frame sensor reduce the potential minimum DOF. Minimum DOF, of course, is a function of the maximum aperture size. A large-sensor camera does not directly provide a shallow DOF.
When a lens designed for a full-frame camera is mounted on a camera with a smaller sensor, the lens' focal length is multiplied by the lens crop factor. (Crop factor equals the ratio of a 35mm frame's 43.3mm diagonal to the diagonal of the image sensor.) A Sony APS-C camera, for example, has a crop factor of 1.5. A 50mm “normal” lens becomes a 75mm tele lens.
When shooting video, a 16:9 window on the sensor is employed. This has three ramifications. First, the viewfinder image will shrink when switching a DSLR to video mode. (This shift can be minimized by shooting 16:9 photos.) Second, the number of pixels read out will be reduced, which is a positive. Third, the lens crop factor will slightly increase. For example, when a Sony APS-C camera is switched to video mode, the crop factor increases to 1.8, thus a 50mm lens acts as a 90mm lens.
The earliest 35mm movie film had a 22mm × 18mm image, with perforations on the sides of each frame. In 1929, the Academy ratio was established. It has a 21mm × 15mm image that has a 1.37:1 aspect ratio. To obtain wide-screen, but not anamorphic, images, a Super 35 frame can be employed.
A 24.9mm × 13.9mm Super 35 frame has a native aspect ratio of 1.79:1 — a perfect match to 1.78 (16:9) HD. It also matches Quad-HD (3840 × 2160 pixels) and almost matches 4K2K, which is 4096 × 2160 pixels — a 1.90:1 aspect ratio. Not surprisingly, frame sizes that come from cinema cameras do not require the use of a 16:9 window when shooting video.
Lens Zoom System
While the videographer likely knows that DSLR lenses do not have power zoom, he or she may not know that photo lenses have other issues. For example, the zoom ring may have high friction because of the need to significantly extend the lens when zooming. Pressure exerted to start a zoom while shooting can easily cause a visible disturbance.
Better Sony lenses, such as A-mount lenses that use micro ball bearings, may cause noise that will be picked up by an on-camera mic.
AF System
Photographers are used to trusting auto-focus — even on action shots where the shooter is following a moving subject. When a DSLR's mirror is in the 45-degree-position in order for the shooter to see the subject, AF is possible. A portion of the image passes through a semitransparent area of the mirror, reflects off a small mirror mounted on the back of the mirror and is cast onto a small sensor at the bottom of the camera. The sensor, in conjunction with a processor, sends commands to the lens' AF motor to move to a position calculated to be correct for precise focus. This system is called phase detection AF.
and a processor to calculate precise focus.
DSLRs employ a different AF system when shooting video because the mirror must be continuously up. The processor, therefore, obtains information from the CMOS image sensor, which is why it is called contrast detection AF.
Mirrorless cameras such as the Panasonic GH2 and Sony NEX-5N must use contrast detection AF. (Strictly speaking, digital cameras without a mirror do not have a reflex system and, therefore, are not DSLRs.)
Contrast detection AF systems work by having a microprocessor rapidly command the lens servomotor to step forward and backward by a tiny amount. The processor notes whether contrast increases or decreases. If contrast increases, then current focus is not perfect. Therefore, stepping forward and backward continues. When there is no change in contrast, the current focus is the best possible.
Contrast detection tends to be slower than phase detection and becomes slower at low light levels. And, unless the lens is designed to be quiet, AF noise may be recorded.
Aperture System
Photography lenses are designed to click into key f-stops: f/2.8, f/4, f/5.6, f/8, f/11, f/16 and f/22. Cinema and video lenses are designed so the aperture changes in a continuous manner. One solution is to use camera lenses designed by the camera's manufacturer for video shooting. The other solution is to use cinema lenses.
ND Capability
To obtain a shallow DOF with a large-chip camera under bright light — at the slow shutter speed required for the correct amount of video motion blur — an ND is a must. (ND filtration also will be required to keep the aperture under f/11 to minimize diffraction.) When a camcorder does not have a built-in filter, a shooter has three choices: mount the camera on rails on which a matte box is mounted, attach one of several ND filters to the lens or employ a vario-ND filter.
Lens Mount Type
Both cameras and camcorders that employ large sensors use a lens mount designed to work with their brand of lenses. For example, Sony's NEX family — including the FS100 and VG20 camcorders — uses Sony's E-mount. Sony markets the LA-EA2 adaptor, which enables the use of Sony and Minolta A-mount lenses. The LA-EA2 has a translucent mirror system that provides phase detection AF to many A-mount lenses.
For most interchangeable lens cameras, third-party adaptors are available. These enable you to use your favorite photo lenses on a new camera or camcorder. For example, a Sony NEX camera can use Nikon F, Canon 5D, Leica M, Leica R, Pentax, Konica Minolta MD, Olympus and Contax/Yashica lenses by using an E-mount adaptor.
Only a few adaptors, such as the LA-EA2, provide electrical signals to a lens. Without electrical connections, in-lens optical stabilization, AF and aperture control cannot function and no information from the lens is received by the camera. Therefore, modern photo lenses that send the aperture ring's setting to the AE system cannot do so.
Solutions to these issues include working with still camera and cinema lenses in a fully manual way (which may be a camera operator's first choice) or using a manufacturer's lenses that have electrical contacts.
Bringing it Home
No matter whether you shoot with a still camera or camcorder, images from the sensor must be compressed and recorded. Currently, two codecs are used for recording 4K2K: the Sony F65RAW (16-bit RAW) codec to a docking SRMaster field recorder that records to SRMemory cards or the RED R3D wavelet codec to a REDMAG solid-state drive.
Future 4K2K codec options include H.264 (as a single stream or as four HD streams) and 4K2K versions of current HD formats.
By Steve Mullen, Broadcast Engineering
DPP Unveils Technical & Metadata Standards for File-based Programme Delivery
The Digital Production Partnership (DPP) – a partnership between ITV, Channel 4 and the BBC – has unveiled its new Technical and Metadata Standards for File-based programme delivery in the UK.
Through the DPP, seven major broadcasters (BBC, ITV, C4, Sky, Channel Five, S4C and UKTV), have all agreed the UK’s first common file format, structure and wrapper to enable TV programme delivery by digital file. These new guidelines will complement the common standards already published by the DPP for tape delivery of HD and SD TV programmes.
Working closely with the Advanced Media Workflow Association (AWMA) in the US, the DPP has been the driving force behind the creation of the organisation’s ‘AS-11,’ a new international file format for HD Files. The new DPP guidelines will require files delivered to UK broadcasters to be compliant with a specified subset of this new, internationally recognised standard.
By implementing one set of pan-industry technical standards for the UK, the DPP aims to minimise confusion and expense for programme-makers, and avoid a situation where a number of different file types and specifications proliferate.
The new DPP standards aim to remove any ambiguity during the production and delivery process. A key aspect is the inclusion of editorial and technical metadata, which will ensure a consistent set of information for the processing, review, and scheduling of programmes, as well as their onward archiving, sale and distribution.
As part of the file-based guidelines, the DPP’s member broadcasters have agreed a minimum set of common metadata to be delivered with a file-based programme. And, in a bid to encourage international adoption of its metadata standards, the DPP has worked closely with the European Broadcasting Union (EBU), mapping its minimum set of common metadata to existing ‘EBU-Core’ and ‘TV-Anytime’ metadata sets.
Alongside these new standards, the DPP is currently building a free-to-use, downloadable, metadata application to enable production companies to enter the required editorial and technical metadata easily. The new application is due to launch in spring 2012.
The agreement of these new file based technical standards does not signal an immediate move to file based delivery. Instead, the DPP seeks to provide clarity around digital delivery that will become the expected standard in the future.
During 2012 BBC, ITV and Channel 4 will begin to take delivery of programmes on file on a selective basis. Production companies wishing to deliver by file should discuss this at the point of commission, and seek formal agreement with their broadcaster at the outset of production. After a period of selective piloting, file based delivery will be the preferred delivery format for these Broadcasters by 2014.
Source: Digital Production Partnership
AS-11: MXF for Contribution
AS-11 is a vendor-neutral subset of the MXF file format to use for delivery of finished programming from program producers and distributors to broadcast stations. AS-11 files are intended to be complete and ready for playout.
AS-11 supports playout while the file transfer is in progress, a workflow is referred to as “late delivery”. It is preferable for AS-11 files to be used by playout servers directly without rewrapping of the MXF data structures.
The content may be delivered at the ultimate bit-rate, picture format and aspect ratio, or it may be transcoded at the broadcast station to the required bit-rates and formats. Similar transcoding may be applied to audio and captions; additionally, specific audio and caption tracks may be selected for different broadcast channels.
The content may be pre-packaged for broadcast without further splicing or it may be segmented for ease of insertion or replacement of interstitials.
AS-11 supports SD video encoded as D-10, 50Mbit/s, and HD as AVC-Intra Class 100. Audio can be PCM, AC-3 or Dolby E.
AS-11 defines a minimal core metadata set required in all AS-11 files, a program segmentation metadata scheme, and permits inclusion of custom shim-specific metadata in the MXF file.
Source: Advanced Media Workflow Association
EBU-TT Subtitling Format Published for Industry Comments
The EBU has published a new Subtitling Format specification (EBU Tech 3350). The new format is called EBU Timed Text (EBU-TT) and provides an easy-to-use method to interchange and archive subtitles in XML.
EBU-TT is based on the W3C Timed Text Markup Language (TTML) specification. The EBU format can be seen as a constrained version of the W3C spec, aimed at providing a solution more tailored to broadcast operation. This is especially relevant as broadcasters are increasingly moving to file-based HDTV facilities, where subtitles are created, edited, exchanged and archived together with the content.
The previous EBU subtitling format was EBU STL (Tech 3264), developed at a time when information was still exchanged on floppy discs. However, as many broadcasters still use STL or have archived STL files, great care was taken in the development of EBU-TT to make sure that it provides backwards compatibility with its predecessor.
The EBU is also providing an XML Schema for EBU-TT.
Source: EBU
Tech Upstarts Kicking Glasses in 3D
There's no shortage of innovation from the major TV manufacturers on display at the huge booths at CES: OLED, 4K -- even 8K -- resolution, new interfaces, connectivity, exclusive content. What's absent here, though, are any prototypes to indicate that glasses-free (autostereo) 3D TV is anywhere close to market.
That's not to say that autostereo TV can't be found at CES. It's just coming from smaller companies in smaller booths -- one with barely a booth at all. They are pressing ahead with -- and showing off -- autostereo screens for television, tablets and smartphones while the big makers remain oddly quiet on the topic.
"Consumer electronics companies wanted to get into the home market quickly," said Raja Rajan, chief operating officer of Stream TV, whose booth in Central Hall, of the mammoth Las Vegas Convention Center, is not far from Sony's. "The consumer electronics companies have tremendous financial pressures to get to market with the fastest, easiest technologies."
That is echoed by one of Rajan's competitors, Stephen Blumenthal of 3D Fusion, a late addition to the floor that has one of its models tucked into the 3D Bee booth at the periphery of Central Hall.
"They brought (3D with glasses) to the market as a very straightforward consumer play, and until they burn through the opportunity to make as much revenue off of it as possible, this adventure with the next step is on the back burner," Blumenthal said.
His partner Ilya Sorokin noted, "The 3D with glasses technology was much easier to incorporate into their existing infrastructure because it was already there, and just lying on a shelf."
Both 3D Fusion and Stream TV are using advanced, lens-based tech that, according to Rajan, was abandoned by the big companies.
Rajan said he toured Asia showing Stream TV's screens and its real-time 2D-to-3D converter to major hardware makers, who responded enthusiastically. Stream TV is looking to be a technology provider, not to manufacture under its own name.
"We expect in the next few weeks to start announcing some of the first brands and products rolling out," Rajan said.
He said there is strong interest from Hollywood in the converter box, because it can be built into cable and satellite boxes, enabling all channels to be in 3D. At the same time, Stream's units come with controllers so the consumer can turn the 3D down, or off altogether, for comfort or personal preference.
"Our cost is incrementally 10% to 15% max over the cost of goods for a 2D television," Rajan said. "That's significant because a big re-seller can get into the consumer market at a cost consumers can afford."
MasterImage 3D, which has a solid worldwide business projecting 3D in theaters, is in the South Hall. It has been in the autostereo screen business for some time, and this year is at CES with two screens aimed straight at state-of-the-art mobile devices: a 720p 4.3-inch smartphone display and a WUXGA (1920x1200) display for tablets.
Royston Taylor, exec VP and general manager for MasterImage, said he welcomes the competition from Stream TV, which is also showing tablet screens.
"First, it validates what you're trying to do," Taylor said. "Being on your own is nice in terms of no competition, but it's very lonely in terms of being the only voice saying how great something is. The second thing is competition is always good for the consumer."
Despite strong sales of the Nintendo 3DS, the poor critical response to the 3DS, the HTC Evo 3D phone and the LG Optimus 3D phone have made some makers nervous, Taylor said. He now expects to be making announcements of deals with consumer electronics companies by April and to have gear with MasterImage 3D screens in stores by Thanksgiving.
One hurdle that had to be overcome was the lack of technical standards for judging the quality of a 3D display.
"Right now it's almost entirely subjective," he said. "Big companies won't risk a $250 million phone line on 3D just because it looks nice."
But a French company, Eldim, has come up with a product for testing 3D displays on objective, technical measurements. With standards in place, it will be possible to compare products and establish quality control in manufacturing.
3D Fusion is already selling autostereo TVs for use in digital signage. Blumenthal said the company is selling its turnkey solution, which includes a 42-inch autostereo display, at CES. Cost is $8,000. His sales are to retailers, small mom-and-pop chains, malls. Blumenthal and Sorkin recognize that their company is small and they're in no position to ramp up to consumer volumes on their own. Like Stream TV, they'd be happy to license their technology.
By David S. Cohen, Variety
Friday, January 13, 2012
Labels: 3D, Analysis and Measurement, Autostereoscopy, Displays and Projection
3-D Cameras for Cellphones
Researchers at Massachusetts Institute of Technology (MIT) have developed a system that uses specially designed algorithms to produce a detailed 3D image with just a cheap photodetector and the processor power found in a smartphone.
Like other sophisticated depth-sensing devices, CoDAC uses the “time of flight” of light particles to gauge depth: A pulse of infrared laser light is fired at a scene, and the camera measures the time it takes the light to return from objects at different distances.
Traditional time-of-flight systems use one of two approaches to build up a “depth map” of a scene. LIDAR (for LIght Detection And Ranging) uses a scanning laser beam that fires a series of pulses, each corresponding to a point in a grid, and separately measures their time of return. But that makes data acquisition slower, and it requires a mechanical system to continually redirect the laser.
The alternative, employed by so-called time-of-flight cameras, is to illuminate the whole scene with laser pulses and use a bank of sensors to register the returned light. But sensors able to distinguish small groups of light particles — photons — are expensive: A typical time-of-flight camera costs thousands of dollars.
The MIT researchers’ system, by contrast, uses only a single light detector — a one-pixel camera. But by using some clever mathematical tricks, it can get away with firing the laser a limited number of times.
The first trick is a common one in the field of compressed sensing: The light emitted by the laser passes through a series of randomly generated patterns of light and dark squares, like irregular checkerboards. Remarkably, this provides enough information that algorithms can reconstruct a two-dimensional visual image from the light intensities measured by a single pixel.
In experiments, the researchers found that the number of laser flashes — and, roughly, the number of checkerboard patterns — that they needed to build an adequate depth map was about 5 percent of the number of pixels in the final image. A LIDAR system, by contrast, would need to send out a separate laser pulse for every pixel.
To add the crucial third dimension to the depth map, the researchers use another technique, called parametric signal processing. Essentially, they assume that all of the surfaces in the scene, however they’re oriented toward the camera, are flat planes. Although that’s not strictly true, the mathematics of light bouncing off flat planes is much simpler than that of light bouncing off curved surfaces. The researchers’ parametric algorithm fits the information about returning light to the flat-plane model that best fits it, creating a very accurate depth map from a minimum of visual information.
Indeed, the algorithm lets the researchers get away with relatively crude hardware. Their system measures the time of flight of photons using a cheap photodetector and an ordinary analog-to-digital converter — an off-the-shelf component already found in all cellphones. The sensor takes about 0.7 nanoseconds to register a change to its input.
That’s enough time for light to travel 21 centimeters, Vivek Goyal from MIT’s Research Lab of Electronics, says. “So for an interval of depth of 10 and a half centimeters — I’m dividing by two because light has to go back and forth — all the information is getting blurred together”.
Because of the parametric algorithm, however, the researchers’ system can distinguish objects that are only two millimeters apart in depth. “It doesn’t look like you could possibly get so much information out of this signal when it’s blurred together,” Goyal says.
The researchers’ algorithm is also simple enough to run on the type of processor ordinarily found in a smartphone. To interpret the data provided by the Kinect, by contrast, the Xbox requires the extra processing power of a graphics-processing unit, or GPU, a powerful special-purpose piece of hardware.
“This is a brand-new way of acquiring depth information,” says Yue M. Lu, an assistant professor of electrical engineering at Harvard University. “It’s a very clever way of getting this information.” One obstacle to deployment of the system in a handheld device, Lu speculates, could be the difficulty of emitting light pulses of adequate intensity without draining the battery.
But the light intensity required to get accurate depth readings is proportional to the distance of the objects in the scene, Goyal explains, and the applications most likely to be useful on a portable device — such as gestural interfaces — deal with nearby objects. Moreover, he explains, the researchers’ system makes an initial estimate of objects’ distance and adjusts the intensity of subsequent light pulses accordingly.
Telecoms company Qualcomm has awarded the research team one of $100,000 Innovation Fellowship grants to continue the research.
By Larry Hardesty, Massachusetts Institute of Technology
BetterView
BetterView's up-conversion technology utilizes Super-Resolution (SR) reconstruction that performs a fusion of low quality images into a higher quality result with improved optical resolution. This task encompasses scaling-up of the visual content by introducing true (optical) resolution enhancement.
It has been known for the past 20 years that, in principle, one could take several low-quality images and fuse them into a single, higher-resolution outcome. This has been demonstrated by scientists, adopting various techniques and algorithms. This process became a hot field in image processing, with thousands of academic papers published during the past two decades on the problem and ways to handle it. The classical approach to fuse the low-quality images requires finding an exact correspondence between their pixels, a process known as "motion estimation".
BetterView technology is based on a recently developed and patent-pending novel family of SR algorithms, proposed by a world-leader in this field, Prof. Michael Elad (Technion – Israel Institute of Technology). Elad and his collaborator, Dr. Matan Protter devised the first method that overcomes the requirement for very accurate and explicit motion estimation in previous SR technologies.
Exact motion estimation has been a crucial stage in every earlier SR algorithm, considerably limiting the scenes that can be handled. The new family of SR techniques avoids the exact motion estimation and replaces it by a probabilistic estimate. This enables handling successfully general content scenes containing extremely complex motion patterns. The results are impressive, with no visual artifacts, and the process is completely robust.
iPlayer3D Update
The Digital Service Development group led by Phil Layton in BBC R&D was involved in the previous trial of 3D at the Wimbledon Tennis Championships this year and also the recently broadcast Strictly Come Dancing Grand Final. In this post Dr Peter Cherriman and Paul Gorley outline the work they did to determine if it was possible to put 3D content onto the Freeview and Freesat versions of TV iPlayer.
Generally 3D requires a high bitrate to achieve good stereoscopy. If the video bitrate is too low, depth cues are lost and the 3D becomes tiring to watch. However, due to varying Internet speeds, the higher the bitrate on iPlayer, the less people that are able to watch it. So we had the challenging task of trying to producing high quality 3D at as low a bitrate as possible.
Our 3D television broadcasts on Freeview, Freesat, Sky and Virgin all use a side-by-side frame-compatible format. This combines the Left and Right eye views into a single HD signal, by anamorphically squashing horizontally each eye's view into half of the HD frame, so they appear side-by-side. This HD signal was compressed at resolution of 1920x1080i25, which means 25 interlaced frames per second each of which is comprised of 1920 pixels across and 1080 lines. Each interlaced frame comprised of two fields, each field is 1920x540 pixels, and the fields are captured 1/50th of a second apart.
In order to produce the best quality 3D we decided to use a recording made in Blackpool, rather than use the broadcast feed received via satellite. This had a number of advantages, it meant we weren't limited to the side-by-side 3D format and the recording would have less compression artefacts.
We did a number of experiments with different resolutions and determined the best compromise for bitrate and quality was to convert the recorded 1920x1080i25 interlaced signal for each eye into a non-interlaced 1280x720p50 signal using a professional cross-converter. The intermediate signal created is at 50 frame per second, where each frame is 1280x720 pixels for each eye.
We then needed to convert the pair of 1280x720p50 signals into a standardised frame-compatible format. The preferred frame-compatible format for 1280x720p50 signals is to anamorphically squash vertically each eye signal into half the HD frame, the so-called top-bottom or over-under format. This results in 1280 by 360 pixels per eye per frame.
Our broadcasts use side-by-side format, however this requires horizontal squashing of the picture which degrades the stereoscopic depth cues. These would be further degraded for a 1280x720 image format. Vertical squashing used in the top-bottom format preserves more of this depth information, but the rescaling required in a receiver is much harder for the interlaced broadcast format which is why it's not used.
We found that our broadcast video encoders were much more efficient at encoding this 1280x720p50 signal than the existing software encoders. The bespoke workflow of this experiment allowed us to trial the use of broadcast encoders for iPlayer content. We tested with a wide range of 3D material, but the most challenging of which was the Strictly Come Dancing 3D footage, shown in cinemas, for last year's Children in Need. By using HE-AAC audio encoding we were able to minimise the audio bitrate required. This enabled us to create good quality 3D at a constant total bitrate of less than 5Mbit/s. You should notice the improved quality of the iPlayer 3D pictures in terms of less compression artefacts, and smoother motion due to the 50 frames per second, which is twice the framerate of standard iPlayer.
The next challenge was to modify the file to be suitable to upload to the iPlayer platform. The Freesat receivers required a MPEG Transport Stream (ISO/IEC 13818-1), which is produced directly by the broadcast video encoders.
However, FreeviewHD receivers require a mp4 file. When we used our standard software tool to create the mp4 files we found that the audio and video was not in-sync on some receivers. The mp4 files contain metadata to indicate how to synchronise the video and audio. However, it seems some receivers assume the first frame of video should be synchronised with the first frame of audio and don't make use of this metadata. Using a alternative tool, we were able to create mp4 files which played in-sync on all receivers available to us, including those which previously seemed to ignore this synchronisation metadata.
We don't yet know what the future of 3D will be, but these experiments have demonstrated another platform on which 3D content can be delivered to viewers.
By Ant Miller, BBC R&D
JPEG 2000
JPEG 2000 has caught the attention of the professional media world for good reason. First, it closely matches some workflows, where the production process operates on each frame of a video stream as a discrete unit. This is different from MPEG-2 and MPEG-4 AVC Long-GOP flavors, where, during the reconstruction process, algorithms reference frames before and after the frame being reconstructed.
The ability to compress each frame as a free-standing unit has made it popular in the digital intermediate space in Hollywood. JPEG 2000 is also of interest to those who want lossless compression. It can provide a bit-perfect reconstruction of the original compressed image, although at a cost in terms of bandwidth. Also, the wavelet compression used in JPEG 2000 provides some unique opportunities that are not available in other compression methods.
The wavelet transform separates the image into four sub-bands. The first sub-band is a lowpass horizontal and lowpass vertical (LL). Images that have gone through this separation are basically lower-resolution images of the original. The other sub-bands are as follows: lowpass horizontal and highpass vertical (LH); highpass horizontal and lowpass vertical (HL); and highpass horizontal and highpass vertical (HH).
Using wavelet transforms and some clever thinking, implementers can do some interesting things. For example, they can send only the LL image, if they know that they are feeding a low-resolution display. Or they can send the LL sub-band in a highly-protected stream, in order to ensure the original image arrives intact. That said, they can then send the higher-resolution sub-bands unprotected since a momentary loss of these sub-bands is not likely to be noticed.
Given JPEG 2000's popularity, it is not surprising there have been some developments that make it particularly interesting for professional applications. First, the ITU has created an amendment (1) that outlines specific configurations for broadcast contribution applications. These configurations are intended to establish interoperability points for those implementing JPEG 2000 in professional applications. This is important because, until the amendment was released, there were so many variables in the compression tool set that interoperability was unlikely. The second important development, Amendment 5 to the MPEG 2 standard (2), provides a mapping of the JPEG 2000 Program Elementary Stream (PES) onto the MPEG-2 Transport Stream (TS).
Finally, some time ago, the Pro-MPEG Forum started to develop a standardized way to transport MPEG-2 TS over IP networks. The Video Services Forum picked up on this work and continued to develop it, finally submitting a draft for standardization within the SMPTE. This standard, SMPTE 2022-2 (3), describes a method for mapping MPEG-2 Transport Streams onto IP networks using RTP and UDP. The document was approved in 2007 and is the most common standard deployed today for professional video transport applications.
So these three developments — development of broadcast profiles; a mapping of JPEG 2000 Program Elementary Streams to MPEG 2 Transport Streams; and wide availability of MPEG-2 TS over IP transport equipment — mean now it is possible to transport JPEG 2000 over IP networks.
broadcast contribution profiles, JPEG 2000 PES to MPEG-2 mapping,
and SMPTE 2022-2 for MPEG-2 transport over IP networks.
Starting with a video source, the image is compressed using a compression engine. This engine is configured to one of the Broadcast Contribution profiles in ITU-T Amendment 3. The compression engine produces a JPEG 2000 PES. This stream is then fed to an MPEG-2 encapsulator. The encapsulator uses the mapping rules established in the MPEG-2 specification, Amendment 5, to map the PES onto an MPEG-2 TS. This MPEG-2 TS is fully compliant with MPEG-2 specifications because, from the outside, it looks just like a normal MPEG-2 Transport Stream. As such, the output of the MPEG-2 encapsulator can be fed into a SMPTE 2022-2 compliant video transport device. This device encapsulates the MPEG-2 TS in standard RTP and UDP packets, and then those packets are wrapped in IP packets. These IP packets can now be fed into an IP network.
You might wonder why we take a relatively new compression algorithm such as JPEG 2000 and encapsulate it in MPEG-2. There are several reasons for this. First and foremost, there are already a number of specifications for how to encapsulate a number of different audio formats into MPEG-2 Transport Streams. Remember: JPEG 2000 says nothing about audio. Using MPEG-2 TS allows us to transport and present the audio alongside the JPEG 2000 compressed video using well-known audio standards. Also, this approach allows us to leverage the existing SMPTE 2022-2 MPEG-2 TS over IP standard. Finally, there are no technical issues in MPEG-2 TS that need to be fixed in this application space, so re-use of Transport Streams rather than inventing something entirely new seems like a good solution.
So, the good news is that the time is ripe for development of an interoperable, open solution for the transport of JPEG 2000 video and audio over IP networks. The standards exist, and there is a clear path forward. But, there are a few issues that need addressed.
JPEG 2000 has been around for quite some time. As such, some proprietary JPEG 2000 over IP transport solutions have already been created. Of course, these were developed in response to customer demand, so existing implementations may need to be changed. Another issue is that while the broadcast contribution profiles in Amendment 3 go a long way toward interoperability in the JPEG 2000 PES space, recent analysis suggests, without further definition, implementations based upon these profiles will not be interoperable. Finally, until the industry actually tries to connect devices from different manufacturers together, interoperability cannot be assured.
Fortunately, the industry is becoming aware of these issues, and steps are being taken to begin work in earnest on interoperable, open transport of professional JPEG 2000 images over IP networks. I would expect to see some developments around this in the first half of the coming year.
Footnotes:
- “Profiles for Broadcast Applications” ISO/IEC 15444-1:2004 Amd.3-2010 (ISO/IEC, Geneva, Switzerland: 2010) |Rec. ITU-T T.800 Amd.3 (06/2010) (ITU, Geneva, Switzerland:2010)
- Amendment 5: Transport of JPEG 2000 Part 1 (ITU-T Rec T.800 | ISO/IEC 15444-1) video over ITU-T Rec H.222.0| ISO/IEC 13818-1
- SMPTE ST20 22-2:2007 “Unidirectional Transport of Constant Bit Rate MPEG-2 Transport Streams on IP Networks”
By Brad Gilmer, Broadcast Engineering
Streaming Standards for Worship
If you produce streaming video in the worship market and have your ear to the ground, you may be experiencing sensory overload right now. HTML5 is being promoted as a panacea for all plug-in-related woes; Adobe threw the mobile market into turmoil by ceasing development of the Flash Player, and there’s a new standard called DASH that supposedly will create a unified approach for adaptive streaming to all connected devices. Seems like getting that sermon out over the Internet has gotten a lot more complicated.
Well, maybe not. In this article I’ll describe what’s actually happening with HTML5, Flash, and DASH, and make some suggestions as to how to incorporate these changes into your video-related technology plans.
About HTML5
Let’s start with HTML5, which has one potential show stopper for many houses of worship: the lack of a live capability. Apple has a proprietary technology called HTTP Live Streaming that you can use to deliver to iDevices and Macs but not Windows computers. So if live video is a requirement, HTML5 is out—at least for the time being.
If on-demand video is your sole requirement, HTML5 is a tale of two marketplaces: desktops and mobile. By way of background, HTML5-compatible browsers don’t require plug-ins like Flash or Silverlight to play web video. Instead, they rely on players that are actually incorporated into and shipped with the browser. Integrating video into a webpage for HTML5 playback uses a simple tag rather than a complicated text string to call a plug-in.
Today, the installed base of HTML5-compatible browsers on desktop computer is only around 60 percent, which makes it an incomplete solution, particularly for houses of worship whose older parishioners may be technology laggards who don’t quickly upgrade to new browsers. However, in the link that you use to display your video, it’s simple to query the browser used by the viewer to test for HTML5-playback capabilities. If the viewer’s browser is HTML5-compatible, the video will play in the HTML5 player. If not, you can code the page to “fall back” to the existing Flash Player or other plug-in, which will then load and play normally. While this sounds complicated, Flash fallback is totally transparent to the viewer and occurs in just a millisecond or two.
Why HTML5 first? Because as we’ll see in a moment, this is a very solid strategy for supporting Apple and Android devices. However, before jumping in, keep in mind that HTML5 is not as mature as Flash in several important respects. First, it lacks true streaming, or the ability to meter out video as it’s played, which is more efficient than progressive download. HTML5 also can’t adaptively stream or dynamically distribute multiple streams to your target viewers to best suit their connection speed and CPU power. It’s these two issues that the aforementioned DASH standard hopes to address.
However, the DASH standard doesn’t address HTML5’s biggest implementation hurdle, which is that all HTML5 browsers don’t support a single compression technology or codec. Specifically, Microsoft’s Internet Explorer 9 and Apple Safari include an HTML5 player for the H.264 codec, while Mozilla Firefox and the Opera browser support only Google’s open-source codec, WebM. Today, Google Chrome browser includes both codecs, but Google has stated that they intend to remove the H.264 codec sometime in the future. It’s actually a bit worse than this sounds because Firefox version 3.6, which is still more than 5 percent of the installed base of desktop browsers, only supports a third codec, Ogg Theora.
To fully support the universe of HTML5-compatible browsers, you’d have to encode files in three formats and still fall back to Flash for viewers without HTML5-compatible browsers. Or you could just continue to solely support Flash and wait a year or two (or more) until the penetration rate of HTML5 browsers exceeds 95 percent, and then reevaluate.
If your only concern was desktop players, this might be a good strategy. Include mobile in the equation, however, and creating an HTML5 player with fallback to Flash might be a great strategy for your on-demand streams.
HTML5 on Mobile Devices
In the two key mobile markets, Apple and Android, HTML5 support is ubiquitous, as is support for the H.264 codec. So the simplest way to enable on-demand playback for Apple and Android devices is to add an HTML5 player on your website using only the H.264 codec, with fallback to Flash. Android and Apple devices would use the HTML5 player, as would desktop viewers running HTML5 browsers that support H.264 playback. All other desktop viewers would fall back to Flash.
While on the topic of mobile, let’s talk about Adobe’s recent mobile-related decision, starting with precisely what they decided to do. Here’s a quote from the Adobe blog that discusses this decision:
“Our future work with Flash on mobile devices will be focused on enabling Flash developers to package native apps with Adobe AIR for all the major app stores. We will no longer continue to develop Flash Player in the browser to work with new mobile device configurations (chipset, browser, OS version, etc.) following the upcoming release of Flash Player 11.1 for Android and BlackBerry PlayBook.”
Adobe will discontinue development of the Flash Player on mobile devices, pushing their key content producers to produce native apps for the mobile platforms. According to the blog post, Adobe will also continue development of the Flash Player on the computer desktop, focusing on markets where Flash “can have most impact for the industry, including advanced gaming and premium video.”
Why the decision to cease development for mobile? There are a number of reasons. The sheer number of Android device configurations made it very expensive to provide device-specific support. Now Adobe has passed the problem of ensuring device compatibility to its app developers. In addition, Adobe was locked out of the iOS market—which doesn’t support Flash—and Microsoft has also said that they won’t enable plug-ins like Flash in its upcoming Windows 8 tablet OS.
In contrast, Android, iOS and Windows 8 all support HTML5, making it the best multiple-platform solution for deploying browser-based content to the mobile market. Adobe saw the writing on the wall and decided to exit a market that they couldn’t affordably and adequately serve.
What’s the key takeaway? At a higher level, simple video playback of on-demand content is becoming commoditized, and it can be performed just as well in HTML5 as in Flash. In addition, HTML5 also has much greater reach, allowing one player to serve mobile and desktop markets—though Flash fallback is clearly necessary on the desktop.
Adobe is positioning Flash as a premium technology that offers many advantages that HTML5 can’t offer, including all those mentioned above. Other noteworthy features that HTML5 doesn’t offer include multicasting and peer-to-peer delivery, which are particularly important in enterprise markets. However, if these features aren’t important to your organization, it’s time to start implementing HTML5 for your on-demand streams—if only with H.264 support to serve the iOS and Android markets.
DASH
DASH stands for Dynamic Adaptive Streaming over HTTP, and it’s an International Standards Organization (ISO) standard that one day may provide standards-based live and on-demand adaptive streaming to a range of platforms—including mobile, desktop, and over the top (OTT) television consoles. It’s a web producer’s dream, since by supporting a single technology, your video can play on all these platforms. The specification enjoys significant industry support, with more than 50 companies contributing to the specification.
Unfortunately, there are some implementation hurdles that may delay or even derail some of this promise. First, at this point, it’s unclear whether DASH will be royalty free. Many companies have contributed intellectual property to the specification. Many companies—such as Microsoft, Cisco, and Qualcomm—have waived any royalties from their contributions, though this is not yet universal. In fact, the current status of DASH in this regard is so uncertain that Mozilla has announced that it’s “unlikely to implement at this time.” Obviously, taking Firefox out of the equation limits the effectiveness of DASH in the HTML5 marketplace.
In addition, while some companies such as Microsoft have publicly announced that they will support DASH once finalized, two critical companies—Adobe and Apple—have not done the same. This isn’t unusual in its own right since neither company typically discusses unannounced products. Still, because these companies dominate the mobile and desktop browser plug-in markets, it’s tough to plot a strategy until you know their intent.
The Bottom Line?
If you’re broadcasting live, HTML5 isn’t an option in the short term. DASH may change things in early 2012, but until we’re certain which platforms will support it and when, you shouldn’t change your existing strategy. For most producers, live broadcasting means one stream (or set of streams) for Flash and another for iOS devices using Apple’s HTTP Live Streaming (HLS). In this regard, Flash should be available for Android devices for the foreseeable future, and Android 3.0 devices should be able to play HLS streams.
For your on-demand streams, it may be time to consider switching over to an HTML5 first with H.264 support with Flash fallback. This is the most efficient mechanism for reaching iOS, Android, and other HTML5-compatible mobile devices while continuing to support legacy desktop browsers.
By Jan Ozer, Sound & Video Contractor
Netflix Sees Cost Savings in MPEG DASH Adoption
"The biggest advantage to us of a standard like MPEG DASH is that everything can be encoded one way and encapsulated one way, and stored on our CDN servers just once. That's a benefit both in terms of saving our CDN costs from a storage perspective and a benefit because you have greater cache efficiency," said Mark Watson, senior engineer for Netflix.
Watson made his comments in a red carpet interview at the recent Streaming Media West conference in Los Angeles, shortly before taking part in a panel on the MPEG DASH specification. MPEG DASH would be a great help to Netflix, he said, because then it could avoid saving several different copies of its entire movie and TV show library.
While there are several different profiles defined in MPEG DASH, Netflix will use the on-demand profile, Watson said, because all of its online content is on-demand. Between the two types of stream segments defined -- MPEG-2 Transport Streams and fragmented MP4 files -- Netflix sides with fragmented MP4. It works well for adaptive streaming and is simpler, he offered.
Netflix, Watson said, contracts with multiple CDNs and allows the client devices to determine which works best for them at any time. The company is also sensitive to the amount of traffic it's putting across networks.
By Troy Dreier, Streaming Media