Upload
duongkiet
View
212
Download
0
Embed Size (px)
Citation preview
Corralling the Chaos
of
Ancillary Data
within
Multiple File Formats
Sara Kudrle
Miranda Technologies
A BELDEN Brand
Ingest and Quality Control
Playout
and Graphics
Publishing for
On Demand
Editing and Content Prep
Live Production
Media Management and Storage
Automation
Linear Channels
Pre- Produced
Content and Ads
WAN Non Linear Delivery WAN
Typical Television Broadcast Facility and Workflow
Broadcast Facilities Have Been Stream / SDI Centric
Ingest and
Quality Control
Playout and Graphics
Non Linear Publishing
Editing and Content Prep
Live
Production
Media Management and Storage
Automation
Linear Channels
Pre- Produced Content and
Ads
WAN
Linear Feeds
Tape (Linear Medium)
Linear Streams
Isolated Islands of IT / File Based Infrastructure
Broadcast Industry Specific SDI Infrastructure
Non Linear Delivery
WAN
Today Facilities Becoming Increasingly “File Based”
Ingest and
Quality Control
Playout
and Graphics
Non Linear Publishing
Editing and Content Prep
Live Production
Media Management and Storage
Automation
Editing and Content Prep
Live Production
Linear Channels
Pre- Produced Content and Ads
WAN
Tape (Linear Medium) Non Linear
Delivery WAN
Media Files
Media Files
Linear Feeds
Linear Streams
Isolated Islands of
IT / File Based Infrastructure
Broadcast Industry Specific SDI Infrastructure
Broadcast Industry Specific SDI Infrastructure
IT / File Base Infrastructure
Contrasting the good ‘ol days versus today
• File wrapper format
“standards” that are too
loose and interpreted in too
many different ways (e.g.
MXF)
WAN Files
• Missing / misplaced meta
data
SDI
• “Iron Clad” tape formats
• “Cast in stone” device
interconnect
(SMPTE SDI) with precise
• No provisions for descriptive
meta data
Then (the good ‘ol days)
Today (the wild west)
The Early Days of File Based
Same Vendor Provides Ingest and Playout
SDI
SDI
SDI
Third party
media server
Homogeneous Environment
ONE Common File Format
INGEST PLAYOUT
STORE
Today’s Reality – The Wild West of Files
SDI
Third party
media server
INGEST
PLAYOUT
STORE
Networked VTRS
WAN File
Delivery
Service
IMX
LXF
?XF Ethernet /
IP Network
MXF
Flv A
Anatomy of a Media File
Ties Essence together & describes it • Structural metadata such as SOM, EOM,
Duration, types of media included in file, etc.
• Descriptive metadata such as creation date, usage
rules, key words
• Time References and Index
• Video (typically compressed)
• Multiple Audio (often compressed)
• Ancillary Data (e.g. Captions, AFD, …)
“Essence”:The actual media which includes:
Wrapper:
Media
File
Two Storage Approaches
Wrapper
Descriptive Metadata
Timecode / index
Video Essence
Audio Essence(s) Audio Essence(s)
Ancillary Data Essence
Method 1:
Embedded Essence
Examples:
• MXF XDCAM
• GXF
• LXF
One File
Structural Metadata
Method 2:
Reference Files / External Essence
Wrapper
Descriptive Metadata
Timecode / index
Video Essence
Audio Essence(s) Audio Essence(s)
Ancillary Data Essence
Structural Metadata
Main
File
Multiple
“Referenced”
Files
Examples:
• Omneon MOV
• MXF P2
• MXF AS-02
Two Storage Approaches
Wrapper
Descriptive Metadata
Timecode / index
Video Essence
Audio Essence(s) Audio Essence(s)
Ancillary Data Essence
Method 1:
Embedded Essence
Examples:
• MXF XDCAM
• GXF
• LXF
One File
Structural Metadata
Method 2:
Reference Files / External Essence
Wrapper
Descriptive Metadata
Timecode / index
Video Essence
Audio Essence(s) Audio Essence(s)
Ancillary Data Essence
Structural Metadata
Main
File
Multiple
“Referenced”
Files
Examples:
• Omneon MOV
• MXF P2
• MXF AS-02
Benefit:
one file includes
everything, easier
media management
Benefit:
Additional essences can be
more quickly added after
the fact (e.g multiple
languages audio and
captions)
Popular Wrapper Formats
Vendor Specific
Wrappers
Associated with
Camera and VTR
Formats
Vendor Specific
Server Formats
Generic
Industry
Wrappers
.MOV
P2 MXF
(Panasonic)
Wrappers
Originating in
PC/Multi-Media
MXF OP1a
MXF-AS-XX
GXF
(Grass Valley)
LXF
(Harris)
IMX/XDCAM MXF
(Sony)
.AVI
MPEG TS File
Television
Distribution
format
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Key Features of Popular Media Wrappers
MPEG-2 TS
MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2 MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Popular Wrapper Formats
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Key Features of Popular Media Wrappers
MPEG-
2 TS MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2 MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Key Features / Functionality
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Key Features of Popular Media Wrappers
MPEG-
2 TS MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2 MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Not
Supported
(Bad)
Strongly
Supported
(Very Good)
Partially
Supported
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Key Features of Popular Media Wrappers
MPEG-
2 TS MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2
MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Observation 1: There are a lot of wrapper formats
and they are obviously very different
Observation 2: No ONE wrapper format supports all
the identified functionality
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Key Features of Popular Media Wrappers
MPEG-
2 TS MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2
MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Observation 3: There are a couple
clear “bad choices”
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
MPEG-2 TS
MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2
MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Observation 4: At first glance MXF appears promising but in
reality has turned out to have major
interoperability issues
Key Features of Popular Media Wrappers
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Key Features of Popular Media Wrappers
MPEG-
2 TS MOV / MP4 AVI
GXF &
LXF Generic
MXF
IMX / XDCAM
MXF P2
MXF
Avid MXF
(OP Atom)
MXF AS-02
Editing
NLE Support
Random Access
Edit-while-ingest
Playout
Transcoder Support
Native Server Support
ANC carriage
Additions/Changes without full remux
Archiving
Descriptive Metadata Carriage
Future-proofness
Observation 5:
Hard-core constraints (Tape fmt Heritage),
Clearly Documented, No Ambiguity
Reasonable bit rate (HD at 50 Mbps),
Wide NLE and Server Support
IMX / XDCAM MXF has turned out to
be a very popular Generic File Format
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Why has MXF Failed and What Are Alternatives?
More
Specific
More
constraints
Better
Generic MXF
LXF Leitch/Harris
GXF Grass Valley
IMX/XDCAM
MXF Sony
Vendor
Specific
Variants
Generic
Variants
• Far too loose &
flexible
• Subject to broad
interpretation
• No hard reference
• Vendor specific
• Some published
some proprietary
• Some overly
constrained
• Harder to Extend
AS-02 Versioning
AMWA
Application
Specifications
AS-03 Program
Delivery AS-10
Production AS-12 Ad
Delivery
AS-11 Contribution
• Constraints on
MXF for different
Applications
• Industry
Standard ?
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
AMWA MXF AS-XX Application Specifications
• Defined by AMWA (Advanced Media Workflow Association)
• AS-XX MXF Specifications Define a set of rules / constraints on MXF
to suit a particular use
• Defines MXF use for different situations / applications
• Shims constrain further to meet needs of a user
MXF
Constrained
Parameters,
and Values
Available Codecs
Constraints on
Codecs and
Bit-rate ranges
AS-XX
Application Specification
User A SHIM User B SHIM
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
AS-XX Format Summary
Number
Application
Domain Essences CODECs
HD Bit
Rates (Mbps)
Meta
Data
Audio
Chans
AS-02 Versioning External Not Specified Depends
on
Codec
Not
Specified
AS-03 Program
Delivery Embedded MPEG2, H.264
LGOP 5-50
Mb/s 2-16
AS-10 Production
(e.g. Sony
XDCAM MXF)
Embedded MPEG-2 MP@HL,
MPEG-2 422P@HL
MPEG-2 MP@H14
LGOP
Depends
on
Codec
2-8
AS-11 Contribution Embedded
?
AVC Intra 100 Mbps 2-64
AS-12 Ad Delivery Embedded MPEG2, H.264
LGOP
5-50
Mb/s
2-16
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Different AS-XX Specifications for Different Apps
AS-12 Ad Delivery
Regional Playout
Facility
Central Broadcaster / Playout Content Creation
AS-02 Versioning
AS-10 End to End ( Acquisition – Production – Playout)
AS-11 Contribution
Aquisition Post Contribution Playout
Preparation Distribution
AS-03 Program Delivery
Packaging of multiple
versions of same program
Multiple Languages,
Multiple Editorial Versions
Delivery of finished
programs. Single File, single
audio language, constrained
codecs
Delivery of finished
Commercials. Identical to
AS-03 but with Advertising
specific Meta Data
Enable Interoperation and an end to end
workflow without transcoding.
Specifies Long GOP MPEG2 codec at a decent
bit rate
Designed as extension to SMPTE RDD 9 and
is used by IMX & XDCAM-HD
Vendor-neutral file format
for the delivery of finished program material.
Similar to AS-03 but features higher bit rate
AVC-Intra video and uncompressed Audio
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Different AS-XX Specifications for Different Apps
AS-12 Ad Delivery
Regional Playout
Facility
Central Broadcaster / Playout Content Creation
AS-02 Versioning
AS-10 Production using Long GOP Codec (e.g. XDCAM)
AS-11 Contribution
Aquisition Post Contribution Playout
Preparation Distribution
AS-03 Program Delivery
Delivery of finished
programs. Single audio
language, constrained
codecs
= User A SHIM User B SHIM Observation :
Even within versions, AMWA has
specified that “An Application
Specification can be further constrained
to meet the needs of a user through use
of a SHIM”
In other words, they are
different
© 2012 SMPTE · The 2012 Annual Technical Conference & Exhibition · www.smpte2012.org
Why has MXF Failed and What Are Alternatives?
• Constraints on
MXF for different
Application
• Industry
Standard ? More
Specific
Better
MXF
LXF Leitch/Harris
GXF Grass Valley
IMX/XDCAM
MXF Sony
Vendor
Specific
Variants
Generic
Variants
AS-02 Versioning
AMWA Application
Specifications
AS-03 Program
Delivery AS-10
Production AS-12 Ad
Delivery
AS-11 Contribution
Ancillary Data
Four main ways to carry Ancillary data in files
In the VBI portion of video (IMX only)
In the MPEG GOP Header (limited)
In the wrapper (limited)
As a s436m payload
ANC Data In Legacy File Formats Was Inconsistent
and Incomplete
HANC
Audio
T/C
VANC AFD, Dolby, CC, VCHIP,
SCTE Triggers, Private Data
Active
Video
File Wrapper
Video Essence
Audio Essence (s)
Some ANC in Codec Header
Some ANC in Wrapper Meta
SDI
Time Code
And
/ Or
Some,
Not All
Some ANC in VBI (SD only)
Legacy File
Formats
Ancillary Data
Dolby
Metadata
SCTE-104
Ad Triggers
Broadcast
Flag
OP47
Teletext Private
Data
$
XDS
V-Chip
WSS
AFD
Timecode
Closed
Captions
Dolby
Metadata
SCTE-104
Ad Triggers
Broadcast
Flag
OP47
Teletext Private
Data
$
VBI (IMX Only) XDS
V-Chip
WSS
AFD
Timecode
Closed
Captions
Ancillary Data
Dolby
Metadata
SCTE-104
Ad Triggers
Broadcast
Flag
OP47
Teletext Private
Data
$
VBI (IMX Only)
Codec Header
XDS
V-Chip
WSS
AFD
Timecode
Closed
Captions
Ancillary Data
Dolby
Metadata
SCTE-104
Ad Triggers
Broadcast
Flag
OP47
Teletext Private
Data
$
VBI (IMX Only)
Codec Header
XDS
V-Chip
WSS
AFD
Timecode
Closed
Captions
Wrapper
Metadata
Ancillary Data
Dolby
Metadata
SCTE-104
Ad Triggers
Broadcast
Flag
OP47
Teletext Private
Data
$
VBI (IMX Only)
Codec Header
XDS
V-Chip
WSS
AFD
Timecode
Closed
Captions
Wrapper
Metadata Key ANC
Data not
Captured
Ancillary Data
ANC Data In Legacy File Formats Was Inconsistent
and Incomplete
HANC
Audio
T/C
VANC Various ANC Data
Active
Video
Original Source Video Legacy File
Formats
File Wrapper
Video Essence
Audio Essence (s)
Time Code
Playback Video
Ingest
HANC
Audio
T/C
VANC Various ANC Data
Active
Video
Playback
Missing
ANC Data
Can’t Easily
Add or Modify
ANC Data to File
Original
ANC Data Subset of
ANC Data
Codec Header
Wrapper
Metadata
SMPTE 436m
Key ANC
Data NOW
Captured
XDS
V-Chip
WSS
AFD
Timecode
Closed
Captions Broadcast
Flag
OP47
Teletext Dolby
Metadata
Private
Data
SCTE-104
Ad Triggers $
Ancillary Data
SMPTE 436 Dramatically Improves ANC Carriage
HANC
Audio
T/C
VANC AFD, Dolby, CC, VCHIP,
SCTE Triggers, Special Data
Active
Video
SMPTE 436 ANC Essence
SDI
MXF File with
436M Essence
File Wrapper
Video Essence
Audio Essence (s)
Time Code
ALL
ANC
Wrapper
Video Essence
Audio Essence
SMPTE 436 ANC Essence CC
AFD
DOLBY
SCTE
Propri
Frame Accurate Index DID
ANC
Data
Line
Num H
C
S
MXF File with
436M Essence
CC
AFD
DOLBY
SCTE
Propri
CC
AFD
DOLBY
SCTE
Propri
CC
AFD
DOLBY
SCTE
Propri
Typical SMPTE
436 Packet
SMPTE 436 Dramatically Improves ANC Carriage
HANC
Audio
T/C
VANC DID tagged streams
Active
Video
VANC
Audio
T/C
VANC DID tagged streams
Active
Video SMPTE 436 ANC Essence
MXF File with
436M Essence
File Wrapper
Video Essence
Audio Essence (s)
Time Code
Original Source Video Playback Video
Ingest Playback
Add and
Modify ANC Easily
Original
ANC Data
Original
ANC Data
Augmented
ANC Data
Original
ANC Data
Augmented
ANC Data
SMPTE 436 Dramatically Improves ANC Carriage
Ancillary Data Using SMPTE 436M
Standardized by SMPTE in 2006
Can carry any DID/SDID (i.e. future-proof)
Carried as a separate essence payload in MXF
Can be embedded or referenced
Easy to re-wrap
Easy to Add / Modify
Supported in MXF and AS-N Application Specifications
Now being added to SMPTE RDD9 for AS-10 / IMX/XD-CAM
SMPTE-436M
Ancillary Data
Carriage in
Files
It solves the biggest file interoperability
problems we have experienced
The most important thing to happen
to File Based since the folder was invented
The authors are solely responsible for the content of this technical presentation. The technical presentation does not necessarily reflect the official position of the Society of Motion Picture and Television Engineers (SMPTE), and its printing and distribution does not constitute an endorsement of views which may be expressed. This technical presentation is subject to a formal peer-review process by the SMPTE Board of Editors, upon completion of the conference. Citation of this work should state that it is a SMPTE meeting paper. EXAMPLE: Author's Last Name, Initials. 2011. Title of Presentation, Meeting name and location.: SMPTE. For information about securing permission to reprint or reproduce a technical presentation, please contact SMPTE at [email protected] or 914-761-1100 (3 Barker Ave., White Plains, NY 10601).
SMPTE Meeting Presentation
Corralling the Chaos of Ancillary Data within Multiple File Formats
Corralling the Chaos of Ancillary Data within Multiple File Formats
Sara Kudrle, Sr. Software Engineer Miranda Technologies, 125 Crown Point Ct. Grass Valley, CA 95945, [email protected].
Written for presentation at the SMPTE 2012 Technical Conference, Hollywood, CA.
Abstract. The baseband Serial Digital Interface (SDI) workflows and formats were “iron clad” and well defined. These were the "good old days" when devices interconnected with ease thanks to the rigor and breadth of SMPTE standards for SDI. Nowadays, with the great flexibility of file based media workflows and the multitude of formats needed for different applications, the industry is dealing with incompatible wrappers and inconsistent or non-extendable Ancillary (ANC) data carriage. This paper will look at these evolving workflows and the resulting "Wild West" of files. More specifically, the paper explores the challenges faced with the handling of ANC data such as Active Format Descriptor (AFD) , closed captions, ad insertion triggers, audio metadata, etc., within various file formats. The paper then describes the unified and extensible approach offered by SMPTE 436M for the carriage of ANC data within Material Exchange Format (MXF) wrapped files. Could SMPTE 436M be the champion the industry needs to restore order to the Wild West and corral some of the chaos?
Keywords. SMPTE 436M, AFD, ANC, Ancillary Data, ad insertion, closed captions, metadata, MXF, standards, SDI, file based workflows, wrappers.
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
2
Introduction As broadcast facilities move towards file based workflows, they are leaving behind the well defined linear workflows that they are accustomed to. They are instead entering the Wild West of Files with various formats and buckets of bits that are easy to define and re-define leading to interoperability problems. Gone are the days of linear workflows starting with linear feeds and linear mediums, with tapes on the input side and linear streams on the output side. These are being replaced by media files.
At some point we started seeing isolated islands of infrastructure based on files. These were generally handled by one vendor from start to end. This homogenous environment made file based transactions relatively painless since the definition of the bits was in the hands of one entity. Nowadays these islands have multiplied and are occupied by many vendors with many definitions of the same bits. We are receiving preproduced content and ads as media files; many different types of media files. What ensues is eventually chaos as the definitions of ancillary data in particular clash.
How can we corral the chaos? We need one definition to follow; a standard, if you will. SMPTE 436M is such a standard.
The good old days. Let’s go back and look at the good old days. These were the days of SDI and hardware based definitions. These were the days of “Iron Clad” tape formats. The “Cast in Stone” device interconnect (SMPTE SDI) with precise specifications on Video, Audio and Ancillary Data. Having everything defined in the hardware made interoperability with other devices such as third party devices, manageable and simple. They just connected! However, there were no provisions for descriptive meta data.
Figure 1. Use the Figure Caption Style for a caption below each figure, outside of the graphics
box. The graphic itself is in the Figure Style.
Then started the days of file based workflows. Suddenly the media was defined in bits in software. These bits could be easily defined and redefined by third parties wanting to make use of spare bits. Due to a homogenous environment a vendor could get away with it since their devices understood the definitions of the bits and handled them appropriately. For in the early days of file-based playout systems, most often users would continue using their existing SDI workflow and ingest from a tape of satellite feed, transforming the data into files, and then storing in a central storage before sending to a playout device. This even included server to server transfers. Often all of these devices from ingest to playout were supplied by the same vendor making the workflow appear smooth and simple. But that solution has been displaced.
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
3
The reality of today. Today we have multiple Codec formats and multiple file wrapper formats. We have metadata that can be misplaced or placed in various locations depending on format. We have metadata that can be missing or end up missing after passing through various equipment. We have file wrapper format “standards” that are flexible (perhaps too flexible) and allow inpretation in many different ways. One that comes to mind is MXF (Material eXchange Format): a fantastic idea but someone forgot to shut the barn door, letting formats run wild.
At the same time, we have content being ingested from different sources using these various, different formats. For instance, Networked VTRs using IMX and third party media servers using LXF and Satelite feeds putting SDI in a flavor of MXF and then files coming in over the Wide Area Network (WAN). All of these are being stored together in centralized servers and then shuffled to the same playout server with varying degrees of success, depending on their particular format flavor. In some cases these file are not interoperable and must first pass through another device to convert them.
Today’s Reality – The Wild West of Files
SDI
Third partymedia server
INGEST
PLAYOUT
STORE
Networked VTRS
WANFileDeliveryService
IMX
LXF
?XFEthernet /IP Network
MXF Flv A
Figure 1. Use the Figure Caption Style for a caption below each figure, outside of the graphics
box. The graphic itself is in the Figure Style.
As the amount of content received or transferred as files increases, becoming the normal mode of operation, we will need to rely on files being interoperable in order for workflows to continue to be smooth and functional. Currently, the workflows often depend on the destination system to accurately and nimbly interpret the correct format as these files race towards them. Welcome to the “Wild West of Files!”
What Does a Media File Look Like? Basically a media file is made up of a wrapper and forms of essence. The wrapper provides a description of the essence and ties it all together. Meanwhile, the essence is the actual media or the content of the file. Let’s look at these more closely, starting with the wrapper.
The wrapper, which describes the media and ties it together includes the structural metadata such as SOM, EOM, duration, types of media included in the file, also known as essence descriptors, and other structural information. The wrapper can also include descriptive
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
4
metadata such as creation date, usage rules, key words and other information that describes the contents. The wrapper usually has time references and index as well.
The “Essence” is the actual media or the actual contents of the file such as the video which is typically compressed and the audio. There can be multiple audio essences in a file and these are often compressed. Another type of essence would be Ancillary Data such as captions, Active Format Description (AFD), Dolby, Ad Insertion Triggers, Interactive Triggers, and any other data that would enhance or be useful to the other media essences contained in the file.
Storage?
There are two different ways that these are generally stored. With the first approach, the wrapper “wraps” around the various forms of essence and keeps them embedded all in one file. This is useful for keeping information together which makes media management simpler. An example of this method is MXF XDCAM, GXF and LXF.
Figure 1. Use the Figure Caption Style for a caption below each figure, outside of the graphics
box. The graphic itself is in the Figure Style.
The second method keeps the wrapper which can also be called a reference file separate from the essence which are kept in external files. There could be one or several external files depending on the various forms of essence required for a particular media. By keeping them separate, adding, deleting or modifying the essence containers is simpler since they are easily accessible. This makes it easy to add extra language essences or changing the captions without having to reprocess the whole file. Examples of this method include Omneon MOV, MXF P2 and MXF AS-02.
Figure 1. Use the Figure Caption Style for a caption below each figure, outside of the graphics
box. The graphic itself is in the Figure Style.
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
5
What about a Hybrid approach? Another complication can be hybrid files which comprise of both storage methods at once. These files embed some of the essence withing the wrapper while others are referenced as external essence files.
Sounds like File Format Fiasco! Since there are so many different forms of essence and so many variants of codecs and wrapper formats along with differing storage techniques, one of the most important and difficult challenges in a file-based world is choosing the correct set of variables for a particular operation. Another challenge is making certain that all of these different variables play nice together and do not further complicate the situation by causing interoperability problems.
How to choose a wrapper? With all the formats available, how are we supposed to choose? Let’s look at a few of the wrapper formats and what they are used for within the three main areas of interest: editing, playout and archive. The first one that comes to mind is the MPEG Transport Stream file which is one of the oldest formats still in use and is actually a television distribution bitstream captured to a file. This format may be future proof, but it does not support editing nor the carriage of descriptive metadata needed for archive.
Another set of old formats, MOV and AVI, originated as personal computer formats over 20 years ago. Although these formats provide some editing support they do not handle editing while ingesting and they only have limited support for ancillary data. AVI does not support necessary characteristics for playout such as native server support, allowing changes without requiring a full remux, and ANC carriage.
Now the next two formats, XDCAM and P2 MXF, are closely associated to VTR and camera manufacturers
Key Features of Popular Media Wrappers
MPEG-‐2 TS
MOV / MP4 AVI
GXF&LXF
Generic MXF
IMX / XDCAM MXF
P2MXF
Avid MXF
(OP Atom)
MXFAS-‐02
EditingNLE SupportRandom AccessEdit-‐while-‐ingestPlayoutTranscoder SupportNative Server SupportANC carriageAdditions/Changeswithout full remuxArchivingDescriptive Metadata CarriageFuture-‐proofness
KEY:
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
6
Not Supported Partially Supported Strongly Supported Figure 1. Use the Figure Caption Style for a caption below each figure, outside of the graphics
box. The graphic itself is in the Figure Style.
Next are the generic industry wrappers such as MXF OP1a and AS-02. The AS-02 format does well with archive and playout but is lacking Non Linear Editing support.
And, of course, there are the vendor specific house formats, such as GXF and LXF and Sony XDCAM MXF, that can be difficult to decipher especially when they contain areas of undocumented metadata.
One thing that becomes obvious looking at all of these formats is that not one of them does everything well. Some are strong in certain areas and weak in others. They each have their strengths based on which functionality they were designed to serve. Now, if we were to choose one wrapper to fit everywhere, it is clear that a few of the mentioned formats are bad choices. For intstance, MPEG2 TS with its lack of support for editing and AVI with its lack of support for playout clearly would not be first choices in covering all areas. Other choices such as generic MXF types have a certain appeal but in reality are plagued by interoperability problems. One of the generic MXF formats has become a surprisingly popular format. That format would be IMX/XDCAM MXF. It works well for editing and has strong transcoder and native server support for playout and is future proofed for archiving. If only it had a little more support for Ancillary data and descriptive metadata carriage, it would be perfect!
MXF keeps popping up. What is MXF? MXF (Material eXchange Format) is an open file format wrapper designed to improve interoperability among content creation components, video servers and other devices in a file based work-flow. MXF was designed to make the interchange of Essence (Video, Audio and Data), Descriptive Metadata and Structural Metadata be simple and easy. It is a file transfer format which is not compression scheme specific and also metadata aware. One important distinction of MXF over other wrappers is its ability to support descriptive metadata which is becoming increasingly important as we archive the plethora of essences out there and wish to retrieve them later. MXF has become the de facto standard and should allow devices to work together despite the manufacturer, which should improve the smoothness of file based workflows. However different use cases that have arisen have given rise to MXF variants.
MXF Variants? MXF covers a lot of ground in making file transfers seamless, however, its generic definition does not suit all cases. First of all, the generic variant is generally regarded as too loose and flexible. There are no hard references and it is subject to broad interpretation.
This has caused Application Specifications within MXF to crop up. There is MXF AS-02 which is specifically designed for versioning and there is AS-03 for program delivery, AS-10 for production, AS-11 for contribution and AS-12 for ad delivery. These specifications put constraints on the MXF format.
Then, there are some vendors whom have decided they need to create their own version of MXF to meet their unique needs, such as Grass Valley with GXF and Leitch (aka Harris) with LXF. However, these constrain the format even more and make it hard to extend. Some of
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
7
these are proprietary and are not even published. Furthermore, Application Specifications can be further confined to the particular needs of a user by the use of SHIMs. Essentiall, two versions of the same specification can be different.
What about ANC Data? As we become more and more file based and file based delivery and playout becomes required by broadcast stations, the ANC Data becomes increasingly important. This auxillary data provides us with the necessary information to compliment the essence and delivery it in its entirety and expected format reliably. A video without its closed captions for the hearing impaired would not bring much value to a deaf person. If the video is missing its ad insertion triggers, the station will not be making much income. If the Active Format Descriptor (AFD) is wrong or missing the odds of it being displayed correctly decrease substantially. Without the correct audio metadata, the sound will not be played correctly.
Legacy file formats do not meet all the needs. The ANC data of legacy formats was inconsistent and incomplete. Some of the data could be carried in the wrapper metadata and some of the data could be carried in the codec header or in the VBI for SD.
The most common types of data needed for playout are timecode, AFD, Closed Captions, XDS, V-Chips, and WSS. All of these can be carried in the VBI, but this is only possible with IMX files. In order to get past this limitations some codecs have started encoding this data as part of the MPEG user data. This approach is not practical since the data is not always supported and therefore the files are not interoperable. Another way to carry the data is with the wrapper structural data, but this approach is limited and not very useful. In the end, during playout, it is easy to end up with missing, incomplete or incorrect ANC data. Plus key ancillary data such as Broadcast Flag, OP47 Teletext, Dolby Metadata, SCTE Ad Triggers and other private data are not captured at all.
Has ANC Data been neglected? Yes! Ancillary data in the file-based domain is akin to audio in the SDI realm. Both were considered the poor second cousin, meaning that not much thought was given to them. They were an afterthought and therefore one could say that ANC data has been neglected. When setting up new HDTV environments, the video codecs are payed close attention to, but often the audio and now especially the ANC data are easily overlooked and considered not as important and are forgotten until later.
What is SMPTE 436M? SMPTE 436M is a standard that defines a space for Ancillary data within an MXF file. It defines mappings for VBI Lines and Ancillary data packets. SMPTE 436M works with MXF to provide a more solidly defined space in which to carry ANC data. In order for systems to work together with different types of data, the data has to be explicity defined and more reminiscent of the rigid hardware standards that facilitated SDI interchanges.
This standard works to define what types of data goes where and makes it possible for multiple devices from multiple vendors to make sense of the data they are passing between them. This standard defines the sequence order of the ANC elements, such as audio and data, and the corresponding picture elements.
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
8
Why is it useful? SMPTE 436M gives a definite set of rules to follow which will make interoperability more predictable and reliable between various devices. If the area where the Ancillary data is clearly defined and defined in a rigid way that will not change regardless of the essence that it contains nor due to the functionality that it is used for then it will have dependable results.
SMPTE 436M appears to be a functional solution for the problem of ancillary data. It allows the carriage of all basic ancillary data, such as captions and AFD and timecode. It also allows for the carriage of key data left out of legacy file systems such as OP47 Teletext, Dolby Metadata, SCTE-104 Ad Triggers and much more. Finally there is a space for ANY ancillary data to be placed which should help future-proof this standard for a while.
With SMPTE 436M all ancillary data has a defined space in the file format and no longer has to be scattered across sections. This drastically improves the carriage of the data and adds to it being reliably preserved, discovered and decoded correctly. SMPTE 436M describes the line number where each piece of data will reside. SMPTE 436M captures all ancillary data and allows the augmented data to be added and modified easily.
Can the Wild West of Files be Tamed? With the help of guidelines from SMPTE 436M, the Wild West of Files can indeed be tamed. The Wild West of Files can have their formats corralled into a usable subset that will help with interoperability. The files, using SMPTE 436M, can carry ANC data throughout the workflow and have it maintained and usable.
By having solid definitions for ancillary data reminiscent of the “iron clad” SDI days, the interchange of media between various devices could become seamless and invisible to the user. The users in a sense could become format agnostic and not have to worry where data is maintained and instead concentrate on the creative aspects or the functional aspects of their jobs.
What can be done to further the cause? Make sure that your vendors are aware of SMTPE 436M. Make sure that your vendors are making their products work within the specifications of SMPTE 436M to allow interoperability between devices in your workflow and allowing the carriage of ANC data from one device to another without any hiccups. The SMPTE 436M standard has been around for many years, becoming a standard in 2006. It is currently undergoing a retrofit within SMPTE to make it up to date and take into account 3D. If you have any input, join the SMPTE Standards Community and get involved in this standard. While the SMPTE 436M standard is being reworked, careful attention is being paid to the use of language in order to avoid the standard being interpreted in multiple ways. A great standard works when it is well-known and understood and its interpreted correctly each time allowing it to always works the same way. That is the goal of SMPTE 436M!
Conclusion The explosion of file formats and wrappers along with the need for more ancillary data at the same time that broadcasters are requiring files at ingest is making the importance of file interoperability paramount to success. In order for files to interchange reliably and correctly broadcasters need to follow some clear cut rules and definitions. Up until the use of SMPTE 436 and MXF, files have been defined in various documented and undocumented ways leading to unknown sets of ancillary data being lost, incorrectly decoded, or just misplaced. It is a wild
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.
9
west free for all out in File Land. But SMPTE 436M can corral the ancillary data into a manageable solution so that File Land can stop being the Wild West of Files and instead become an Ancillary Data Utopia.
Format for Reference and Bibliography Entries Bibliography entries are in alphabetical order; references, in numerical order according to their citations in text. In bibliographies, the last name is listed first; if there is no author, the entries begin with "Anon." and are alphabetized by title. In references, the listing is as in the examples below. If more than three authors are listed, use the first name followed by et al. The following are examples of format for various types of publications or non published documents. Use standard abbreviation for journal articles when known; otherwise, spell out.
Journal Articles M. C. Gruszka, “Bumps on the All-Digital Road,” TV Technology, 12:12-39, June 1994.
Articles Published in Proceedings M. Shibata, “A Temporal Segmentation Method for Video Sequence,” Proc. SPIE Visual Comm. and Image Proc. '92, pp. 1194-1205 1992.
Books L. S. Birks, Electron Probe Microanalysis, Wiley: New York, 1982.
Chapter in a Book P. E. Wilcock, “Analog Components,” in Video Pictures of the Future, SMPTE: Scarsdale, N.Y., 1983.
Unpublished Paper W. E. Glenn, J. Marcinka, and R. Dhein, “Subband Coding Compression System for Program Production,” presented at the 136th SMPTE Technical Conference, Los Angeles, Oct. 1994.
Dissertation or Thesis T. L. Gilbert, “Rate of Decay of Auditory Sensation,” Ph.D. dissertation, Dept. of Chemistry, University of Massachusetts, 1984.
Standards SMPTE 259M, “Television -10-Bit 4:2:2 Component and 4fsc NTSC Composite Digital Signals—Serial Digital lnterface," SMPTE Mot. Imag. J., 102:174-179, Feb. 1993.
Copyright © 2012 Society of Motion Picture and Television Engineers. All rights reserved.