EIA-708
Encyclopedia
CEA-708 is the standard for closed captioning for ATSC digital television
streams in the United States
and Canada
. It was developed by the Electronic Industries Alliance
.
Unlike most DVB captions, CEA-708 captions are textual like traditional Line 21
captions. However, unlike Line 21 captions, CEA-708 captions include most of the Latin-1 character set, and include stubs to support full UTF-32 captions, and downloadable fonts.
CEA-708 captions are injected into MPEG-2
video streams in the picture user data
. The packets are in picture order, and must be rearranged just like picture frames are. This is known as the DTVCC Transport Stream. It is a fixed-bandwidth
channel that has 960 bit/s allocated for backward compatible Line 21 captions, and 8640 bit/s allocated for CEA-708 captions. The ATSC A/53 Standard contains the encoding specifics. Either the EIT or the PMT tables contain a caption descriptor which tells you what language the captions are in and tells you if they are regular captions or 3rd grade level captions for language learners.
CEA-708 caption decoder
s are required in the U.S. by FCC regulation in all 13" (33 cm) diagonal or larger digital televisions. Further, some broadcasters are required by FCC regulations to caption a percentage of their broadcasts.
This layering is based on the OSI Protocol Reference Model:
This section will describe the various packets, the Coding Layer and Presentation Layers are described in the remainder of this document.
When the picture user data contains a caption user_data_start_code is set to 0x00001B2, the ATSC_identifier is 0x47413934, and the user_data_type_code will be 0x03. If the user_data_type is not 0x03, or 0x06 for bar data, then the packet will be terminated with the bytes 0x0, 0x0, 0x1. If the data describes caption data, it will be as follows:
user_data
Marker bits and reserved bits should all be set by default. If the additional_data_flag is set then the ATSC_reserved_user_data will be at the tail of the packet, terminated by the bytes 0x0,0x0,0x1. If the process_cc_data_flag is set the cc_data_pkt's should be parsed as follows:
At some future time the process_em_data_flag will indicate whether to process the em_data bit string.
If cc_valid is not set the cc_data_pkt's should be considered padding and discarded.. If it is set, cc_type will be one of four values NTSC_CC_FIELD_1 = 0, NTSC_CC_FIELD_2 = 1, DTVCC_PACKET_DATA = 2, DTVCC_PACKET_START = 3. If it is either 0 or 1, the cc_data fields should be interpreted as EIA-608
Captions. If cc_type is 3 then a decoder should begin assembling a Caption Channel Packet with the cc_data as described below, and if the cc_type is 2 it should append the cc_data to any Caption Channel Packet being assembled. If a DTVCC packet is already being assembled and either cc_valid is set and the cc_type is 3 or cc_valid is clear and cc_type is 2 or 3, then the packet should be considered complete.
NOTE: In a caption decoder cc_data packets must be reassembled in the correct order to create the DTVCC packets. The standard is not clear on this, but it appears this should be in frame display order, not encoded frame order. This means in encoder DTVCC Packets should probably be broken up and inserted into the picture user data as cc_data packets in display order as well.
NOTE: To avoid this bug in the CEA-708 standard some encoders encode captions only
on one frame type, such as only B frames, or only I frames, since if you only use one frame type the frame display and frame encoded order are the same.
Within the packet_data, there is only one type of packet. This is known as the Service Block. This further subdivides the DTVCC Transport Stream into 63 substreams, each of which describes a discrete captioning service. Service 1 is designated as the Primary Caption Service, while Service 2 is the Secondary Language Service. The Caption Descriptor describes any other services offered.
Optional Data 1 is only present if service_number is 7, in which case the extended_service_number should be used instead of the service_number. Optional Data 2 is present only if block_size is greater than 0. However if block_size is 0, the service_number must be zero as well, and then this is known as a Null Service Block Header. A Null Service Block Header can be used for padding.
Note: Service Blocks may not cross Caption Channel Packet Boundaries. This means each Caption Channel Packet can be parsed without keeping any state for the Service Blocks themselves.
. There are four logical code sub-groups: CL, GL, CR, and GR. These each have single and multi-character code sets.
Whenever a command character is seen any text accumulated in the parser should be flushed. Since text might need to be flushed when there is no command pending, there is a null command known as the ETX command in the C0 command set.
There are also two special commands, the Reset and DelayCancel. These must be parsed with lookahead. A Delay command issued previously can be canceled at any time with a DelayCancel command, so once a Delay is seen a decoder must look ahead for a DelayCancel, and only look for a DelayCancel. A Reset command on the other hand is sent to break out from an unknown decoder state and all data before it must be ignored.
NUL, BS, FF, and CR are interpreted as they are in ASCII control codes. HCR is
undefined in the CEA-708-A standard. ETX is the NULL command mentioned earlier,
which is used to flush text to the current window when no other command is pending.
EXT1 is used to escape to the 'C2', 'C3', 'G2', and 'G3' tables for the following
byte. Finally, P16 can be used to escape the next two bytes for Chinese and other
large character maps.
It should be noted that all characters in the range 0x10-0x17, which currently
includes EXT1, are followed by one byte which needs to be interpreted differently.
And, all characters in the range 0x18-x1f, which currently includes P16, are followed
by two bytes that need to be interpreted differently. If a decoder encounters one
of these and does not know what to do, it should still skip the next byte or two,
as appropriate, before continuing.
The C1 Table contains all the currently defined caption commands.
These will be described in detail in the next section.
The G0 Table consists of ASCII characters for the most part. SP here is short hand for
Space, and MN is a musical note, which replaces the Delete command code in ASCII.
The G1 Table is basically the ISO 8859-1 Latin-1 character set. Note character 0xa0 is the non-breaking space
, which is to be used to prevent word wrap from separating two words onto separate lines.
TSP and NBTSP are the Transparent Space, and Non-Breaking Transparent Space, respectively. The G2 Table contains miscellaneous characters that may not be displayed in all browsers. BLK indicates a solid block which fills the entire character block with a solid foreground color.
If a window is not locked (see Define Window) and the SMALL font is in effect the location can be outside the otherwise valid addresses.
Fill Color is specified with 6 bits, 2 for each of blue, green and red. The lowest order bits are for blue, the next two for green and the highest order bits represent red. Fill Opacity is represented by two bits, they represent SOLID=0, FLASH=1, TRANSLUCENT=2, and TRANSPARENT=3. The window's Border Color is specified the same way. However, the Border Type is split into two fields. They should be combined, with border type 01 representing the low order bits, and border type 2 the high order bit. Once combined the Border Type has 6 valid values: NONE=0, RAISED=1, DEPRESSED=2, UNIFORM=3, SHADOW_LEFT=4, and SHADOW_RIGHT=5.
Colors, text painting, effects, and border types can be customized with the SetWindowAttributes and SetPenAttributes commands. However, the caption provider may wish to use predefined standard window styles. A set of predefined styles will be hard stored in receivers. This set will anticipate the most widely used types of caption windows in order to conserve caption channel bandwidth by eliminating the need to transmit superfluous SetWindowAttributes and SetPenAttributes commands.
Predefined window and pen styles can be specified by the window style and pen style ID parameters in the DefineWindow command.
The parameters are as follows.
These are used to tell the caption decoder how to expand the text box as text is added to a caption window. A window is assigned an anchor point, or location and an anchor ID. If the anchor point is say 0,0, and the anchor ID is 0, then the window will expand down and right from the upper left corner of the caption area. If the anchor point is 50%,50% and the anchor ID is 4 the window will expand equally in all directions from the center of the caption area.
characters in this standard. Below are some font examples, for more see the Wikipedia Fonts article.
The window size should be scaled based on the font size. With this in mind, rows longer than 32 characters are discouraged even on 16:9 ratio screen so that larger than specified fonts may be selected by the user.
Totally four combinations are provided
1)Row /Column locked
2)Row Unlocked Column locked
3)Row locked and Column unlocked
4)Row unlocked and column unlocked
1)Row /Column locked: If both rows and columns are locked then window size in terms of columns and rows can’t be extended.
For a window if number of rows and columns are defined like 3 and 10
Then the text ‘ROWS AND COLUMNS ARE NOT LOCKED FOR EVER AND EVER AND EVER ‘which comes in the 0 row looks like below (assume that word wrapping is disabled)
1. ROWS AND C
2.
3.
Since both are locked, text cont be extended beyond 10 columns and also row cont be extended beyond 0.
2. Row Unlocked /Column locked: In this case window can be extended up to max row given in the window define command.
The same above text will look like below
1. ROWS AND C
2. OLUMNS ARE
3. NOT LOCKED
Row is unlocked so text can be extended up to max rows of a define command.
3. Row locked and Column unlocked: In this case window can be extended up to max number columns.
As per the CEA-708 standard Max number of columns for any window is 32.
The same above text then look like below
1. ROWS AND COLUMNS ARE NOT LOCKED
2.
3.
Column is unlocked so text can be extended up to max columns.
4.Row unlocked and column unlocked: In this case window can extended in terms of rows are columns.
The same above text then look like below
1. ROWS AND COLUMNS ARE NOT LOCKED
2. FOR EVER AND EVER AND EVER
Since both are unlocked so the text can extended up to 32 columns and as well as total rows.
Digital television
Digital television is the transmission of audio and video by digital signals, in contrast to the analog signals used by analog TV...
streams in the United States
United States
The United States of America is a federal constitutional republic comprising fifty states and a federal district...
and Canada
Canada
Canada is a North American country consisting of ten provinces and three territories. Located in the northern part of the continent, it extends from the Atlantic Ocean in the east to the Pacific Ocean in the west, and northward into the Arctic Ocean...
. It was developed by the Electronic Industries Alliance
Electronic Industries Alliance
The Electronic Industries Alliance was a standards and trade organization composed as an alliance of trade associations for electronics manufacturers in the United States. They developed standards to ensure the equipment of different manufacturers was compatible and interchangeable...
.
Unlike most DVB captions, CEA-708 captions are textual like traditional Line 21
EIA-608
EIA-608, also known as line 21 captions, used to be the standard for closed captioning for NTSC TV broadcasts in the United States and Canada...
captions. However, unlike Line 21 captions, CEA-708 captions include most of the Latin-1 character set, and include stubs to support full UTF-32 captions, and downloadable fonts.
CEA-708 captions are injected into MPEG-2
MPEG-2
MPEG-2 is a standard for "the generic coding of moving pictures and associated audio information". It describes a combination of lossy video compression and lossy audio data compression methods which permit storage and transmission of movies using currently available storage media and transmission...
video streams in the picture user data
MPEG user data
The MPEG user data feature provides a means to inject application-specific data into an MPEG elementary stream. User data can be inserted on three different levels:*The sequence level*The group of pictures level*The picture data level...
. The packets are in picture order, and must be rearranged just like picture frames are. This is known as the DTVCC Transport Stream. It is a fixed-bandwidth
Bandwidth (computing)
In computer networking and computer science, bandwidth, network bandwidth, data bandwidth, or digital bandwidth is a measure of available or consumed data communication resources expressed in bits/second or multiples of it .Note that in textbooks on wireless communications, modem data transmission,...
channel that has 960 bit/s allocated for backward compatible Line 21 captions, and 8640 bit/s allocated for CEA-708 captions. The ATSC A/53 Standard contains the encoding specifics. Either the EIT or the PMT tables contain a caption descriptor which tells you what language the captions are in and tells you if they are regular captions or 3rd grade level captions for language learners.
CEA-708 caption decoder
Decoder
A decoder is a device which does the reverse operation of an encoder, undoing the encoding so that the original information can be retrieved. The same method used to encode is usually just reversed in order to decode...
s are required in the U.S. by FCC regulation in all 13" (33 cm) diagonal or larger digital televisions. Further, some broadcasters are required by FCC regulations to caption a percentage of their broadcasts.
Packets in CEA-708
Caption streams are transmitted with many packet wrappers around them. These are the picture user data, which contains the caption data, which contains the cc_data, which contains the Caption Channel packets, which contains the Service Block, which contains the caption streams. These packets are described in detail in this section. But the streams themselves are described in the following sections.This layering is based on the OSI Protocol Reference Model:
|
This section will describe the various packets, the Coding Layer and Presentation Layers are described in the remainder of this document.
Picture User Data
The MPEG-2 Stream's picture user data syntax is as follows:| 32 bits | user_data_start_code (lsbf) |
| 32 bits | ATSC_identifier (lsbf) |
| 8 bits | user_data_type_code (msbf) |
| X*8 bits | user_data |
When the picture user data contains a caption user_data_start_code is set to 0x00001B2, the ATSC_identifier is 0x47413934, and the user_data_type_code will be 0x03. If the user_data_type is not 0x03, or 0x06 for bar data, then the packet will be terminated with the bytes 0x0, 0x0, 0x1. If the data describes caption data, it will be as follows:
user_data
| 1 bit (b7) | process_em_data_flag |
| 1 bit (b6) | process_cc_data_flag |
| 1 bit (b5) | additional_data_flag |
| 5 bit (b0-b4) | cc_count (msbf) |
| 8 bits | em_data (meaning not yet defined in ATSC a/53 standard) |
| cc_count*24 bits | cc_data_pkt's |
| 8 bits | marker_bits (all 1's) |
| OPT: Y bits + bytes 0x0,0x0,0x1 | ATSC_reserved_user_data |
Marker bits and reserved bits should all be set by default. If the additional_data_flag is set then the ATSC_reserved_user_data will be at the tail of the packet, terminated by the bytes 0x0,0x0,0x1. If the process_cc_data_flag is set the cc_data_pkt's should be parsed as follows:
At some future time the process_em_data_flag will indicate whether to process the em_data bit string.
cc_data_pkt
| 5 bits (b7-b3) | marker_bits (should be all 1's) |
| 1 bit (b2) | cc_valid |
| 2 bits (b1-b0) | cc_type (lsbf) |
| 8 bits | cc_data_1 (lsbf) |
| 8 bits | cc_data_2 (lsbf) |
If cc_valid is not set the cc_data_pkt's should be considered padding and discarded.. If it is set, cc_type will be one of four values NTSC_CC_FIELD_1 = 0, NTSC_CC_FIELD_2 = 1, DTVCC_PACKET_DATA = 2, DTVCC_PACKET_START = 3. If it is either 0 or 1, the cc_data fields should be interpreted as EIA-608
EIA-608
EIA-608, also known as line 21 captions, used to be the standard for closed captioning for NTSC TV broadcasts in the United States and Canada...
Captions. If cc_type is 3 then a decoder should begin assembling a Caption Channel Packet with the cc_data as described below, and if the cc_type is 2 it should append the cc_data to any Caption Channel Packet being assembled. If a DTVCC packet is already being assembled and either cc_valid is set and the cc_type is 3 or cc_valid is clear and cc_type is 2 or 3, then the packet should be considered complete.
NOTE: In a caption decoder cc_data packets must be reassembled in the correct order to create the DTVCC packets. The standard is not clear on this, but it appears this should be in frame display order, not encoded frame order. This means in encoder DTVCC Packets should probably be broken up and inserted into the picture user data as cc_data packets in display order as well.
NOTE: To avoid this bug in the CEA-708 standard some encoders encode captions only
on one frame type, such as only B frames, or only I frames, since if you only use one frame type the frame display and frame encoded order are the same.
DTVCC packet
| 6 bits (b0-b5) | packet_size |
| 2 bits (b6-b7) | sequence_number |
| ((packet_size*2-1)&0xFF) * 8 bits | packet_data (Service Block) |
Within the packet_data, there is only one type of packet. This is known as the Service Block. This further subdivides the DTVCC Transport Stream into 63 substreams, each of which describes a discrete captioning service. Service 1 is designated as the Primary Caption Service, while Service 2 is the Secondary Language Service. The Caption Descriptor describes any other services offered.
Service Block Packet
| 5 bits (b0-b4) | block_size |
| 3 bits (b5-b7) | service_number |
| opt 1: 6 bits (b0-b5) | extended_service_number |
| opt 1: 2 bits (b6-b7) | null_fill |
| opt 2: block_size * 8 bits | block_data |
Optional Data 1 is only present if service_number is 7, in which case the extended_service_number should be used instead of the service_number. Optional Data 2 is present only if block_size is greater than 0. However if block_size is 0, the service_number must be zero as well, and then this is known as a Null Service Block Header. A Null Service Block Header can be used for padding.
Note: Service Blocks may not cross Caption Channel Packet Boundaries. This means each Caption Channel Packet can be parsed without keeping any state for the Service Blocks themselves.
Caption Stream Encoding
The 63 caption service sub-streams contain a mixed command and text stream, much like TelnetTELNET
Telnet is a network protocol used on the Internet or local area networks to provide a bidirectional interactive text-oriented communications facility using a virtual terminal connection...
. There are four logical code sub-groups: CL, GL, CR, and GR. These each have single and multi-character code sets.
| CL Group: C0 | 0x00-0x1F | Subset of ASCII Control Codes |
| CR Group: C1 | 0x80-0x9F | Caption Control Codes |
| CL Group: C2 | 0x1000-0x101F | Extended Miscellaneous Control Codes |
| CR Group: C3 | 0x1080-0x109F | Extended Control Code Set 2 |
| GL Group: G0 | 0x20-0x7F | Modified version of ANSI X3.4 Printable Character Set (ASCII) |
| GR Group: G1 | 0xA0-0xFF | ISO 8859-1 Latin 1 Characters |
| GL Group: G2 | 0x1020-0x107F | Extended Control Code Set 1 |
| GR Group: G3 | 0x10A0-0x10FF | Future characters and icons |
Whenever a command character is seen any text accumulated in the parser should be flushed. Since text might need to be flushed when there is no command pending, there is a null command known as the ETX command in the C0 command set.
There are also two special commands, the Reset and DelayCancel. These must be parsed with lookahead. A Delay command issued previously can be canceled at any time with a DelayCancel command, so once a Delay is seen a decoder must look ahead for a DelayCancel, and only look for a DelayCancel. A Reset command on the other hand is sent to break out from an unknown decoder state and all data before it must be ignored.
C0 Table
| 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0a | 0x0b | 0x0c | 0x0d | 0x0e | 0x0f | |
| 0x00 | NUL | ETX | BS | FF | CR | HCR | ||||||||||
| 0x10 | EXT1 | P16 |
NUL, BS, FF, and CR are interpreted as they are in ASCII control codes. HCR is
undefined in the CEA-708-A standard. ETX is the NULL command mentioned earlier,
which is used to flush text to the current window when no other command is pending.
EXT1 is used to escape to the 'C2', 'C3', 'G2', and 'G3' tables for the following
byte. Finally, P16 can be used to escape the next two bytes for Chinese and other
large character maps.
It should be noted that all characters in the range 0x10-0x17, which currently
includes EXT1, are followed by one byte which needs to be interpreted differently.
And, all characters in the range 0x18-x1f, which currently includes P16, are followed
by two bytes that need to be interpreted differently. If a decoder encounters one
of these and does not know what to do, it should still skip the next byte or two,
as appropriate, before continuing.
C1 Table
| 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0a | 0x0b | 0x0c | 0x0d | 0x0e | 0x0f | |
| 0x80 | CW0 | CW1 | CW2 | CW3 | CW4 | CW5 | CW6 | CW7 | CLW | DSW | HDW | TGW | DLW | DLY | DLC | RST |
| 0x90 | SPA | SPC | SPL | SWA | DF0 | DF1 | DF2 | DF3 | DF4 | DF5 | DF6 | DF7 |
The C1 Table contains all the currently defined caption commands.
These will be described in detail in the next section.
C2 Table
The C2 Table contains no commands as of CEA-708 revision A. However, if a command is seen in these code sets a decoder must skip an appropriate number of the following bytes.| 0x00-0x07 | +0 bytes | |
| 0x08-0x0f | +1 byte | |
| 0x10-0x17 | +2 bytes | |
| 0x18-0x1f | +3 bytes |
C3 Table
The C3 Table contains no commands as of CEA-708 revision A. However, if a command is seen in these code sets a decoder must skip an appropriate number of the following bytes.| 0x80-0x87 | +4 bytes | |
| 0x88-0x8f | +5 bytes |
G0 Table
| 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0a | 0x0b | 0x0c | 0x0d | 0x0e | 0x0f | |
| 0x20 | SP | ! | " | # | $ | % | & | ' | ( | ) | * | + | , | - | . | / |
| 0x30 | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | : | ; | < | = | > | ? |
| 0x40 | @ | A | B | C | D | E | F | G | H | I | J | K | L | M | N | O |
| 0x50 | P | Q | R | S | T | U | V | W | X | Y | Z | \ | ^ | _ | ||
| 0x60 | ` | a | b | c | d | e | f | g | h | i | j | k | l | m | n | o |
| 0x70 | p | q | r | s | t | u | v | w | x | y | z | { | | | } | ~ | MN |
The G0 Table consists of ASCII characters for the most part. SP here is short hand for
Space, and MN is a musical note, which replaces the Delete command code in ASCII.
G1 Table
| 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0a | 0x0b | 0x0c | 0x0d | 0x0e | 0x0f | |
| 0xa0 | NBS | ¡ | ￠ | ￡ | ¤ | ￥ | ¦ | § | ¨ | © | ª | « | ￢ | - | ® | ￣ |
| 0xb0 | ° | ± | ² | ³ | ´ | µ | ¶ | · | ¸ | ¹ | º | » | ¼ | ½ | ¾ | ¿ |
| 0xc0 | À | Á | Â | Ã | Ä | Å | Æ | Ç | È | É | Ê | Ë | Ì | Í | Î | Ï |
| 0xd0 | Ð | Ñ | Ò | Ó | Ô | Õ | Ö | × | Ø | Ù | Ú | Û | Ü | Ý | Þ | ß |
| 0xe0 | à | á | â | ã | ä | å | æ | ç | è | é | ê | ë | ì | í | î | ï |
| 0xf0 | ð | ñ | ò | ó | ô | õ | ö | ÷ | ø | ù | ú | û | ü | ý | þ | ÿ |
The G1 Table is basically the ISO 8859-1 Latin-1 character set. Note character 0xa0 is the non-breaking space
Non-breaking space
In computer-based text processing and digital typesetting, a non-breaking space or no-break space is a variant of the space character that prevents an automatic line break at its position. In certain formats , it also prevents the “collapsing” of multiple consecutive whitespace characters into a...
, which is to be used to prevent word wrap from separating two words onto separate lines.
G2 Table
| 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0a | 0x0b | 0x0c | 0x0d | 0x0e | 0x0f | |
| 0x20 | TSP | NBTSP | … | Š | Œ | |||||||||||
| 0x30 | BLK | ‘ | ’ | “ | ” | • | ™ | š | œ | SM | Ÿ | |||||
| 0x40 | ||||||||||||||||
| 0x50 | ||||||||||||||||
| 0x60 | ||||||||||||||||
| 0x70 | ⅛ | ⅜ | ⅝ | ⅞ | │ | ┐ | └ | ─ | ┘ | ┌ |
TSP and NBTSP are the Transparent Space, and Non-Breaking Transparent Space, respectively. The G2 Table contains miscellaneous characters that may not be displayed in all browsers. BLK indicates a solid block which fills the entire character block with a solid foreground color.
G3 Table
The G3 Table contains only a single character, the [CC] Icon, with square corners. This chacter is at 0xa0.Caption commands
| bits | Command Name | Parameters | |
|---|---|---|---|
| ETX 0x03 | 8 | EndOfText | |
| CW0–CW7 0x80–0x87 | 8 | SetCurrentWindow0–7 | |
| CLW 0x88 | 16 | ClearWindows | window bitmap |
| DSW 0x89 | 16 | DisplayWindows | window bitmap |
| HDW 0x8A | 16 | HideWindows | window bitmap |
| TGW 0x8B | 16 | ToggleWindows | window bitmap |
| DLW 0x8C | 16 | DeleteWindows | window bitmap |
| DLY 0x8D | 16 | Delay | tenths of seconds |
| DLC 0x8E | 8 | DelayCancel | |
| RST 0x8F | 8 | Reset | |
| SPA 0x90 | 24 | SetPenAttributes | pen size, font, scripting, italics, underline |
| SPC 0x91 | 32 | SetPenColor | foreground color, foreground opacity, background color, background opacity, edge color, edge type |
| SPL 0x92 | 24 | SetPenLocation | row, column |
| SWA 0x97 | 40 | SetWindowAttributes | justify, print direction, scroll direction, word wrap, display effect, effect direction, effect rate. fill color, border color, border type, opacity |
| DF0–DF7 0x98–0x9F | 56 | DefineWindow0–7 | priority, anchor number, anchor vertical, anchor horizontal, row count, column count, locked, visible, centered, style ID |
EndOfText (0x03)
The EndOfText command is a Null Command which can be used to flush any buffered text to the current window. All commands force a flush of any buffered text to the current window, so this command is only needed when no other command is pending.SetCurrentWindow0-7 (0x80-0x87)
SetCurrentWindow tells the caption decoder which window the following commands describe: SetWindowAttributes, SetPenAttributes, SetPenColor, SetPenLocation. If the window specified has not already been created with a DefineWinow command then SetCurrentWindow and the window property commands can be safely ignored.ClearWindows (0x88 + 1 byte)
ClearWindows clears all the windows specified in the 8 bit window bitmap.DisplayWindows (0x89 + 1 byte)
DisplayWindows displays all the windows specified in the 8 bit window bitmap.HideWindows (0x8A + 1 byte)
HideWindows hides all the windows specified in the 8 bit window bitmap.ToggleWindows (0x8B + 1 byte)
ToggleWindows hides all displayed windows, and displays all hidden windows specified in the 8 bit window bitmap.DeleteWindows (0x8C + 1 byte)
DeleteWindows deletes all the windows specified in the 8 bit window bitmap. If the current window, as specified by the last SetCurrentWindow command, is deleted then the current window becomes undefined and the window attribute commands should have no effect until after the next SetCurrentWindow or DefineWindow command.Delay (0x8D + 1 byte)
Delay suspends all processing of the current service, except for DelayCancel and Reset scanning. The period of suspension is set to by the one byte parameter. The paramater specifies the delay in tenths of a second, so the minimum delay is 0.1 seconds, and the maximum delay is 25.5 seconds. A zero second delay can safely be ignored in a decoder, but should not be emitted from an encoder. A delay should be cancelled if the caption decoder's input buffer becomes full, a DelayCancel or Reset is received, or the specified delay time elapses.DelayCancel (0x8E)
DelayCancel terminates any active delay and resumes normal command processing. DelayCancel should be scanned for during a Delay.Reset (0x8F)
Reset deletes all windows, cancels any active delay, and clears the buffer before the Reset command. Reset should be scanned for during a Delay.SetPenAttributes (0x90 + 2 bytes)
The SetPenAttributes command specifies how certain attributes of subsequent characters are to be rendered in the current window, until the next SetPenAttributes command. This command has the following parameters:- pen size, 2 bits, { SMALL=0, STANDARD=1, LARGE=2, ILLEGAL_VAL=3 }
- offset, 2 bits, { SUBSCRIPT=0, NORMAL=1, SUPERSCRIPT=2, ILLEGAL_VAL=3 }
- text tag, 4 bits, { dialog=0, source_or_speaker_id=1, electronically_reproduced_voice=2,
dialog_in_other_language=3, voiceover=4, audible_translation=5,
subtitle_translation=6, voice_quality_description=7, song_lyrics=8,
sound_effect_description=9, musical_score_description=10, oath=11,
undefined_0=12,undefined_1=13,undefined_2=14, invisible=15 }
- font tag, 3 bits, { default=0, monospaced_serif=1, proportional_serif=2, monospaced_sanserif=3,
proportional_sanserif=4, casual=5, cursive=6, smallcaps=7 }
- edge type, 3 bits, { NONE=0, RAISED=1, DEPRESSED=2, UNIFORM=3,
LEFT_DROP_SHADOW=4, RIGHT_DROP_SHADOW=5, ILLEGAL_VAL0=6, ILLEGAL_VAL1=7 }
- underline, 1 bit, { NO=0, YES=1 }
- italic, 1 bit, { NO=0, YES=1 }
SetPenColor (0x91 + 3 bytes)
SetPenColor sets the foreground, background, and edge color for the subsequent characters. Color is specified with 6 bits, 2 for each of blue, green and red. The lowest order bits are for blue, the next two for green and the highest order bits represent red. Opacity is represented by two bits, they represent SOLID=0, FLASH=1, TRANSLUCENT=2, and TRANSPARENT=3. The edge color is the color of the outlined edges of the text, but the outline shares its opacity with the foreground, so the highest order bits of the third paramater byte should both be cleared. The parameters are as follows:- foreground color, 6 bits
- foreground opacity, 2 bits
- background color, 6 bits
- background opacity, 2 bits
- edge color, 6 bits
SetPenLocation (0x92 + 2 bytes)
SetPenLocation sets the location of for the next bit of appended text in the current window. It has two parameters, row and column.If a window is not locked (see Define Window) and the SMALL font is in effect the location can be outside the otherwise valid addresses.
- row, 4 bits, normally 0-14
- null padding, 4 bits
- column, 6 bits, normally 0-31 for 4:3 formats, and 0-41 for 16:9 formats
- null padding, 2 bits
SetWindowAttributes (0x97 + 4 bytes)
SetWindowAttributes Sets the window attributes of the current window.Fill Color is specified with 6 bits, 2 for each of blue, green and red. The lowest order bits are for blue, the next two for green and the highest order bits represent red. Fill Opacity is represented by two bits, they represent SOLID=0, FLASH=1, TRANSLUCENT=2, and TRANSPARENT=3. The window's Border Color is specified the same way. However, the Border Type is split into two fields. They should be combined, with border type 01 representing the low order bits, and border type 2 the high order bit. Once combined the Border Type has 6 valid values: NONE=0, RAISED=1, DEPRESSED=2, UNIFORM=3, SHADOW_LEFT=4, and SHADOW_RIGHT=5.
- fill color, 6 bits. Window interior color.
- fill opacity, 2 bits. { SOLID=0, FLASH=1, TRANSLUCENT=2, and TRANSPARENT=3 }
- border color, 6 bits. Window border color.
- border type 01, 2 bits. See discussion above.
- justify, 2 bits. For Left-to-Right and Right-to-Left print directions the values are: {LEFT=0, RIGHT=1, CENTER=2, FULL=3}, for Top-to-Bottom and Bottom-to-Top print directions the values are: {TOP=0, BOTTOM=1, CENTER=2, FULL=3
For Left justification, decoders should display any portion of a received row of text when it is received. For center, right, and full justification, decoders may display any portion of a received row of text when it is received, or may delay display of a received row of text until reception of a row completion indicator. A row completion indicator is defined as receipt of a CR, ETX or any other command, except SetPenColor, SetPenAttributes, or SetPenLocation where the pen relocation is within the same row.
Receipt of a character for a displayed row which already contains text with center, right or full justification will cause the row to be cleared prior to the display of the newly received character and any subsequent characters. Receipt of a justification command which changes the last received justification for a given window will cause the window to be cleared.
- scroll direction, 2 bits. This specifies which direction text will scroll when the end of a caption "line" is reached. It has one of four values: LEFT_TO_RIGHT=0, RIGHT_TO_LEFT=1, TOP_TO_BOTTOM=2, and BOTTOM_TO_TOP=3.
- print direction, 2 bits. This specifies how order text is added to a window. It has one of four values: LEFT_TO_RIGHT=0, RIGHT_TO_LEFT=1, TOP_TO_BOTTOM=2, and BOTTOM_TO_TOP=3.
- word wrap, 1 bit. If set word wrapping is enabled, otherwise word wrap should not be employed.
- border type 2, 1 bits. See discussion above.
- display effect, 2 bits. This specifies an effect to be used to display or hide a window. It has one of three valid values: SNAP=0, FADE=1, and WIPE=2. SNAP means the window should assume full opacity immediately. FADE means the window should fade in or out at effect speed. Finally, WIPE means the window should fly onto or off the screen from the border of the screen border specified in effect direction at the rate specified in effect speed
- effect direction, 2 bits. This specifies where a wipe effect comes from on window display. It has one of four values: LEFT_TO_RIGHT=0, RIGHT_TO_LEFT=1, TOP_TO_BOTTOM=2, and BOTTOM_TO_TOP=3. When the window is wiped off the screen it should be wiped off in the opposite direction from how it was wiped onto the screen.
- effect speed, 4 bits. This specifies in half-seconds how long a caption display or hide effect, such as FADE, and WIPE, should take. The maximum time is 7.5 seconds, and the minimum non-zero value is 0.5 seconds.
Colors, text painting, effects, and border types can be customized with the SetWindowAttributes and SetPenAttributes commands. However, the caption provider may wish to use predefined standard window styles. A set of predefined styles will be hard stored in receivers. This set will anticipate the most widely used types of caption windows in order to conserve caption channel bandwidth by eliminating the need to transmit superfluous SetWindowAttributes and SetPenAttributes commands.
Predefined window and pen styles can be specified by the window style and pen style ID parameters in the DefineWindow command.
DefineWindow07 (0x98-0x9F, + 6 bytes)
DefineWindow0-7 creates one of the eight windows used by a caption decoder. This command should be sent periodically by a caption encoder even for pre-existing windows so that a newly tuned in caption decoder can begin displaying captions. When issued on a pre-existing window the pen style and window style can be left null, this tells the decoder not to change the current styles if the exist, and initialize both to style 1 if the window does not exist in its context.The parameters are as follows.
- priority, 3 bits, 0-7. A decoder is only required to display up to four windows. If more than four displayed windows are requested, the decoder should display the four highest priority windows.
- column lock, 1 bit. If set column lock fixes the absolute number of columns to be displayed. If not set, a caption decoder may display more columns of text when the font size permits it, and a SetPenLocation command may go to a location outside the defined window size.
- row lock, 1 bit. If set row lock fixes the absolute number of rows to be displayed. If not set, a caption decoder may display more rows of text when the font size permits it, and a SetPenLocation command may go to a location outside the defined window size.
- visible, 1 bit, if set this flag causes the window to be displayed upon creation, if not it is initially hidden.
- null, 2 bits. Null padding.
- anchor vertical, 7 bits. Vertical position of the window's anchor point. The range is normally 0-74. When the relative positioning bit is set however the range is 0-99.
- relative positioning, 1 bit, if set the anchor horizontal and anchor vertical represent relative coordinates, percentages, instead of regular coordinates
- anchor horizontal, 8 bits. Horizontal position of the window's anchor point. The range is normally 0-209 when the stream's aspect ratio is 16:9, and 0-159 when the stream's aspect ratio is 4:3. When the relative positioning bit is set however the range is 0-99.
- row count, 4 bits. This is the number of rows of text, assuming the STANDARD font size, the window will hold. The range is 0-31 for 4:3 streams, and 0-41 for 16:9 streams. NOTE: In practice a decoder must add one to the number to get the indended effect. i.e. 0 -> 1, 1 -> 2, etc.
- anchor ID, 4 bits. Valid Values: { UPPER_LEFT=0, UPPER_CENTER=1, UPPER_RIGHT=2, MIDDLE_LEFT=3, MIDDLE_CENTER=4, MIDDLE_RIGHT=5, LOWER_LEFT=6, LOWER_CENTER=7, LOWER_RIGHT=8 }
- column count, 6 bits. This is the number of columns of text, assuming the STANDARD font size, the window will hold. The range is 0-31 for 4:3 streams, and 0-41 for 16:9 streams. NOTE: In practice a decoder must add one to the number to get the indended effect. i.e. 0 -> 1, 1 -> 2, etc.
- null, 2 bits. Null padding.
- pen style, 3 bits. If the value is zero this is a new window, pen style one should be used for future characters.
If the value is zero and this is an existing window the previous pen should continue to be used. For non-zero values the pen style should be set as if SetPenStyle were called with the parameters in the pen style table below.
- window style, 3 bits. If the value is zero this is a new window, window style one should be used for future characters.
If the value is zero and this is an existing window the previous window style should continue to be used. For non-zero values the window style should be set as if SetWindowStyle were called with the parameters in the window style table below.
- null, 2. Null padding.
Predefined Pen style
| Pen Style | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|
| Font Size | Standard | Standard | Standard | Standard | Standard | Standard | Standard |
| Font Style | "Default" | Monospaced Serif | Proportional Serif | Monospaced sans Serif | Proportional sans Serif | Monospaced sans Serif | Proportional sans Serif |
| Offset | Normal | Normal | Normal | Normal | Normal | Normal | Normal |
| Italics | No | No | No | No | No | No | No |
| Underline | No | No | No | No | No | No | No |
| Edge Type | None | None | None | None | None | None | None |
| Foreground Color | White | White | White | White | White | White | White |
| Foreground Opacity | Solid | Solid | Solid | Solid | Solid | Solid | Solid |
| Background Color | Black | Black | Black | Black | Black | Black | Black |
| Background Opacity | Solid | Solid | Solid | Solid | Solid | Transparent | Transparent |
| Edge Color | Black | Black | Black | Black | Black | Black | Black |
Predefined Window style
| Window Style | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|
| Justify | Left | Left | Center | Left | Left | Center | Left |
| Print Direction | Left-to-Right | Left-to-Right | Left-to-Right | Left-to-Right | Left-to-Right | Left-to-Right | Top-to-Bottom |
| Scroll Direction | Bottom-to-Top | Bottom-to-Top | Bottom-to-Top | Bottom-to-Top | Bottom-to-Top | Bottom-to-Top | Right-to-Left |
| Word Wrap | No | No | No | Yes | Yes | Yes | No |
| Display Effect | Snap | Snap | Snap | Snap | Snap | Snap | Snap |
| Effect Direction | Null | Null | Null | Null | Null | Null | Null |
| Effect Speed | Null | Null | Null | Null | Null | Null | Null |
| Fill Color | Black | Black | Black | Black | Black | Black | Black |
| Fill Opacity | Solid | Transparent | Solid | Solid | Transparent | Solid | Solid |
| Border Type | None | None | None | None | None | None | None |
| Description | NTSC Style PopUp | PopUp w/Transparent Background | NTSC Style PopUp Centered | NTSC Style RollUp | RollUp w/Transparent Background | NTSC Style Centered RollUp | Ticker Tape |
Word wrap
It may sometimes be desired that word wrap be performed in a caption decoder. This may happen because the end user of the caption decoder specifies a different font than the encoder requests, or the end user wishes to see more of the caption text than normally possible. Note that SetWindowAttributes sets a word wrap flag, when set this indicates the subtitles are written with word wrap in mind, and this may be used as a hint to the decoder that word wrapping is safe. Word wrap can be performed on carriage return, space, and hyphen characters, however both the non-breaking space (0xA0 in the G1 Table), and the non-breaking transparent space (0x21 in the G2 Table) should not be considered safe characters to rewrite.Anchor ID
There are nine valid anchor ID's, shown below:These are used to tell the caption decoder how to expand the text box as text is added to a caption window. A window is assigned an anchor point, or location and an anchor ID. If the anchor point is say 0,0, and the anchor ID is 0, then the window will expand down and right from the upper left corner of the caption area. If the anchor point is 50%,50% and the anchor ID is 4 the window will expand equally in all directions from the center of the caption area.
Fonts
CEA-708 supports eight font tags: undefined, monospaced serif, proportional serif, monospaced sans serif, proportional sans serif, casual, cursive, small capitals. The first is not defined and should probably be avoided. However these fonts are implemented it should be possible to underline them, and italicize them. Bold versions are not needed, but it should be possible to draw the outline of each letter in a different color and opacity than the fill. Finally, these fonts must allow superscripts, subscripts, and be able to support Latin-1 plus the additional symbols in CEA-708, such as the [CC] symbol and the dozen or so UnicodeUnicode
Unicode is a computing industry standard for the consistent encoding, representation and handling of text expressed in most of the world's writing systems...
characters in this standard. Below are some font examples, for more see the Wikipedia Fonts article.
| Proportional Serif | |
|---|---|
 | |
| Proportional Sans Serif | |
Windows
The window addressable area should always be within the Safe-Title area, so that all addressable locations are within the display window if the monitor overscans the image onto a non-rectangular screen. If the video stream has a 16:9 aspect ratio the addresses should be in the range 0..74 for the vertical addresses, and 0..209 for the horizontal addresses. If the video stream has a 4:3 aspect ratio the addresses should be in the range 0..74 for the vertical addresses, and 0..159 for the horizontal addresses. For other aspect ratios relative addressing should be used and both vertical and horizontal addresses should be in the range 0..99%.The window size should be scaled based on the font size. With this in mind, rows longer than 32 characters are discouraged even on 16:9 ratio screen so that larger than specified fonts may be selected by the user.
Row and column locking
Row and column locking feature is supported in CEA-708-B standard but the later version CEA-708-C it has been assumed that both rows and columns are locked. The basic functionality is like belowTotally four combinations are provided
1)Row /Column locked
2)Row Unlocked Column locked
3)Row locked and Column unlocked
4)Row unlocked and column unlocked
1)Row /Column locked: If both rows and columns are locked then window size in terms of columns and rows can’t be extended.
For a window if number of rows and columns are defined like 3 and 10
Then the text ‘ROWS AND COLUMNS ARE NOT LOCKED FOR EVER AND EVER AND EVER ‘which comes in the 0 row looks like below (assume that word wrapping is disabled)
1. ROWS AND C
2.
3.
Since both are locked, text cont be extended beyond 10 columns and also row cont be extended beyond 0.
2. Row Unlocked /Column locked: In this case window can be extended up to max row given in the window define command.
The same above text will look like below
1. ROWS AND C
2. OLUMNS ARE
3. NOT LOCKED
Row is unlocked so text can be extended up to max rows of a define command.
3. Row locked and Column unlocked: In this case window can be extended up to max number columns.
As per the CEA-708 standard Max number of columns for any window is 32.
The same above text then look like below
1. ROWS AND COLUMNS ARE NOT LOCKED
2.
3.
Column is unlocked so text can be extended up to max columns.
4.Row unlocked and column unlocked: In this case window can extended in terms of rows are columns.
The same above text then look like below
1. ROWS AND COLUMNS ARE NOT LOCKED
2. FOR EVER AND EVER AND EVER
Since both are unlocked so the text can extended up to 32 columns and as well as total rows.
Implementation notes
- The minimum buffer size for each of the 63 possible services (Service Input Buffers) is 128 bytes.
- In a caption decoder the DelayCancel and Reset commands should be interpreted outside the buffering mechanism. It should be safe to
just for the 0x8E and 0x8F codes.
- In a caption encoder the 0x8E and 0x8F commands should encoded in a parameter to another command. Commands can be split into several subcommands to avoid this problem.
- The closed caption icon in the G3 code set must not be rendered with rounded corners in a WTO country, due to trademark licensing problems.
External links
- Critique of CEA-708 caption fonts