Cavena 890
The binary subtitle files of Cavena's systems, from Stockholm, which Nordic broadcasters exchange. A header about the programme, then a 128-byte block for each subtitle: its frames in and out, its alignment and two lines of text.
At a glance
| Extension | .890 |
|---|---|
| Media type | None registered |
| Specification | None published. Subtitld follows Subtitle Edit, whose reader and writer were worked out from broadcasters' files. |
| Time precision | A frame, counted from the programme's start or from 0. The file does not say at what rate. |
| Text encoding | A table chosen by the header's language: Latin, with accents before their letter; Russian, Greek, Arabic, Hebrew a byte a letter; Chinese in UTF-16 |
| Styling | Italics, seven colours and white, and a box behind the text |
| Positioning | Left, centre or right. No place up or down is known. |
| Speakers | None |
| In containers | None. A file delivered with the programme. |
Anatomy of a file
Not text but bytes: a 388-byte header, then a 128-byte block for each subtitle.
Header · 388 bytes Tears of Steel · English HLV23N.V · 10:00:00:00 Block · 128 bytes · number 16 frame 900,575 to 900,613 centred · You're a jerk, Thom. Block · 128 bytes · number 32 …
- 1 · Header
The programme's title, translated and original; the translator; the language of each line; the font; how many subtitles; and the programme's start.
- 2 · Timing
The block's number, then its first and last frame, counted from 00:00:00:00.
- 3 · Text
Its alignment, then two lines of 51 bytes. A subtitle of one line has it in the second, the bottom one.
The header
What is known of it, by where it is:
| Byte | What it holds |
|---|---|
| 40 | The translated programme title, 28 characters |
| 68 | The translator |
| 146, 147 | The language of the first line and of the second: 09 English, 07 Danish, 28 Swedish, 1E Norwegian, 56 Russian, 70 Greek, 80 Arabic, 8F Hebrew, 90 and 91 Chinese |
| 179 | How many subtitles, low byte first |
| 187 | The font: HLV23N.V for Latin, KYRIL4.V, GREEK4.V, ARA19N.V, HEBNOA.V, CCKM44.V for the others |
| 218 | The original programme title |
| 256 | The programme's start, as text: 10:00:00:00 |
The rest is written as Subtitle Edit found it in files, its meaning not known.
A block, byte by byte
The first subtitle above, as Subtitld writes it, from a programme start of 10:00:00:00 at 25 frames a second:
| Byte | Bytes | What they are |
|---|---|---|
| 0 | 00 10 |
Its number, 16. Blocks that share a number are one subtitle. |
| 2 | 0D BD DF |
In Frame 900,575: 10:00:23:00 |
| 5 | 0D BE 05 |
Out Frame 900,613: 10:00:24:13 |
| 8 | 54 |
Centred: 50 left, 52 right, 54 centre |
| 16 | 7F 7F … |
The first line, empty: 7F is unused room |
| 70 | 00 |
A0 puts a box behind the text |
| 73 | 59 6F 75 … |
You're a jerk, Thom. |
| 124 | 00 00 00 00 |
The end of the block, which the last block may leave off |
In the text, 88 and 98 turn italics on and off. F1 to F6 and F8 colour the text red, green, yellow, blue, magenta, cyan or orange; F7 makes it white again.
The timecode
Times are counts of frames, not timecodes: 10:00:23:00 is frame 900,575 at 25 a second. The file does not say the rate. Nordic television is 25 frames a second, and Subtitld counts at the video's rate, else 25.
Top subtitles go to the bottom
No place up or down. No one has found where a Cavena 890 says a subtitle sits higher on the picture. Subtitld writes every subtitle at the bottom, keeping only left, centre or right: a subtitle at the top or in the middle comes back at the bottom.
Cavena 890 or EBU STL?
| Cavena 890 | EBU STL | |
|---|---|---|
| Specification | None published | EBU Tech 3264 |
| Header | 388 bytes | 1,024 bytes |
| A subtitle | 128-byte blocks, two lines each | 128-byte blocks, 112 bytes of text |
| Times | Frame counts | Timecodes |
| Frame rate | Not in the file | In the header |
| Up or down | Not known | A row |
Cavena 890 in Subtitld
- Every subtitle's text and line breaks
- Italics, in Latin, Hebrew and Chinese
- Colours, except in Chinese: the nearest of red, green, yellow, blue, magenta, cyan, orange and white
- A box behind the text
- Left, centre or right
- The language, and the programme's title
- Start and end, to the frame
- Places at the top or in the middle, written at the bottom
- Bold and underline, which have no code known
- Italics in Greek, Russian and Arabic, which Subtitld writes without
- Characters outside the table, which become their letter without its accent, or "?"
- Speakers, translations beside the original
- Dubs, speakers' faces and colours
Save a .usfx project to keep them.
Opening a Cavena 890 file
Subtitld reads:
- the table from the language bytes and the fonts: Latin, Russian, Greek, Arabic, Hebrew or Chinese;
- blocks that share a number as one subtitle, its lines in order;
- italics, colours and the box;
- the Latin letters and digits in Hebrew lines, which the files keep back to front, in reading order;
- times from the programme's start in the header, when every subtitle comes after it; else from 0;
- frames at the video's rate, else 25.
The header's language becomes the project's, and its title the project's title. The frame rate and the programme start are kept, to save the file again as it was. The format has nothing to recognise it by, so any .890 long enough to hold a subtitle is read as one.
Exporting Cavena 890
Export, then Subtitles and CAVENA890, offers:
- TextThe original, a translation, or both, the translation below.
- Speaker namesLeft out, or before the text (Thom: …).
- FormattingKept, or removed.
- Shift timesEvery subtitle earlier or later by the same amount.
- Frame rateThe video's, unless you choose. The file does not say it.
- Programme starts at00:00:00:00, or 10:00:00:00: where the frames count from.
The table is the language's, or the one the text's letters need. The title is the project's, else the video's name. The header's programme start is 10:00:00:00 either way, as Subtitle Edit writes it; times counted from 0 come before it, and readers take them as they are. A subtitle of more than two lines takes more blocks under one number, and a line longer than 51 bytes is broken at a space.
Not checked against Cavena's tools. What Subtitld writes follows Subtitle Edit's reader and writer. No file it wrote has been opened in Cavena's own software.
Trivia
The other name the 890 format goes by in subtitling software that reads it.
Edgeware, of Stockholm, bought Cavena Image Products. The purchase closed in February 2019.
Bytes of text a line, two lines a block. A Chinese line holds 25 characters, at two bytes each.
Related formats
Open it in Subtitld with the video and retime it on the waveform.