Chris Rossi

Well 2 days ago my copy of the great series “The IT Crowd series 2″ arrived on DVD from the UK (Region 2 DVD).

Series 1 had a whole bunch of cool geeky features, however series 2 has gone above and beyond this time.

When you select “l33t” subtitles and watch episode 1, you see this:

The subtitles are actually a base64 encoded file! This is the same format used for attachments in emails.

Hidden Files
Episode 1 – “EasterEgg1.jpg”
Episode 2 – “Barber.z5″
Episode 3 – “EggDump”
Episode 4 – Not a file, but some programming code (see below)
Episode 5 – “EasterEgg3.part.2″
Episode 6 – Not a file, it is some kind of encoded light pattern in the bottom of the screen.

Programming Code (Episode 4)
Episode 4 contains the following programming code:

10	REM > Sophie's Choice

20	MODE 0

30	*LOAD EggDump ffff3000

40	ENVELOPE 1, 3, 1, -1, 1, 1, 2, 1, 30, -10, 0, -40, 127, 80

50	PROCPLAY2

60	END

70	DEF PROCPLAY2

80	RESTORE 290

90	REPEAT

100	A = 1

110	READ C, N$, D

120	P = 37 + INSTR("A BC D EF G", LEFT$(N$,1)) * 4

130	N2$ = MID$(N$, 2, 1)

140	IF N2$ = "#" THEN P = P + 4

Etc…
It appears to be in BASIC language, a game called “Sophie’s Choice”.

The JPEG image in episode 1 will probably reveal more though.

The main problem with DVD Subtitles is that they are not text, they are images. The only way to convert them to plain text is to use Optical Character Recognition (OCR). The best program I’ve seen to do this is SubRip, however this unfortunately truncates the subtitles and only translates the first 8 or so lines.
Currently I’m using a program called Avidemux, it’s the only one I’ve found (out of about ten) that works. The drawback is that although it’s accurate, it also takes much longer to train the OCR.

I’ve written a base64 decoder with PHP, so far I’ve only extracted enough of the subtitles to know that the JPEG photo was taken with an Olympus Digital Camera E-10 at 11:15pm on May 4th 2007.

Update: Part 2 Continues here