Yepp... this should fix it:
@fsphil The VBI sequences seems to be wrong. each byte of a line is equal to the XOR of the same bytes of each of the
4 other lines of the same group ( 49, 02, 15, 5E, 73 ) or ( A1, EA, FD, B6, 9B )
This is what hacktv produces:
15 72 00 10 |0E| 4C AB 9A |22| 44 8E AD 00
73 8C 2F 54 |35| 67 73 86 |F0| 9E C5 99 00
5E 4D 4D 59 |45| 4D 4D 44 |55| 4D 4D 59 45
49 FE 28 B1 |00| 00 00 00 |00| 00 00 00 00
02 4D 4D 59 |45| 4D 4D 44 |55| 4D 4D 59 45
first line from top to bottom:
0e xor 35 xor 45 xor 00 !=45
next line:
22 xor F0 xor 55 xor 00 !=55
So this should be right this way: (see Syster VBI paper from zip)
15 7b 02 35 |86| 47 f9 d6 |dc| 0c 7b fb ...
02 4D 4D 59 |45| 4d 4d 44 |55| 4d 4d 59...
49 7F 08 B1 |00| 00 00 00 |00| 00 00 00 ...
5E FF 21 01 |00| F6 FE F0 |E3| F0 A1 00 ...
73 B6 66 DC |C3| FC 4A 62 |6A| 41 25 52 ...
86 xor 45 xor 00 xor 00 == C3
or:
dc xor 55 xor 00 xor e3 == 6A
73 can be random data, so the only thing we have to calculate are the data for 5E... the other lines are fixed patterns... this is for error correction and seems to confuse some decoders.