[4candles]

I've now compared the "VIDEO_TS.IFO" file from the DVD burned using toast by GSMYTHE (thanks to Graham for a copy of the file) with the original torrented on this site.

There is only one byte in the IFO file which is different between the two versions - but this is obviously enough to change the MD5 checksums.

The difference is in the VMG_PTT_SRPT (Video Manager Title Play Map Table) and is the value of the "Title set starting sector" for title 1.

What that means is that there is absolutely no difference in the content of the two DVDs, just that Toast is positioning the files on the DVD in a different way to that predicted by the DVD authoring program (or by the first program to burn those files to DVD). Because of that, it has needed to adjust one of the sector pointers in the VIDEO_TS.IFO file for hardware players to still play it properly.

You could compare this problem with MD5 checksums of WAV files - two WAV files could contain slightly different headers (and therefore have different MD5 checksums), but the PCM data inside the WAV file (the "FLAC fingerprint" or "shntool MD5") could be identical.

So the solution would be to calculate MD5 checksums of just the important data inside the IFO and VOB files - but that's not an easy thing to do, especially as the DVD specifications are not published.