right, it's vital to know accurately which byte the header ends so that you know which byte the encrypted volume begins. using a graphical hex editor could work, e.g. perhaps there is a byte sequence at the end of the header that's always the same.
if not, knowing exactly the length of a key slot, exactly how many key slots your header has, and the exact length of the data before them is very important.
an extra trick I thought of: I expect that the header for a disk partition is smaller than a basic disk encryption header. Instead of replacing the encryption header entirely with random data, why not:
- find out the size of a partition header
- subtract it from the size of your encryption header
- overwrite the start of the disk encryption header with a partition
- fill the remaining space with random data (only as far as the last byte of the encryption header! use the number you found in step 2)
then, instead of having a "suspicious" random data disk, you have a disk that an OS filesystem window would recognize when you plug it in. Sure the rest of the data is your encrypted volume, but it "looks" otherwise like a normal disk that's got nothing on it.
you: "really? nothing on it? damn, must've taken the wrong disk with me, my bad"
or
you: "yeah, that's my disk to put the holiday photos on, would you like to see my photos of the church organs of Europe?"
that sort of thing
Someone with forensic tools would just be able to see a relevant chunk of random data irrespective of how your headers look. Sure having a reasonable looking header increases your chances of plausible deniability but I don't see how tricks anyone that has been instructed in this field. Also playing around with the header is quite dangerous if you screw up in the process. Id rather never put myself on a situation where someone gets to my encrypted data to begin with. So far I have concluded that you just cannot cross any borders with encrypted stuff.