sensor data - raw vs linear tiff

Hi

I have a question on the difference between the representation of data in RAW and Linear decoded TIFF. Clearly RAW is not demosiaced and has separate pixels for R G and B while in a decoded TIFF they are combined.

That issue aside, if I use a tool such as dcraw with options -4 -T, I thought that essentially it created a demosiaced file and that this was the only difference? Meaning that there should be little differences in quantization of light levels in a specific "pixel" between the two.

I ask this because when trying to load a single TIFF into Photomatix VS using the CRW file from my camera I get substantially different results.

I asked people at photomatix about this but found their replys to be typical of support which does not read the question and assumes you don't know what you're asking.

Doesn't the rest of the RAW information kick-in at some stage?

It's not just the demosaicing that happens while converting raw to a displayable image, as far as I know.

There is also the various algorithms used for such - ufraw has, like, 4 or 5 of them, with distinct advantages for each.

Then you got the actual conversion curve. Each sensor maker may have a far from linear light response curve, which means a correction might have to be applied otherwise you get the equivalent of a weird monitor gamma.

And then of course you get the colour temp information to be considered. That's another curve modifying factor.

None of those apply to the tiff case, where the image is simply stored in an already "decoded", visible fashion.

Did I completely miss your concern or does the above rambling make any sense?