[EMAIL PROTECTED] mused: > > In a message dated 3/12/2005 2:01:42 PM Pacific Standard Time, > [EMAIL PROTECTED] writes: > Oh, good. I really didn't want to have more RAW stuff to think about. My > head already hurts. Good thing Bayer makes aspirin as well as matrixes. I > can cure my headache while at the same time getting one. <LOL> > > Actually, I'm and impressed that you guys know so much technical stuff. > > Shel > ======= > Ditto. And Godfrey's explanation was very impressive, although I am not sure > I understand it all yet, but I am getting the gist. > > Got to get that book and read up. > > What I am wondering is, is there *any* compression in a RAW file (let's say > as generic a RAW file as possible)? Or is it just data as recorded? (I am > sure > I could ask that more techno-geeky way but that is what I am wondering about > without knowing the buzz words yet.)
It's possible. Your dRebel, for example, compresses RAW files. Adobe's new DNG format uses compression, as does Pentax (for the DS). The compression schemes they use are, in theory, lossless. But there is a rumour that Canon slightly tweak the raw data to make it compress better. An example (to help explain things): If you've got a large area of blue sky, you can just say "the next 100 pixels are this colour", rather than storing the same value 100 times. That can be done using less space than would be needed for the values. Lo and behold - you've got a compressed RAW file! That's a true lossless compression scheme. Now imagine that one of those 100 pixels is almost, but not quite, the same colour. That's going to need more space to describe. Still not as much as storing all the original values, but more that the compressed version. If you ignore that different pixel, and pretend it's the same colour as the other 99, you don't need to use the extra space. In other words you've given up a small amount of the original data in trade for an even smaller file.
