Originally posted by Clarkey My understanding of this is a little limited, but with respect to s-raw, it is my understanding that the number of either photo sites sampled or pixels taken is reduced.
There is no official information being published.
The following is some insider statement from some forum:
Quote: sRaw is the original data resampled and redistributed in such a way that the resulting sRAW, with 1/4 the pixels of the RAW, have 1/2 as many samples as the original RAW, but with two of them in each pixel such that there are two types of pixels, green+red and green+blue, 30 bits each (I don't recall if it is 15+15 or not), in a checkerboard pattern.
They still need a form of demosaicing for the red and blue channels.
The problem is this:
The original RAW file contains monochrome pixels according to the Bayer pattern. They cannot simply be resampled to a smaller size (as it would be a gray image).
Therefore, there a two RAW formats which may contain a resampled image:
1. Linear DNG: the Bayer pattern is demosaiced and the resulting color values are losslessly compressed and stored in full bit depth.
The size of 1/4 #pixels Linear DNG roughly is 3/4 of the original RAW.
2. sRaw (1/4 #pixels) and mRaw (1/2 #pixels).
The size of sRaw is about 1/2 of the original RAW (and mRaw about 3/4). This slightly better compression capability comes at the prize of a remaining 2 color checkboard demosaicing (cf. quote above) which only Canon software (if at all) may be able to do properly. sRaw can lead to 1/4 size images which aren't as sharp as full size demosaiced images, then downsampled.
Personally, I couldn't care less about sRaw (being dead end special solutions) and look forward to downsampled (linear) DNG from cameras sporting 30 MPixels and more.
Another solution for lossy compression at higher bitdepths will be the forthcoming "JPEG XR" file format which will make the RAW file format redundant for many applications which now need RAW, like using fill color in post processing.