scott kirkpatrick
Member
There's a heated debate over on a Leica forum between several posters who are proving just how dangerous a little knowledge can be. Here's the problem: The Leica DMR produced 20 MB raw files from its 10 MP Kodak imager, which is essentially the same silicon that will be used in the M8. But the M8 specs appear to say that its RAW files will be a constant size, 10MB. People poking at the handful of bootleg M8 .DNG files that are around are saying that each pixel is in fact coded as an 8-bit number, although there are plausible reasons to suspect bugs, misinterpretations, etc... And it would seem like a dangerous thing to do, even if the bits are perceptually encoded to represent, say, the logarithm of the linear signal off the CCD, because of the fragile nature of 8-bit images when you start to postprocess them.
I have trouble imagining a loss-less encoding done for the M8 which would not
(a) produce files of varying size
(b) possibly end up with less than one Byte of data per pixel, and certainly not line up by the byte.
But can some of the Canon/Nikon owners let us know how big their RAW files are (expressed as the ratio of Bytes per Pixel). And are they constant size? (I know the answer is NO, and .85 bytes per pixel for .CR2 files from a friend's 20D.)
scott
I have trouble imagining a loss-less encoding done for the M8 which would not
(a) produce files of varying size
(b) possibly end up with less than one Byte of data per pixel, and certainly not line up by the byte.
But can some of the Canon/Nikon owners let us know how big their RAW files are (expressed as the ratio of Bytes per Pixel). And are they constant size? (I know the answer is NO, and .85 bytes per pixel for .CR2 files from a friend's 20D.)
scott