mogul wrote:
This may be a dumb question, since jpeg is so important to the Canon\'s file, could this be affecting their limited DR?
No. The compression algorithm is lossless. That means you get back exactly the same data, bit for bit when you decompress the file.
Lossless JPEG works like this. First you compress the image using the regular lossy JPEG algorithm (using the highest quality settings). As this is a RAW file there is no \"image\", but you can compress all the red pixels, all the blue pixels, all the odd green pixels and all the even green pixels as four separate \"images\".
The resulting JPEG will be smaller than the original, but it will be slightly different, because JPEG is lossy. So the next step is to compare the JPEG to the original and calculate the difference, pixel by pixel. Now as the JPEG and original will be similar these differences will be small integers, hence mostly zeros in binary, so the difference values will compress well using a standard lossless compression algorithm (just like zipping a text file).
Since we have the JPEG, and a copy of the difference between the JPEG and the original, we can reconstruct the exact original data.
Jun 12, 2015 at 10:17 AM
Previous versions of Matt Grum's message #13056542 « Announced: Sony A7R II »
