I usually shoot Canon, but this last month I’ve been shooting exclusively with the D600 both recreationally and professionally. For the most part, I have been loving it, but I recently noticed something especially peculiar….
The Canon 6D has a 20 MP sensor
The Nikon D600 has a 24 MP sensor
One would expect that file sizes of the D600 would be about 20-25% larger than the 6Ds based on sensor size… at least ball park right?
Wrong…
Truth of the matter is, the Canon 6D’s JPEG’s are in the 3-5 MB size range. The Nikon D600’s seem to be falling in the 8-12 MP range, over 140% larger in file size. The RAW files seemed closer, in the 20MB range for the Canon, with closer to the 30 MB range for the D600, still 50% larger is significant.
I am curious if it also has something to do with the fact that the D600 has better Dynamic Range (14.2 vs 12.1 EV) and Color Gamut (25.1vs 23.8) than the 6D, which results in a wider range of colors and shades, it would explain the much larger file size as this would have to be accounted for on a pixel level after compression for JPEG files.. 20-30% larger size would be understandable….100-140% larger? There are going to be workflow issues because it is multiplied at every step, from downloading off the card, to opening, editing exporting, saving and sharing, it really adds up.
Ill need to do some more tests when I get back to Maui, but this is a big deal, especially if you are a wedding photographer who takes 4,000-5000 images per wedding.
If you own either of these cameras, you might want to check out:
Canon 6D Crash Course Training Tutorial Download Video
Nikon D600 Crash Course Training Tutorial Download Video I usually shoot Canon, but this last month I’ve been shooting exclusively with the D600 both recreationally and professionally. For the most part, I have been loving it, but I recently noticed something especially peculiar….
The Canon 6D has a 20 MP sensor
The Nikon D600 has a 24 MP sensor
One would expect that file sizes of the D600 would be about 20-25% larger than the 6Ds based on sensor size… at least ball park right?
Wrong…
Truth of the matter is, the Canon 6D’s JPEG’s are in the 3-5 MB size range. The Nikon D600’s seem to be falling in the 8-12 MP range, over 140% larger in file size. The RAW files seemed closer, in the 20MB range for the Canon, with closer to the 30 MB range for the D600, still 50% larger is significant.
I am curious if it also has something to do with the fact that the D600 has better Dynamic Range (14.2 vs 12.1 EV) and Color Gamut (25.1vs 23.8) than the 6D, which results in a wider range of colors and shades, it would explain the much larger file size as this would have to be accounted for on a pixel level after compression for JPEG files.. 20-30% larger size would be understandable….100-140% larger? There are going to be workflow issues because it is multiplied at every step, from downloading off the card, to opening, editing exporting, saving and sharing, it really adds up.
Ill need to do some more tests when I get back to Maui, but this is a big deal, especially if you are a wedding photographer who takes 4,000-5000 images per wedding.
If you own either of these cameras, you might want to check out:
Canon 6D Crash Course Training Tutorial Download Video
Nikon D600 Crash Course Training Tutorial Download Video
Michael,
JPEG is a well defined standard. It works with images with 24 bits per pixel (bpp), i.e. 8 bits per color component, so that it doesn’t matter what the camera gathers, as it must transform that into a 24bpp image. JPEG then uses a 4 step compression algorithm, as follows:
– The first step splits the image in several 8×8 pixel blocs;
– The second step gets each block to go through a mathematical transform called Discrete Cosine Transform (DCT). The result is the data organized by frequency rather than pixel (our eyes are less sensible to high frequencies, which brings up the next step);
– The third step is the quantization, that is the only one that trows away some information. The quantization. has to do with whole number division, i.e. 27/12=2 (whole number), which in the decompression will lead to 2×12=24 (henceforth the loss of information). This is done in such a way that high frequencies are divided by larger numbers while low frequencies are divided by smaller numbers, henceforth the data thrown away is that we are the least sensible for;
– The last step is composed of a combination of lossless compression methods (runlenght, difference and huffman codings) in the resulting values, after a zig-zag reorganizaton of the data. That is finally organized in a well defined file structure and saved in a given media.
With all that said, quality control is performed by changing the quantization matrix (values for which the data will be divided). The larger the numbers the greater the amount of data thrown away (and the smallest the resulting file). So, JPEG-wise, there is no difference regarding the EV or color depth of the hardware, as that will need to be transformed to a 24bpp image prior to being JPEG compressed.
Some ideas come to mind considering all the above (and some more I couldn’t write in order to keep this not too long):
1- noise leads to high frequency of colors, i.e. an image with more noise will be tougher to compress, as the DCT won’t be as effective due to the unusual high frequency. Would the Nikon sensor generate more noise (or perhaps the Canon process noise out prior to compressing the data).
2- Perhaps Nikon is more conservative in the quantization matrix they use. We often have a few options for quality of our JPEGs. That basically means which quantization matrix is to be used.
3- A mix of options 1 and 2 above…
It is worthy noting that the above is based on the lossy JPEG compression scheme. Lossless JPEG or the more recent JPEG2000 is a whole different story, but that is not what either camera use.
The limited bpp of JPEG is a reason why people still use TIFF, as that allows as much as 48bpp
I believe that I already wrote too much, but this is a topic I like very much. It has to do with what I work with. The above is simplified to some extent, but accurate where relevant.
Cheers!
Have you enabled RAW file compression on the Nikon? Canon RAW formats are compressed by default (using a lossless algorithm).
Interesting comments. I will have to do more tests when I get back, but the reason I believe that EV and DR are significant are because those shades and colors must be represented in the final images to some degree over Canons in the final image. I also believe that Canon and Nikon use different compression algorithms, and that Canon’s is better at reducing noise in the 6D. Should lead to some interesting tests and results.