Hi to Everyone,
My camera - a Canon eos 600D, produces JPG files as well as CR2 (Canon
raw files). Using Darktable, I expect to obtain from the Canon raw
files, better JPGs than those produced by the camera and, most of all,
in modifying, rotating, sampling, cutting etc... from CR2s, I expect not
to loose quality at each re Saving as this happens with lossy jpeg files.
Problem is... those jpg files that I obtain from an export process in
Darktable, are more than TWICE AS BIG as those delivered by my Canon
Camera ! So far I have been unable to understand why, since both use
jpeg quality 98.
These images don't show much difference when analysing them using
ImageMagick's 'identify -verbose'. Both are jpeg quality 98. Both have
18 MegaPixels, both are 8 bit. the only differences that I can see, are
as follows :
'Identify' says that Darktable jpg is 300 x 300 dpi with a print size of
11.54 x 17.3267, while the Canon JPG shows 72 x72 dpi with a print size
of 72 x 48... Am I right to consider that it comes to the same result in
the end ? It simply looks like if the Canon jpg had gone through a
resampling process and, to my opinion (that of a newbee..) it should not
change anything... In the end, when displaying each image, I don't see
much difference in that matter, and print sizes appear to be the same
once 'Gimp' has done its job
Page geometries are the same or just about, except that one image has
been turned, therefore width and length are reverse.
jpg colorspace = 2 for both images
==> jpeg sampling factors : 1x1, 1x1, 1x1 for the Darktable jpg... and,
surprisingly, 2x1, 1x1, 1x1 for the Canon jpg... ???
What is this sampling factor ? What are those three pairs of values ?
Could this explain the difference in filesizes from Darktable generated
and Camera generated JPGs ? What difference does it make in image
quality ? How to change these sampling factors in the Darktable export
process ?
Thanks in advance for any infos and tricks.
Bernard
darktable user mailing list
to unsubscribe send a mail to darktable-user+unsubscr...@lists.darktable.org