Hi Stéphane,
thanks for your thoughts.
In the meantime I used exiftool to extract the embedded jpg from the RAW:
exiftool -b -JpgFromRaw -w _embedded.jpg nef
and those jpgs show exactly the size that is reported by Geequie (and
Nikon) - in all three versions: L, M, S.
And I found another article in the web:
https://photographylife.com/reviews/nikon-d850/4
According to this the creation of the smaller RAW versions is a two-step
process with step 1 in the camera and step 2 in the RAW converter.
So obiously darktable does not support that 2nd step - which is what I
prefer since I get the output resolution according to the file size (for
me it doesn't make much sense to store 25,7MP and get only 11,4MP out of
the RAW converter).
And yes, the difference in size in L mode seems to be due to a certain
overlap for lens corrections and the like - this is only a small
difference of some 20 or so pixels.
--
regards
Bernhard
https://www.bilddateien.de
Stéphane Gourichon schrieb am 14.01.19 um 16:37:
It may make sense after all.
Geeqie (most certainly) and Nikon (very probably) report the
definition of the JPEG file embedded in the RAW, which is the actual
output of the Nikon processing algorithm.
This would be similar to what I observe on the D5200, where the RAW
and JPEG have different definitions (sidecar JPEG and embedded JPEG
are both 6000x4000). Viewing both images and looking for possible crop
and/or resize allows to actually notice the lens correction processing
by Nikon. The point might have been to crop the picture so that most
of the time, we have both 1:1 pixel coverage in the center while
avoiding black areas on the side and corners due to lens correction.
Using ufraw-batch --embedded to extract the embedded JPG from a D5200
RAW file indeed produces a 6000x4000 JPEG.
Interestingly, darktable report the NEF is 6016x4016 while ufraw
reports 6078x4058. This is becoming stranger.
Anyway, D5200 has only one RAW definition.
Perhaps the embedded JPEG size is really much smaller when using
reduced RAWs, because, you know, the point is the camera user asked to
save storage space and, since they have the raw file they're supposed
to have means to regenerate a JPEG from that.
Can you try the commands below for each raw and sidecar-jpeg files?
identify mypicture.JPG
ufraw-batch --embedded mypicture.NEF ; identify mypicture.embedded.jpg
ufraw-batch --out-type=jpg --output=mypicture.ufrawprocessed.jpg
mypicture.NEF ; identify mypicture.ufrawprocessed.jpg
Le 12/01/2019 à 23.58, Bernhard a écrit :
it looks like the RAW files published here https://raw.pixls.us/ show
the same deviation - and the exifdata show the same values like
darktable.
I still do not understand why Geeqie and Nikon report other numbers ...
Bernhard schrieb am 11.01.19 um 14:27:
Nikon enables the use of RAW files of different sizes in the D850.
According to the specifications in the manual, the formats are
defined as follows:
L 8256 × 5504
M 6192 × 4128
S 4128 × 2752
(these figures are also shown in the camera menu)
darktable tells me:
L 8288 x 5520
M 7104 x 4728
S 6216 x 4136
While the deviations for size L are still small (we had a discussion
about them in the past), they are very significant for M and L -
according to Nikon a file of size S 11.3MP, while darktable assumes
over 25MP.
Geeqie again shows the Nikon numbers.
Exiftool shows the darktable numbers.
A jpg or tiff file exported from darktable shows the darktable
numbers in xnview.
What is going on here?
____________________________________________________________________________
darktable user mailing list
to unsubscribe send a mail to [email protected]
