yes but ImageMagick uses QuantumRange instead of actual bit count
Sylvain Ard
0549507724
0778380991
sylvain.ard@xxxxxxxxx
http://sylvain-ard.fr
Entreprise individuelle SIRET : 80079243400022
Appt 26 Bât A Résidence Le Patio
83 rue de la Bugellerie
86000 Poitiers
Le lun. 2 mai 2022 à 11:40, Michael Perry <dmarc-noreply@xxxxxxxxxxxxx> a
écrit :
Wasn’t your scan an 8-bit scan though?
On 2 May 2022, at 07:07, Sylvain Ard <sylvain.ard@xxxxxxxxx> wrote:
Hello,
I finally succeeded by upgrading image magick with version HDRI 16 bits
and calculate the LAB values as follow from the LAB values of Image magick :
l:=l/QuantumRange
a:=255*a/QuantumRange;
if a>127 then
a:=a-255;
b:=255*b/QuantumRange;
if b>127 then
b:=b-255;
Sylvain Ard
0549507724
0778380991
sylvain.ard@xxxxxxxxx
http://sylvain-ard.fr
Entreprise individuelle SIRET : 80079243400022
Appt 26 Bât A Résidence Le Patio
83 rue de la Bugellerie
86000 Poitiers
Le lun. 2 mai 2022 à 07:23, Graeme Gill <graeme@xxxxxxxxxxxxx> a écrit :
Sylvain Ard wrote:
Hi,
hello, as I expected the measured CieLab values of the same photo ofthe color chart have nothing to do with the theoretical CieLab
measurements!
It's hard to know what you mean by that, since you give no examples.
my commands chain is : scanin -dipn -G1.0 -v2 mire.tif ref\it8.chtR202202.txt colprof -v
-D"Scanner A" -qm -ax -ua mire
Output is:
Profile check complete, peak err = 6.512414, avg err = 1.252660
That's how accurate the profile will reproduce the target CIE values.
magick mire.tif +profile ICC -profile mire.icm -profileLab-D50-Identity-elle-V4.icc -define
It's hard to know what you are attempting with this, since I'm not
familiar with "magick" (presumably the "R" package ?).
Better using an ArgyllCMS utility if you can, then we can know exactly
what
is going on and are not trying to debug some other piece of software.
i.e.
cctiff -ia mire.icc mire057.tif cie.tif
Some spot checking (using Photoshop histogram display, and translating
from 8 bit
0..255 to L*a*b*) shows delta E's that are in the expected range indicated
by the above colprof self fit error. i.e.:
Patch R202202.txt value Raster L*a*b* value
A4 22.59 33.49 12.73 22 33 15
E14 70.27 40.06 -26.64 70 40 -26
H15 85.55 1.08 72.02 85 1 73
I19 31.76 25.31 -64.09 31 27 -66
K18 49.24 -50.94 51.70 49 -50 51
so I can't understand where you think there's a problem.
Graeme Gill.
