another question : do you know a libtiff forum or mailing list please ?
thank you
best regards
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 jeu. 21 avr. 2022 à 19:09, Chris Lilley <chris@xxxxxx> a écrit :
This is completely correct, except for the float part.
Using 16bits per component (uint16 for L, int16 for a and b) also
preserves the fractional component.
But no-one should be representing Lab in 8bits unless they don't care
about accuracy at all! Think of it this way - if values are rounded (or,
worse, truncated) to 8bits there is a worst-case error of almost 1 on each
of L, a and b. So a deltaE 76 of sqrt(3) = 1.7, from quantization error
alone!
On 2022-04-21 09:23, graxx@xxxxxxxxxxxx wrote:
Sylvain,
CIE Lab is independent of any computer implementation.
If you look at CIE Publications 15-2 which defines Colorimetry, you won’t
find any mention of how the numbers should be represented inside a computer
memory. CIE Lab numbers are just numbers and don’t need computers to exist.
Actually, you need “floating-point” variables to correctly represent CIE
Lab, otherwise you loose the fractional part.
/ Roger
*From:* argyllcms-bounce@xxxxxxxxxxxxx <argyllcms-bounce@xxxxxxxxxxxxx>
<argyllcms-bounce@xxxxxxxxxxxxx> *On Behalf Of *Sylvain Ard
*Sent:* April 21, 2022 7:33 AM
*To:* argyllcms@xxxxxxxxxxxxx
*Subject:* [argyllcms] Re: ranges of LAB values
I want to understand, it's all. I think you're right the good formula is
255*(0 to 1 value)-128 because I saw on Internet that unsigned bytes come
from -128 to 127 and I think that CieLab is standard unsigned byte.
Do you know my imagemagick commands ?
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 jeu. 21 avr. 2022 à 13:29, Aaron Perelmuter <aaron.746@xxxxxxxxx> a
écrit :
I suppose that depends on from where one derives their definitions. I
would generally do 256*x-128 as 0>>255 contains 256 integers. Also, do the
math and substitute either 127 or 128 and I doubt the resulting colour will
be in any way noticeably different.
I am somewhat confused though, if you already know the answer to your
question, why ask? Please don’t misunderstand what I’m saying, I mean, it’s
better to be as specific as you possibly can in terms of what you’re
asking. If you understand something to be 255*value-127, why not ask about
the time you saw something very slightly different to that? Just makes it
easier to give the exact answer you’re searching for is all.
Keep the questions coming, we can all learn from them.
On Thu, 21 Apr 2022 at 9:19 pm, Sylvain Ard <sylvain.ard@xxxxxxxxx> wrote:
isn't it rather -127 - 128?
because to match 0-1 values to near 127 values we do 255*value-127 ?
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
<https://www.google.com/maps/search/83+rue+de+la+Bugellerie+86000+Poitiers?entry=gmail&source=g>
86000 Poitiers
<https://www.google.com/maps/search/83+rue+de+la+Bugellerie+86000+Poitiers?entry=gmail&source=g>
Le jeu. 21 avr. 2022 à 13:15, Aaron Perelmuter <aaron.746@xxxxxxxxx> a
écrit :
In most cases, for all intents and purposes, Lab is understood to be
within the following ranges: L*0>>100;a*-128>>127;b*-128>>127
As was mentioned in a reply to a previous list message, technically Lab
values can and do sometimes go well outside of the ranges listed but in
almost every instance, for the vast majority of people and use cases, Lab
is understood to be within the bounds listed above.
As was also mentioned in another reply to a list message, it certainly
helps to be able to give a more accurate or more pertinent answer if you
can state your reasons why you’d like to know what it is you’re asking
about.
I’m not 100% sure about this but ttbomk, there’s no such thing as a v2, v4
or any other version number for Lab profiles, for tiff or any other file
format. Is there even a Lab profile anyway? What’s the exact file name for
the Lab profiles you’re asking about? Finally, IT8 files consists of a
great many different sub-types which are used for myriad purposes. Most
commonly IT.7/3 and IT8.7/4 are often used for profiling CMYK processes,
IT8.7/2 (used to be) often used for creating scanner profiles and so on - I
say used to be because not many people in this day and age are creating
scanner profiles or even own/use a scanner. Regardless, there are several
flavours of IT8 type files but again, in almost every case the Lab range is
as listed above.
Best Regards,
Aaron.
On Thu, 21 Apr 2022 at 8:45 pm, Sylvain Ard <sylvain.ard@xxxxxxxxx> wrote:
Hi I would like to know the ranges of L,A and B values in : ICC profiles
v2, TIFF images with LAB profile v2 and v4, IT8 CGATS files and their
matches because the ranges are differents.
Thank you
Best regards
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
<https://www.google.com/maps/search/83+rue+de+la+Bugellerie+86000+Poitiers?entry=gmail&source=g>
86000 Poitiers
<https://www.google.com/maps/search/83+rue+de+la+Bugellerie+86000+Poitiers?entry=gmail&source=g>
--
Chris Lilley
@svgeesus
Technical Director @ W3C
W3C Strategy Team, Core Web Design
W3C Architecture & Technology Team, Core Web & Media
