CMYK & RGB Mismatch about colorbrewer HOT 9 OPEN

Quite curious. The values are correct according to both the RGB and CMYK source spreadsheets. I believe the different specs were created separately, i.e. one is not derived from the other, and thus might differ a bit (because of what works for print and what doesn't perhaps?), but am pretty ignorant beyond that. Will see what kind of info can be dug up here.

Just for reference, AI screenshot full 7-class GnBu using RGB and CMYK is below. Looks like it's not the only scheme with such differences, though.

from colorbrewer.

Thanks! That would make sense, that maybe they were developed to be
different. I've never actually read the relevant publication. If that is
indeed the case, maybe the example map could switch colors as the user
changes from CMYK to RGB numbers.

On Thu, Sep 15, 2016 at 2:11 PM Andy Woodruff [email protected]
wrote:

Quite curious. The values are correct according to both the RGB and CMYK
source spreadsheets. I believe the different specs were created
separately, i.e. one is not derived from the other, and thus might differ a
bit (because of what works for print and what doesn't perhaps?), but am
pretty ignorant beyond that. Will see what kind of info can be dug up here.

Just for reference, AI screenshot full 7-class GnBu using RGB and CMYK is
below. Looks like it's not the only scheme with such differences, though.
[image: screen shot 2016-09-15 at 3 03 43 pm]
https://cloud.githubusercontent.com/assets/3533462/18563656/7a3c6d0e-7b56-11e6-9d79-4b431bff4e4c.png

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#12 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AFMizPu9LiYErhKbBvW_Z_xPIEJCH_d1ks5qqZhbgaJpZM4J-MoQ
.

from colorbrewer.

Could be related to what colour space you have set. Presume there is no absolute RGB-CMYK mapping, since one model is emittive and one reflective. Forgive dodgy terminology: been a while since I looked at this in any depth.

from colorbrewer.

My understanding of the history of this when talking to Cindy was this:

1. The original specs were done in CMYK despite the RGB spec being more readily available.
2. The RGB values were derived from the CMYK values in either Photoshop or Illustrator
3. At that time even Photoshop and Illustrator didn't match with respect to color management algorithms
4. There wasn't any documentation on version of Photoshop / Illustrator or the color management decisions and color profiles used to make the conversions.

I went back to original publications for the CMYK values for creation of styles for ArcGIS Pro. I provided those values to @awoodruff when I noticed they didn't match what the JS implementation of ColorBrewer was showing (it was doing a rough conversion without color management).

from colorbrewer.

Given @williamscraigm's points, I've been pondering this a bit more today. Seems like it might make sense to rebuild the RGB table from the CMYK one (a tedious task, but one I could probably take on), and this time have some sort of documentation as to what that conversion is.

There's no universally accepted CMYK to RGB conversion, since there are so many versions each of CMYK and RGB. Choosing a profile for each end is the tricky part (though also maybe not so tricky, because we're probably talking small changes).

from colorbrewer.

I've recently implemented an RGB-CMYK conversion for another project, if you don't get to it I can make the change (new to the project so will have to orient). Maybe next weekend.

from colorbrewer.

I remember it was a bit of research, but I ended up implementing it and it made sense (been a couple months so I'll look back)

from colorbrewer.

@AlexFiliakov Neat! What are the color profiles at each end?

from colorbrewer.

Hello! Any chance these values will be corrected anytime soon? I'm using 4 classes; qualitative (http://colorbrewer2.org/#type=qualitative&scheme=Set2&n=4) and the RBG values appear to be correct. The CMYK are way off. Thanks!

from colorbrewer.