Unable to convert colors?

[Roger-Breton]

BTW, Coloraide is not for the faint of heart.

I managed to get to this point:

rom coloraide import cat
from coloraide import Color
lab_D65 = Color("lab", [30, 26, -27], illuminant="D65")

xyz_d65 = lab_D65.convert("xyz-d65").coords()

xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyz_d65, method='bradford')

But I can't figure how to convert xyz_d50 to Lab_d50?

Lab_d50 = xyz_d50 ???

[facelessuser]

I think you are greatly overthinking things. ColorAide generally tries to dumb things down a lot for the average user.

>>> Color("lab", [30, 26, -27]).convert('xyz-d65')
color(xyz-d65 0.08955 0.06312 0.16465 / 1)

ColorAide exposes xyz-d65 and xyz-d50. lab uses D50, lab-d65 uses D65 illuminant. If I could go back, I'd probably use the ID lab-d50 for lab even if it serializes to lab for CSS.

Yes, you could subclass the existing XYZ spaces and add more XYZ spaces with different illuminants, but we added the ones that the average user would need to use.

[Roger-Breton]

I'm not overthinking things. I stumbled upon your library doing some research on CAT but I didn't expect using your library would prove so unintuitive. Your two lines of code are not helping with my case which is, having chromatically adapted a D65 XYZ value through the Bradford matrix, to simply convert the resulting XYZ value back to Lab.

I know how to do CAT in Matlab and Excel but I was curious to see how you do it in your library.
I've also used color-science before but evenb if it has a CAT function, it's so poorly documented that there's no way to make it work.
Believe me, I tried....

[facelessuser]

I'm not trying to be insulting, so if it came out that way, I do apologize.

I have examples about chromatic adaptation in the documentation, with examples. It is meant to document functions, but most users will not need to use this because it is built into the conversion process. But here is how I can manually convert to XYZ D65 and also how I would suggest most users do it, by using .convert().

from coloraide import cat
from coloraide import util
lab = Color("lab", [30, 26, -27])
xyz_d65 = Color.chromatic_adaptation(cat.WHITES['2deg']['D50'], cat.WHITES['2deg']['D65'], lab.convert('xyz-d50').coords())
Color('xyz-d65', xyz_d65)
Color("lab", [30, 26, -27]).convert('xyz-d65')

Example

I have an example in the documentation that goes over this exact case, but with Prophoto RGB: https://facelessuser.github.io/coloraide/cat/

I cover the concepts because we expose the functionality, but the average user is just going to convert the shipped, pre-packaged spaces to the other pre-packaged spaces, without having to worry about how they get there. Chormatic adaptation will happen automatically.

[Roger-Breton]

I feel I don't know enough about Object-oriented programming. So when you talk "sub-classes", even though the concept looks crystal-clear to you, I struggle...
I thank you for taking the time to show what steps would be require beyond calling the chromati_adaptation method.
As I was searching your code, I had the intuition that I had to "define" a new variable based on the CAT results but not having enough programming know-how, I ended up throwing the towel.
You see, my VERY humble programming know-how would tell me that, after doing some operation, the result can be readily acted upon, Like, having obtained the xyz_d65, because I'm not familiar enough with the underlying object structures, I was persuaded I could directly use the result in a "convert" operation, naively.
But the way you have it, 1) you don't use any variable names? Which is completely alien to me? And 2) you "seem" to be creating a new instance of a variable to which you assign the 'xyz-d65' color space, and 3) in order to execute an XYZ to Lab conversion, which your last line of code does not seem to do? You've gone back to my starting Lab_D65 variable? Why? 5) I am looking for the way to convert 'xyz_d65" to Lab_d50. So 6) was it my mistake to name the result of the CAT 'xyz_d65'? When the result is actually XYZ "D50"?

I checked out your documentation, on your ProPhoto example.
So I could have typed: XYZ_D50 = Color('xyz-d50', xyz_d50).convert('lab-d50')?

Thank you for all your time and help....

[facelessuser]

I feel I don't know enough about Object-oriented programming. So when you talk "sub-classes", even though the concept looks crystal-clear to you, I struggle...

I understand. Unfortunately, some assumptions must be made about people's understanding of the underlying language (Python in this case), but I understand that that means some users may be confused about some concepts.

But the way you have it, 1) you don't use any variable names? Which is completely alien to me? And 2) you "seem" to be creating a new instance of a variable to which you assign the 'xyz-d65' color space, and 3) in order to execute an XYZ to Lab conversion, which your last line of code does not seem to do? You've gone back to my starting Lab_D65 variable? Why? 5) I am looking for the way to convert 'xyz_d65" to Lab_d50. So 6) was it my mistake to name the result of the CAT 'xyz_d65'? When the result is actually XYZ "D50"?

Let me back up a little, I guess I'm trying to demonstrate 2 things. One, how to manually apply chromatic adaptation. I may be getting a little sloppy with my examples. The playground I link to is akin to doing things in a python terminal session. But let me be more clear with my example.

Here we are simply taking the Lab D50 color and converting it to XYZ D50. We manually apply the chromatic adaptation which requires the starting white point and the target white point (D50 and D65 respectively) and the XYZ coordinates (relative to D50) that require the chromatic adaptation. That will return XYZ coordinates, adapted to XYZ D65. We can take those coordinates and instantiate a new Color object that is using the XYZ D65 color space with those coordinates.

Manual Example

>>> from coloraide import Color
>>> from coloraide import cat
>>> lab = Color("lab", [30, 26, -27])
>>> xyz_d50 = lab.convert('xyz-d50')
>>> white_point_d50 = cat.WHITES['2deg']['D50']
>>> white_point_d65 = cat.WHITES['2deg']['D65']
>>> xyz_d65_coords = Color.chromatic_adaptation(white_point_d50, white_point_d65, xyz_d50.coords())
>>> xyz_d65 = Color('xyz-d65', xyz_d65_coords)
>>> print(xyz_d65)
color(xyz-d65 0.08955 0.06312 0.16465 / 1)

ColorAide uses this internally any time a color is converted. Any conversion that requires a white point change will cause the color to be converted to XYZ (using the current white point), chromatic adaptation will then be applied so that it is in XYZ with the new white point, and then it will be converted to target color space.

As long as the the appropriate color spaces plugins are registered, the user doesn't have to think about chromatic adaptation. It is probably good to know it happens, but it will be taken care of for them. That is the second thing I was trying to demonstrate.

Automatic Example.

>>> from coloraide import Color
>>> lab = Color("lab", [30, 26, -27])
>>> xyz_d65 = lab.convert('xyz-d65')
>>> print(xyz_d65)
color(xyz-d65 0.08955 0.06312 0.16465 / 1)

Here I am being more explicit.

Since chromatic adaptation happens through XYZ, XYZ is kind of the bridge color space, almost all color conversions go through XYZ. Some exceptions are things like sRGB to HSL since HSL is just a different representation of sRGB. But if we wanted to go from HSL to Lab, we would go (roughly) from HSL -> sRGB -> XYZ D65 -> XYZ D50 -> Lab.

The user doesn't need to care about all of this and can simply specify.

>>> Color('hsl', [270, 1, 0.5]).convert('lab')
color(--lab 39.283 74.649 -95.545 / 1)

This "dumbs" down the entire process so the user can just specify the color and convert to whatever they want (assuming it is supported) and go. Obviously, for those who want to know the entire process, this hides a lot of that.

[facelessuser]

If it helps, each supported color space has their name mentioned in their properties description in the documentation: https://facelessuser.github.io/coloraide/colors/xyz_d50/#xyz-d50.

[Roger-Breton]

It took me a while to figure what the correct string to use for this simple conversion:
xyz_d65 = lab_D65.convert("xyz-d65").coords()

[facelessuser]

That is fair. It is possible I need to more clearly provide such information in one place, maybe a table with supported color spaces and their ID, and briefly mention that while also linking that table when talking about conversion.

[facelessuser]

Moving forward, I will likely be creating a table like what is shown below, and will link to it when discussing the conversion:

Supported Color Space IDs

Color Space	ID
A98 RGB	a98-rgb
Linear A98 RGB	a98-rgb-linear
ACES 2065-1	aces2065-1
ACEScc	acescc
ACEScct	acescct
ACEScg	acescg
CAM16 JMh	cam16-jmh
CAM16 LCD	cam16-lcd
CAM16 SCD	cam16-scd
CAM16 UCS	cam16-ucs
CMY	cmy
CMYK	cmyk
Cubehelix	cubehelix
DIN99o	din99o
Display P3	display-p3
Linear Display P3	display-p3-linear
HCT	hct
HPLuv	hpluv
HSI	hsi
HSL	hsl
HSLuv	hsluv
HSV	hsv
Hunter Lab	hunter-lab
HWB	hwb
ICtCp	ictcp
IgPgTg	igpgtg
IPT	ipt
Jzazbz	jzazbz
JzCzhz	jzczhz
Lab (D50)	lab
Lab (D65)	lab-d65
LCh (D50)	lch
LCH (D65)	lch-d65
DIN99o LCh	lch99o
Luv LCh	lchuv
Luv	luv
Okhsl	okhsl
Okhsv	okhsv
Oklab	oklab
OkLCh	oklch
oRGB	orgb
Prismatic	prismatic
ProPhoto RGB	prophoto-rgb
Linear ProPhoto RGB	prophoto-rgb-linear
Rec. 2020	rec2020
Linear Rec. 2020	rec2020-linear
Rec. 2100 HLG	rec2100-hlg
Linear Rec. 2100	rec2100-linear
Rec. 2100 PQ	rec2100-pq
Rec. 709	rec709
RLAB	rlab
RYB	ryb
sRGB	srgb
Linear sRGB	srgb-linear
UCS	ucs
XYB	xyb
xyY	xyy
XYZ (D50)	xyz-d50
XYZ (D65)	xyz-d65

[facelessuser]

I will also state that most people will not need to use chromatic_adaptation directly. It's there, and we use it internally during conversions and such, but the hope is that the average user will not need to touch it. If you are doing advanced stuff, there are times when that may be handy. I don't expect everything that ColorAide exposes to be used by the majority of people, but some of it has to exist so we can provide the easy stuff, like a simple .convert().

Our goal isn't to necessarily be a colour science library, which is a great library, but to appeal to people that just want to use colors.

It is possible that maybe sometimes I delve a little too deep into topics in the documentation 😅. There's a lot to know about colors, so I think it is interesting to help people understand some of it, but the average user is going to just be converting, filtering, mixing, etc. We wanted to make that fairly easy to do.

[Roger-Breton]

Your library has unique "conceptual appeal",
I never seen a way to define a Lab color while simultaneously specify its Illuminant:
lab_D65 = Color("lab", [30, 26, -27], illuminant="D65")

[facelessuser]

Yes, we do not require the user to specify the illuminant manually:

>>> Color('lab', [30, 26, -27])
color(--lab 30 26 -27 / 1)
>>> Color('lab-d65', [30, 26, -27])
color(--lab-d65 30 26 -27 / 1)

The target audience is for the average user, used to dealing with things like CSS colors and such.

>>> Color('lab(30% 26 -27)')
color(--lab 30 26 -27 / 1)

At times, they can barely know about the color concepts involved, they just want to go from one color to the other.

[facelessuser]

We wanted to get a little a way from the fine control of constantly remembering what illuminant you are working with and whether the space you are converting to even supports that illuminate. The spaces are preconfigured with the appropriate illuminant and conditions, and you just convert to them.

If you are an advanced user, and know if the space supports the illuminant you are targeting or can be applied to a different gamut, you can subclass the color space, configure it, register it with your own unique name, and then use it. These aren't the average users we are targeting, but they may exist and the functionality is there. I use the functionality to prototype color spaces and experiment with things.

We don't expose every variant of XYZ with every illuminant, these aren't the kind of people the library is made for, but yes, you can expose more XYZ variants if you know what you are doing, and this allows you to take advantage of the built-in chromatic adaptation.

from coloraide import Color as Base
from coloraide.spaces import xyz_d65
from coloraide.cat import WHITES

class XYZC(xyz_d65.XYZD65):
    NAME = 'xyz-c'
    SERIALIZE = ('--xyz-c',)
    WHITE = WHITES['2deg']['C']

class Color(Base): ...

Color.register(XYZC())

Color('red').convert('xyz-c')
Color('red').convert('xyz-c').convert('xyz-d65')

Example

[facelessuser]

Lastly, the documentation covers color object generally (how to create colors, convert colors, clone colors, etc.) https://facelessuser.github.io/coloraide/color/#converting.

The documentation also covers color manipulation: https://facelessuser.github.io/coloraide/manipulation/.

[Roger-Breton]

Believe me, I searched your "documentation" extensibley. But before I threw the towel, I took a chance to contact you here.

[facelessuser]

I am certainly open to suggestions on ways to improve the documentation. I'm not going to say I've perfectly made everything intuitive, but I've definitely put in a lot of effort and time to try.

[Roger-Breton]

Here is modified code: from coloraide import cat from coloraide import Color lab_D65 = Color("lab", [30, 26, -27], illuminant="D65") # Convert Lab (D65) to XYZ (D65) xyz_d65 = lab_D65.convert("xyz-d65").coords() # Apply chromatic adaptation from D65 to D50 # xyzd50 = Color('prophoto-rgb', [1, 1, 1]).convert('xyz-d50').coords() xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyz_d65, method='bradford') # Convert XYZ (D50) to Lab (D50) lab_d50 = Color('xyz-d50', xyz_d50).convert('lab') The problem I run into is that the adapted D50 XYZ values are not comparable to the adapted D50 XYZ values I get either in Excel or Matlab? Lab Violet 287ª 6500K X D65 8.63 30.00 Y D65 6.24 26.49 Z D65 16.47 -27.25 Lab Violet 287ª D50 X D50 8.36 29.79 Y D50 6.15 23.94 Z D50 12.40 -27.40 ColorAide XYZ Lab D50 9.45 30.56 D50 6.46 29.83 D50 21.88 -48.23 / Roger

________________________________ From: Isaac Muse ***@***.***> Sent: Monday, April 1, 2024 3:50 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Author ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) I feel I don't know enough about Object-oriented programming. So when you talk "sub-classes", even though the concept looks crystal-clear to you, I struggle... I understand. Unfortunately, some assumptions must be made about people's understanding of the underlying language (Python in this case), but I understand that that means some users may be confused about some concepts. But the way you have it, 1) you don't use any variable names? Which is completely alien to me? And 2) you "seem" to be creating a new instance of a variable to which you assign the 'xyz-d65' color space, and 3) in order to execute an XYZ to Lab conversion, which your last line of code does not seem to do? You've gone back to my starting Lab_D65 variable? Why? 5) I am looking for the way to convert 'xyz_d65" to Lab_d50. So 6) was it my mistake to name the result of the CAT 'xyz_d65'? When the result is actually XYZ "D50"? Let me back up a little, I guess I'm trying to demonstrate 2 things. One, how to manually apply chromatic adaptation. I may be getting a little sloppy with my examples. The playground I link to is akin to doing things in a python terminal session. But let me be more clear with my example. Here we are simply taking the Lab D50 color and converting it to XYZ D50. We manually apply the chromatic adaptation which requires the starting white point and the target white point (D50 and D65 respectively) and the XYZ coordinates (relative to D50) that require the chromatic adaptation. That will return XYZ coordinates, adapted to XYZ D65. We can take those coordinates and instantiate a new Color object that is using the XYZ D65 color space with those coordinates. Manual Example<https://facelessuser.github.io/coloraide/playground/?code=from%20coloraide%20import%20Color%0Afrom%20coloraide%20import%20cat%0Alab%20%3D%20Color%28%22lab%22%2C%20%5B30%2C%2026%2C%20-27%5D%29%0Axyz_d50%20%3D%20lab.convert%28%27xyz-d50%27%29%0Awhite_point_d50%20%3D%20cat.WHITES%5B%272deg%27%5D%5B%27D50%27%5D%0Awhite_point_d65%20%3D%20cat.WHITES%5B%272deg%27%5D%5B%27D65%27%5D%0Axyz_d65_coords%20%3D%20Color.chromatic_adaptation%28white_point_d50%2C%20white_point_d65%2C%20xyz_d50.coords%28%29%29%0Axyz_d65%20%3D%20Color%28%27xyz-d65%27%2C%20xyz_d65_coords%29%0Aprint%28xyz_d65%29>

>> from coloraide import Color >> from coloraide import cat >> lab = Color("lab", [30, 26, -27]) >> xyz_d50 = lab.convert('xyz-d50') >> white_point_d50 = cat.WHITES['2deg']['D50'] >> white_point_d65 = cat.WHITES['2deg']['D65'] >> xyz_d65_coords = Color.chromatic_adaptation(white_point_d50, white_point_d65, xyz_d50.coords()) >> xyz_d65 = Color('xyz-d65', xyz_d65_coords) >> print(xyz_d65)

color(xyz-d65 0.08955 0.06312 0.16465 / 1) ColorAide uses this internally any time a color is converted. Any conversion that requires a white point change will cause the color to be converted to XYZ (using the current white point), chromatic adaptation will then be applied so that it is in XYZ with the new white point, and then it will be converted to target color space. As long as the the appropriate color spaces plugins are registered, the user doesn't have to think about chromatic adaptation. It is probably good to know it happens, but it will be taken care of for them. That is the second thing I was trying to demonstrate. Automatic Example<https://facelessuser.github.io/coloraide/playground/?code=from%20coloraide%20import%20Color%0Alab%20%3D%20Color%28%22lab%22%2C%20%5B30%2C%2026%2C%20-27%5D%29%0Axyz_d65%20%3D%20lab.convert%28%27xyz-d65%27%29%0Aprint%28xyz_d65%29>.

>> from coloraide import Color >> lab = Color("lab", [30, 26, -27]) >> xyz_d65 = lab.convert('xyz-d65') >> print(xyz_d65)

color(xyz-d65 0.08955 0.06312 0.16465 / 1) Here I am being more explicit. Since chromatic adaptation happens through XYZ, XYZ is kind of the bridge color space, almost all color conversions go through XYZ. Some exceptions are things like sRGB to HSL since HSL is just a different representation of sRGB. But if we wanted to go from HSL to Lab, we would go (roughly) from HSL -> sRGB -> XYZ D65 -> XYZ D50 -> Lab. The user doesn't need to care about all of this and can simply specify.

>> Color('hsl', [270, 1, 0.5]).convert('lab')

color(--lab 39.283 74.649 -95.545 / 1) This "dumbs" down the entire process so the user can just specify the color and convert to whatever they want (assuming it is supported) and go. Obviously, for those who want to know the entire process, this hides a lot of that. — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5E3IV7QINOKJOCUHSTY3G3CDAVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DSNZWGEZTO>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[facelessuser]

@Roger-Breton Can you post your data in a more readable format? I'm just having trouble understanding what each line means. I'm sure it probably came from a spreadsheet or something, but it is a bit confusing in its current form.

[Roger-Breton]

Please find attached the Excel sheet where you can verify my transformations. Here is the python code : from coloraide import cat from coloraide import Color lab_D65 = Color("lab", [30, 26, -27], illuminant="D65") # Convert Lab (D65) to XYZ (D65) xyz_d65 = lab_D65.convert("xyz-d65").coords() # Apply chromatic adaptation from D65 to D50 xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyz_d65, method='bradford') # Convert XYZ (D50) to Lab (D50) lab_d50 = Color('xyz-d50', xyz_d50).convert('lab') I'm sorry to put you through this mess. I hate to think that this is most likely a problem of "misunderstanding" on my part. / Roger

…

________________________________ From: Isaac Muse ***@***.***> Sent: Wednesday, April 3, 2024 5:02 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Mention ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) @Roger-Breton<https://github.com/Roger-Breton> Can you post your data in a more readable format? I'm just having trouble understanding what each line means. I'm sure it probably came from a spreadsheet or something, but it is a bit confusing in its current form. — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5EWNNWVW4L4NLXPPQDY3RU6FAVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMBSGEYTO>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[facelessuser]

Unfortunately, spreadsheets can't be attached to Github comments via email. Maybe if you can just take a screenshot and attach it directly in GitHub (or attach the spreadsheet if GitHub allows it...it may not).

[facelessuser]

@Roger-Breton I've got an idea. Lets try to compare against an unbiased source. We are going to use https://www.easyrgb.com/en/convert.php#inputFORM to help us compare what we are doing. Now, we need to consider that there may be slight deviation depending on what exactly they use for the D65 white point. Do they use the chromaticity points like we do and generate the XYZ coordinates? Do they take a rounded XYZ white point? Believe me, everyone does not use the exact same D65 white point, and I'm not claiming ours is the one true white point either. Also, we have to consider the precision of their conversion matrices and their CAT matrices, they may be using 32 bit precision, or even less.

Regardless of the exact white point and precisions, we should be approximately the same.

We will use XYZ (D50/2°) = 55.992090 63.32206 63.754826 as the input to their calculator, and when doing so, we get the following output:

XYZ (D50/2°) = 55.992090 63.32206 63.754826
sRGB   0-255 =  173.038  215.993  229.986    D65/2°
sRGB   0-1.0 =  0.67858  0.84703  0.90191
RGB Adobe 98 =  185.143  214.931  228.671
HSL    0-1.0 =  0.54095  0.53236  0.79024   194.74°
HSV    0-1.0 =  0.54095  0.24762  0.90191
HSI    0-1.0 =  0.54095  0.16139  0.80917
CMY    0-1.0 =  0.32142  0.15297  0.09809
CMYK   0-1.0 =  0.24762  0.06084  0.00000  0.09809
CMYK   0-100 =   24.762    6.084    0.000    9.809
XYZ          =   55.992   63.322   63.755    D50/2°
Yxy          =   63.322  0.30585  0.34589
CIE-L*ab     =   83.612  -12.215  -11.774
CIE-L*Ch(ab) =   83.612   16.966  223.947°
CIE-L*uv     =   83.612  -23.984  -13.046
CIE-L*Ch(uv) =   83.612   27.303  208.544°
HunterLab    =   79.575  -11.456  -10.259
Pre-formatted code:
CSS, HTML    = #ADD8E6
Java         = new Color( 173, 216, 230 )
.NET         = Color.FromArgb( 255, 173, 216, 230 );
Objective-C  = [UIColor colorWithRed:0.67858 green:0.84703 blue:0.90191 alpha:1.00000];
openGL       = glColor3f( 0.67858f, 0.84703f, 0.90191f );
Swift        = UIColor(red:0.67858, green:0.84703, blue:0.90191, alpha:1.00000)
Xojo         = rgb( 173, 216, 230, 0 )

So we are going to focus on two things here, the sRGB 0 - 1 output as it requires chromatic adaptation from the D50 white point to D65 white point. And we will consider the CIE-L*ab value which should use a D50 white point because we specified a D50 white point as our input. These are the results we are comparing.

--- Input ---
XYZ (D50/2°) = 55.992090 63.32206 63.754826
--- conversions ---
sRGB   0-1.0 =  0.67858  0.84703  0.90191
CIE-L*ab     =   83.612  -12.215  -11.774

Let's see what we get with ColorAide. We need to scale the XYZ values from 0 - 100 to 0 - 1 by dividing by 100, but then we can convert and see how ColorAide converts to these spaces, and in the case of sRGB, chromatically adapts.

>>> Color('xyz-d50', [c / 100 for c in [55.992090, 63.32206, 63.754826]]).convert('srgb')
color(srgb 0.67843 0.84706 0.90196 / 1)
>>> Color('xyz-d50', [c / 100 for c in [55.992090, 63.32206, 63.754826]]).convert('lab')
color(--lab 83.612 -12.226 -11.782 / 1)

Not bad, we are roughly the same. Deviations are likely due to the exact white point used, along with the precision of the white point, and the precision of the matrices. I think it is safe to say we are not doing anything wildly different from others.

[Roger-Breton]

Starting CIE LCh (uv) D65 value = 30 40 287ª XYZ Source (D65) Lab Source LMS Bradford Matrix Scaler Illuminant XYZ (D65) Violet 287ª 6500K Rho 0.9423475 0.8961 0.2664 -0.1614 0.9504296 X D65 8.63 30.00 Gamma 1.0404467 -0.7502 1.7135 0.0367 1 Y D65 6.24 26.49 Beta 1.0895008 0.0389 -0.0685 1.0296 1.0888005 Z D65 16.47 -27.25 XYZ Destination (D50) Lab Destination LMS Bradford Matrix Scaler Illuminant XYZ (D50) Violet 287ª D50 Rho 0.9972185 0.8961 0.2664 -0.1614 0.9641658 X D50 8.36 29.79 Gamma 1.0204639 -0.7502 1.7135 0.0367 1 Y D50 6.15 23.94 Beta 0.8185249 0.0389 -0.0685 1.0296 0.825096 Z D50 12.40 -27.40 Ratio of Illuminants 1.0582279 0 0 ColorAide 0 0.980794 0 XYZ Lab 0 0 0.7512844 D50 9.45 30.56 D50 6.46 29.83 Inverse of Matrix Scaler D50 21.88 -48.23 0.9860197 -0.1469093 0.1598049 0.4318790 0.5184238 0.0492221 -0.0085203 0.0400416 0.9684881 Inverse * Ratios 1.0434336 -0.144088 0.1200589 0.4570264 0.5084669 0.0369798 -0.009016 0.0392725 0.72761 Final Bradford Matrix 1.0477857 0.0228524 -0.050086 0.029528 0.9904768 -0.017029 -0.009238 0.0150502 0.7520438

…

________________________________ From: Isaac Muse ***@***.***> Sent: Wednesday, April 3, 2024 5:20 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Mention ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) Unfortunately, spreadsheets can't be attached to Github comments via email. Maybe if you can just take a screenshot and attach it directly in GitHub (or attach the spreadsheet if GitHub allows it...it may not). — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5FEOZV4473NKTLW4PTY3RXAPAVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMBSGIZTS>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[facelessuser]

Apologies, but I still can't make sense of this. It just comes out as a column of data, and I think I just lose the context of what I'm looking at. If you can make the spreadsheet available via Google Drive, DropBox, or some other file sharing service, I think we'll be able to move forward. Or even post a screenshot of the data organized with better context. I just really need the data formatted so I can be sure what I'm looking at.

Did you see my earlier post as well comparing against easyRGB? There seems to be some doubt as to whether we are doing things correctly, but compared to another unbiased source, our results are comparable.

[Roger-Breton]

Try this link: [https://res-h3.public.cdn.office.net/assets/mail/file-icon/png/xlsx_16x16.png]Bradford CAT Excel (2024 04 03 for ColorAide).xlsx<https://1drv.ms/x/s!AkD78CVR1NBqm_UZ1YxEgqbDdOlbZw?e=Sjt1LZ> I have to say, the way I'm doing this in Excel is correct. And I come out the same results, to a fraction of a decimal, in Matlab. So I know my method is sound. / Roger

…

________________________________ From: Isaac Muse ***@***.***> Sent: Thursday, April 4, 2024 3:18 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Mention ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) Apologies, but I still can't make sense of this. It just comes out as a column of data, and I think I just lose the context of what I'm looking at. If you can make the spreadsheet available via Google Drive, DropBox, or some other file sharing service, I think we'll be able to move forward. Or even post a screenshot of the data organized with better context. I just really need the data formatted so I can be sure what I'm looking at. Did you see my earlier post as well comparing against easyRGB? There seems to be some doubt as to whether we are doing things correctly, but compared to another unbiased source, our results are comparable. — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5GF4ADLUYDNIPCELLDY3WRR7AVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMJTGY4DQ>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[facelessuser]

@Roger-Breton Okay, I can now make sense of things. Your claims for ColorAide output are wrong. I will demonstrate below.

Your results which we are comparing against. Our whitepoint will be slightly different just due to how we generate it from the chromaticity points, but I will output them for comparison. Generally our results will be comparable.

                Lab
Violet  287ª    6500K
X D65   8.63    30.00
Y D65   6.24    26.49
Z D65   16.47   -27.25
        
                Lab
Violet  287ª    D50
X D50   8.36    29.79
Y D50   6.15    23.94
Z D50   12.40   -27.40

Here are ColorAides actual results. I want you to note that we scale XYZ between 0 - 1, not 0 - 100. We specifically note this here and here. I've also printed our white point as it does slightly vary from yours. I'm just noting it, not to say yours is wrong, but to explain why there will be some slight deviation. Generally, results will be approximately the same.

Source:

>>> from coloraide import Color
>>> from coloraide import util
>>> from coloraide import cat
>>> util.xy_to_xyz(cat.WHITES['2deg']['D65'])
[0.9504559270516716, 1.0, 1.0890577507598784]
>>> util.xy_to_xyz(cat.WHITES['2deg']['D50'])
[0.9642956764295677, 1.0, 0.8251046025104602]
>>> XYZ100 = [8.63, 6.24, 16.47]
>>> XYZ1 = [c / 100 for c in XYZ100]
>>> xyzd65 = Color('xyz-d65', XYZ1)
>>> labd65 = xyzd65.convert('lab-d65')
>>> xyzd50 = xyzd65.convert('xyz-d50')
>>> labd50 = xyzd65.convert('lab')
>>> print(xyzd65)
color(xyz-d65 0.0863 0.0624 0.1647 / 1)
>>> print(xyzd50)
color(xyz-d50 0.0836 0.06155 0.12398 / 1)
>>> print(labd65)
color(--lab-d65 30.01 26.412 -27.228 / 1)
>>> print(labd50)
color(--lab 29.8 23.882 -27.361 / 1)

To clear up whether this works manually, see below.

>>> from coloraide import Color
>>> from coloraide import cat
>>> XYZ100 = [8.63, 6.24, 16.47]
>>> XYZ1 = [c / 100 for c in XYZ100]
>>> xyzd65 = Color('xyz-d65', XYZ1)
>>> xyzd50 = Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords())
>>> print(xyzd50)
[0.08360155956434158, 0.061547933420552856, 0.12397546369244132]
>>> Color('xyz-d50', xyzd50).convert('lab')
color(--lab 29.8 23.882 -27.361 / 1)

This should finally clear up any confusion about us performing the CAT correctly or not.

[Roger-Breton]

Dear Isaac, I believe the differences between our respective CAT adaptation stems from the fact that, in my original (naive?) chromatic_adaptation call, I specified the use of the optional 'Bradford' parameter: xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyz_d65, method='bradford') whereas in your code, today, you don't : xyzd50 = Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords()) You see? And that makes for two different results. I concede to you that, without the use of the optional 'Bradford' parameter, in your call statement, the adapted xyzD50 is correct. But why? Regards, / Roger

________________________________ From: Isaac Muse ***@***.***> Sent: Thursday, April 4, 2024 3:49 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Mention ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) @Roger-Breton<https://github.com/Roger-Breton> Okay, I can now make sense of things. Your claims for ColorAide output are wrong, most likely because I suspect. I will demonstrate below. Your results which we are comparing against. Our whitepoint will be slightly different just due to how we generate it from the chromaticity points, but I will output them for comparison. Generally our results will be comparable. Lab Violet 287ª 6500K X D65 8.63 30.00 Y D65 6.24 26.49 Z D65 16.47 -27.25 Lab Violet 287ª D50 X D50 8.36 29.79 Y D50 6.15 23.94 Z D50 12.40 -27.40 Here are ColorAides actual results. I want you to note that we scale XYZ between 0 - 1, not 0 - 100. We specifically note this here<https://facelessuser.github.io/coloraide/colors/xyz_d65/> and here<https://facelessuser.github.io/coloraide/colors/xyz_d50/>. I've also printed our white point as it does slightly vary from yours. I'm just noting it, not to say yours is wrong, but to explain why there will be some slight deviation. Generally, results will be approximately the same. Source:

>> from coloraide import Color >> from coloraide import util >> from coloraide import cat >> util.xy_to_xyz(cat.WHITES['2deg']['D65'])

[0.9504559270516716, 1.0, 1.0890577507598784]

>> util.xy_to_xyz(cat.WHITES['2deg']['D50'])

[0.9642956764295677, 1.0, 0.8251046025104602]

>> XYZ100 = [8.63, 6.24, 16.47] >> XYZ1 = [c / 100 for c in XYZ100] >> xyzd65 = Color('xyz-d65', XYZ1) >> labd65 = xyzd65.convert('lab-d65') >> xyzd50 = xyzd65.convert('xyz-d50') >> labd50 = xyzd65.convert('lab') >> print(xyzd65)

color(xyz-d65 0.0863 0.0624 0.1647 / 1)

>> print(xyzd50)

color(xyz-d50 0.0836 0.06155 0.12398 / 1)

>> print(labd65)

color(--lab-d65 30.01 26.412 -27.228 / 1)

>> print(labd50)

color(--lab 29.8 23.882 -27.361 / 1) To clear up whether this works manually, see below.

>> from coloraide import Color >> from coloraide import cat >> XYZ100 = [8.63, 6.24, 16.47] >> XYZ1 = [c / 100 for c in XYZ100] >> xyzd65 = Color('xyz-d65', XYZ1) >> xyzd50 = Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords()) >> print(xyzd50)

[0.08360155956434158, 0.061547933420552856, 0.12397546369244132]

>> Color('xyz-d50', xyzd50).convert('lab')

color(--lab 29.8 23.882 -27.361 / 1) This should finally clear up any confusion about us performing the CAT correctly or not. — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5BKIXW6C6PCM2U7MRLY3WVEZAVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMJTHEZDO>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[facelessuser]

I believe the differences between our respective CAT adaptation stems from the fact that, in my original (naive?) chromatic_adaptation call, I specified the use of the optional 'Bradford' parameter:

No, the default IS bradford. See here.

Let me demonstrate so there is no more confusion:

>>> from coloraide import Color
>>> from coloraide import cat
>>> XYZ100 = [8.63, 6.24, 16.47]
>>> XYZ1 = [c / 100 for c in XYZ100]
>>> xyzd65 = Color('xyz-d65', XYZ1)
>>> Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords())
[0.08360155956434158, 0.061547933420552856, 0.12397546369244132]
>>> Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords(), method='bradford')
[0.08360155956434158, 0.061547933420552856, 0.12397546369244132]

Exact same results because "bradford" is the default.

So, why are you getting different results? Let's take a closer look at what you are doing.

xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyz_d65, method='bradford')

The signature of the function is:

Color.chromatic_adaptation(from_white_point, to_white_point, xyz_to_convert)

You are giving it a D65 XYZ color, but saying it is a D50 when it is not, and then converting it to D65, which is not what you want. You want this.

xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D65"], cat.WHITES['2deg']["D50"], xyz_d65, method='bradford')

Hopefully, everything is clear now. If there was some source of confusion that led you to believe that the order you were using was correct over what I did, then please do let me know and I will be happy to correct it.

[Roger-Breton]

I see. So I was converting in the wrong direction? Unknowingly? Well, it wasn't obvious which parameter was considered "Source" and which parameter was considered "Destination" from your documentation Chromatic Adaptation - ColorAide Documentation (facelessuser.github.io)<https://facelessuser.github.io/coloraide/cat/> These four lines of code were the only thing that there are :

>> from coloraide import cat >> xyzd50 = Color('prophoto-rgb', [1, 1, 1]).convert('xyz-d50').coords() >> xyzd65 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyzd50, method='bradford') >> manual = Color('xyz-d65', xyzd65).convert('srgb') >> auto = Color('prophoto-rgb', [1, 1, 1]).convert('srgb') >> manual, auto

(color(srgb 1 1 1 / 1), color(srgb 1 1 1 / 1)) You tell me which parameter is which from this example, Isaac? The documentation ought to specify which argument is Source and which is Destination, along the lines of

>> xyzd65 = Color.chromatic_adaptation(Source, Destination, xyzd50, method='bradford')

I guess I could have done a better job of inferring which is which since the variable to be converted as "xyzd50" and the output of the conversion was "xyzd65". Maybe it was late at night and I was still feebly trying to follow your "object" model? Anyhow, case closed. I thank you for your help and patience. / Roger

________________________________ From: Isaac Muse ***@***.***> Sent: Thursday, April 4, 2024 4:35 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Mention ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) I believe the differences between our respective CAT adaptation stems from the fact that, in my original (naive?) chromatic_adaptation call, I specified the use of the optional 'Bradford' parameter: No, the default IS bradford. See here<https://github.com/facelessuser/coloraide/blob/main/coloraide/util.py#L19>. Let me demonstrate so there is no more confusion:

>> from coloraide import Color >> from coloraide import cat >> XYZ100 = [8.63, 6.24, 16.47] >> XYZ1 = [c / 100 for c in XYZ100] >> xyzd65 = Color('xyz-d65', XYZ1) >> Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords())

[0.08360155956434158, 0.061547933420552856, 0.12397546369244132]

>> Color.chromatic_adaptation(cat.WHITES['2deg']['D65'], cat.WHITES['2deg']['D50'], xyzd65.coords(), method='bradford')

[0.08360155956434158, 0.061547933420552856, 0.12397546369244132] Exact same results because "bradford" is the default. So, why are you getting different results? Let's take a closer look at what you are doing. xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyz_d65, method='bradford') The signature of the function is: Color.chromatic_adaptation(from_white_point, to_white_point, xyz_to_convert) You are giving it a D65 XYZ color, but saying it is a D50 when it is not, and then converting it to D65, which is not what you want. You want this. xyz_d50 = Color.chromatic_adaptation(cat.WHITES['2deg']["D65"], cat.WHITES['2deg']["D50"], xyz_d65, method='bradford') Hopefully, everything is clear now. If there was some source of confusion that led you to believe that the order you were using was correct over what I did, then please do let me know and I will be happy to correct it. — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5A7G3IWTP5XR52ZDTLY3W2RTAVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMJUGM2DC>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[facelessuser]

You tell me which parameter is which from this example, Isaac?

I can't tell if you are directing frustration at me or not. I hope not because I've attempted to add as much documentation as I can to help people, probably more than you will often find in other projects, and I've tried to be patient and help you.

To answer your question, I have an entire explanation as to what is being done leading up the the example. I specify the order before showing the example.

So, we can actually do this manually and compare the results to convert() which automatically handles chromatic adaptation. In order to do this, we need to provide the specified "white point" for the source color and the "white point" for the destination color along with the XYZ coordinates we wish to transform. ColorAide uses the Bradford CAT by default, so we will specify that CAT for consistency.

In the example we start with a xyzd50 variable and send it in as the value to convert and get a xyzd65 variable. It seems clear to me we are converting from D50 to D65. The example call to the function mirrors the description above that mentions the source white point, destination white point, and the XYZ input. The names of the input and output also match the white points being used for the source and the input

>>> from coloraide import cat
>>> xyzd50 = Color('prophoto-rgb', [1, 1, 1]).convert('xyz-d50').coords()
>>> xyzd50
[0.9642956764295677, 1.0, 0.8251046025104602]
>>> xyzd65 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyzd50, method='bradford')
>>> manual = Color('xyz-d65', xyzd65).convert('srgb')

I've spent a lot of time trying to make clear what the library does, granted, what seems obvious to me may not be obvious to others. I will take all of this into consideration when tweaking the documentation in the future.

[Roger-Breton]

Thank you for your kindness and patience. I didn't expect you'd actually be willing to help me clear out MY confusion. I should have picked up on the name of the variables which direction was which 🙁 / Roger

________________________________ From: Isaac Muse ***@***.***> Sent: Thursday, April 4, 2024 8:32 PM To: facelessuser/coloraide ***@***.***> Cc: Roger-Breton ***@***.***>; Mention ***@***.***> Subject: Re: [facelessuser/coloraide] Unable to convert colors? (Discussion #412) You tell me which parameter is which from this example, Isaac? I can't tell if you are directing frustration at me or not. I hope not because I've attempted to add as much documentation as I can to help people, probably more than you will often find in other projects, and I've tried to be patient and help you. To answer your question, I have an entire explanation as to what is being done leading up the the example. I specify the order before showing the example. So, we can actually do this manually and compare the results to convert() which automatically handles chromatic adaptation. In order to do this, we need to provide the specified "white point" for the source color and the "white point" for the destination color along with the XYZ coordinates we wish to transform. ColorAide uses the Bradford CAT by default, so we will specify that CAT for consistency. In the example we start with a xyzd50 variable and send it in as the value to convert and get a xyzd65 variable. It seems clear to me we are converting from D50 to D65. The example call to the function mirrors the description above that mentions the source white point, destination white point, and the XYZ input. The names of the input and output also match the white points being used for the source and the input

>> from coloraide import cat >> xyzd50 = Color('prophoto-rgb', [1, 1, 1]).convert('xyz-d50').coords() >> xyzd50

[0.9642956764295677, 1.0, 0.8251046025104602]

>> xyzd65 = Color.chromatic_adaptation(cat.WHITES['2deg']["D50"], cat.WHITES['2deg']["D65"], xyzd50, method='bradford') >> manual = Color('xyz-d65', xyzd65).convert('srgb')

I've spent a lot of time trying to make clear what the library does, granted, what seems obvious to me may not be obvious to others. I will take all of this into consideration when tweaking the documentation in the future. — Reply to this email directly, view it on GitHub<#412 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AEVBK5HTJDXDDEDVNJNSEWLY3XWIFAVCNFSM6AAAAABFPRWH7KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMJVGY4DQ>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[facelessuser]

No worries. I'm just glad it is cleared up now. Color stuff can be surprisingly complicated at times. I didn't know much about color stuff when I started this project, so I've learned a lot, and most certainly have more to learn.