Build 10 games and learn game development along the way in this fast paced game development course

Get the source code of 12 commercial games, loaded 50+ million times, and learn the secrets of game design

Learn how to make a successful Flash game from a real world example. Fully commented source code

Create realistic Flash physics games from scratch with advanced features, from Angry Birds to Crush the Castle

Color difference algorithm

I will probably write some posts and tutorials about this topic, becaue color differences are very useful when designing a color scheme or some elements to be placed on a given background color.

And for a wide range of applications working with color spaces.

Anyway, at the moment just know Wikipedia says there is a International Commission on Illumination (CIE) who calls distance between colors dE.

Obviously the meaning is removing subjectivity… I can see two colors “almost identical” while you can see the same ones “slighty different”.

In theory, a dE of under 1.0 is supposed to be indistinguishable unless the samples are adjacent to one another

To make things a little bit harder, there are different ways to calculate color difference… three formulas the CIE refined during the years… CIE76, CIE94 and CIEDE2000.

If you want to study in depth color difference, or even small color differences may be critical for your work, then the CIEDE2000 is the most advanced way to determine color difference, and the one with the most complex formula.

Complex formulas aren’t what I am looking for because I have in mind something a little bit complicated than just measuring color difference, and the CIE76 is maybe too basic, so at the moment I am testing the CIE94 formula.

Before examining the formula, let me introduce you some ways to define colors.

If you are used to web design, you’ll probably know the RGB color model. The name of the model comes from the initials of the three additive primary colors, Red, Green, and Blue, that are added to reproduce all colors.

But RGB is not the only way you can code colors.

Color models

Here it is a brief list of color models… some of them are useful, some others not that useful, anyway let’s see them…

RGB: The most famous color encoding, as explained before

CMY: The complement of RGB, named after Cyan, Magenta and Yellow.

CMYK: Same as CMY with the added K parameter which is blacK. It is the system used by printers.

HSV or HSB: A system which represents color as Hue, Saturation and Value. Hue is what is seen as the color. Saturation is how “pure” is the color. 0% is grey and 100% is a pure color. Value represents the Brightness.

HSL: Just like HSV but instead of Value we have lightness. This time, maximum saturation is at the mid-point (50%). 0% is black and 100% is white.

XYZ: This is also called “norm color system” and it’s a superset of RGB. It uses tristimulus values when encoding. X, Y and Z are all calculated through color-matching functions and are always positive.

The human eye has receptors (called cone cells) for short (S), middle (M), and long (L) wavelengths. Thus in principle, three parameters describe a color sensation. The tristimulus values of a color are the amounts of three primary colors in a three-component additive color model needed to match that test color.

Yxy: Known as the Trichromatic coordinates. Y represents the brightness and (x,y) hue and saturation.

CIE Lab: This color scale is based on the opponent color theory. L represents light-dark, a represents red-green and b represents yellow-blue.

There are some more systems, anyway, at the moment we are interested in Lab color space that unlike the RGB and CMYK color models, is designed to approximate human vision.

Since we are used to RGB (and eventually to CYMK if you work in print), we need to convert these formats into Lab color space, and in order to do it we must first convert them in XYZ color space, and then convert them again to Lab values.

RGB -> XYZ conversion

On the Standard Default Color Space for the Internet page on the “Colorimetric definitions and digital encodings” chapter you will find the matrix used for color conversion and the previous adjustments to do.

So, having a RGB value with red, green and blue values as base 10 integers from 0 to 255, you can convert RGB to XYZ this way:

Now, next step:

XYZ -> Lab conversion

In order to convert XYZ to Lab, first we must set the white point, that is a set of tristimulus values or chromaticity coordinates that serve to define the color “white” in image capture, encoding, or reproduction.

Then, following the formula found on this page, we can solve the problem this way:

And in the end, the orginal RGB 39,176,165 is converted in Lab 65.12, -37.39, -4.54 that is very close to Photoshop CS4 Lab 65,-38,-5 that probably uses a different white point.

Next time we’ll see something interesting to do with these formulas.

Rate this post: 1 Star2 Stars3 Stars4 Stars5 Stars (27 votes, average: 4.19 out of 5)
Loading ... Loading ...
Learn how to make a successful commercial Flash game from a real world example: get the fully commented source code of Globez, a Flash game played millions of times which generated a four figure income. Limited copies available.

Get it now
Be my fan on Facebook and follow me on Twitter! Exclusive content for my Facebook fans and Twitter followers

This post has 13 comments

  1. Monkios

    on August 31, 2009 at 4:17 pm

    Nice article. Pretty usefull in understanding color models.

    Just saying but … you could use more functions.

  2. Weekly Shared Items – 1. September, 2009 | TOXIN LABS - weblog of a german design student from wuerzburg

    on September 1, 2009 at 6:04 am

    [...] Color difference algorithm [...]

  3. Color difference algorithm – part 2 : Emanuele Feronato

    on September 8, 2009 at 5:24 pm

    [...] I am turning the scripts explained at Color difference algorithm into functions, this way: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 [...]

  4. pic2col – extract color schemes from a picture : Emanuele Feronato

    on September 11, 2009 at 12:56 pm

    [...] wondered the point in calculating color distance at Color difference algorithm and Color difference algorithm – part [...]

  5. Jon Rista

    on February 25, 2011 at 11:04 pm

    Excellent article. Very helpful, especially to find some useful code examples that demonstrate these conversions.

    A question and a note. First, what white point did you use for your calculations? I know that Photoshop’s color matching algorithms are balanced on a D50 white point. As for a simple note, I am doubtful that the difference between your final numbers and Photoshops are due to a white point difference. It looks like a simple rounding to me: 65.12 -> 65 (round down, valid), -37.39 -> -38 (round up, rather than round down, but that could be due to precision differences in the calculation), -4.54 -> -5 (round up, valid).

    Again, thanks for the example code.

  6. sabr

    on September 12, 2011 at 4:16 pm

    I think the $z color is far less than that it would be …
    Cordially.

  7. sabr

    on September 14, 2011 at 11:52 am

    In fact, In RGB 2 XYZ, I don’t understand why we have to adjust values before applying the matrix ?
    Could someone explain me ?

  8. Anthony Thyssen

    on October 21, 2011 at 4:51 am

    You do not explain of the in put RGB values are linear RGB, sRGB or RGB with a 2.2 gamma.
    You formulas seems to assume linear RGB, but typical (non-mathematical) images are more usually sRGB. The difference is important and quite massive!

  9. How to Choose Colours Procedurally (Algorithms) » devmag.org.za

    on July 29, 2012 at 3:09 pm

    [...] http://www.emanueleferonato.com/2009/08/28/color-differences-algorithm/ [...]

  10. Xterm 256 colors??????? | SanRin?

    on August 21, 2012 at 5:42 pm

    [...] Color difference algorithm – Emanuele Feronato – rgb??XYZ?Lab?????PHP????? [...]

  11. Filipe

    on February 14, 2013 at 7:05 pm

    Thanks a lot, :)

  12. Lukasz Krawczyk blog | Colour detection algorithm

    on February 24, 2013 at 6:03 pm

    [...] http://www.emanueleferonato.com/2009/08/28/color-differences-algorithm/ http://stevehanov.ca/blog/index.php?id=116 http://en.wikipedia.org/wiki/Color_difference http://homepages.inf.ed.ac.uk/rbf/PAPERS/iccv99.pdf http://www.cs.cmu.edu/~har/visapp2006.pdf http://research.cs.wisc.edu/vision/piximilar/ http://mattmueller.me/Piximilar/paper.pdf http://mattmueller.me/blog/creating-piximilar-image-search-by-color [...]

  13. Comparison of colours in Actionscript 3 | Code Lab

    on August 1, 2013 at 11:15 pm

    [...] that compare in the best way colours I really suggest to work with this color space and read this post on this topic, in my case I need something in the middle so a colour space that could guarantee the perfect trade [...]

Thank you for the download!!

I hope you will find it useful and make something interesting out of it

To keep up to date with the blog, why don't you like my Facebook page and follow me on Twitter?

Want to learn more? Don't miss this:

Learn how to make a successful commercial Flash game from a real world example: get the fully commented source code of Globez, a Flash game played millions of times which generated a four figure income. Limited copies available.

Get it now

×