This is a forward of a message from Jens Scheidtmann relevant to this
---------- Forwarded message ----------
> > Best would be (if you want to have them similar) to place a reference
> > point in CIE L*a*b*, where euclidian distance is proportional to
> > percepted color difference and place 5 - 11 colors (odd) around that
> > reference point. Then enumerate these points so that the sequence has
> > maximum color differences (for 5 points --- hope you don't use a
> > proportional font):
> > 1
> > 4 . 3
> > 2 5
> > (You should do that in three dimensions taking luminance into account
> > ;-)
> Jens, thank you very much for the information.
> Some experts are of the opinion that the euclidian distance between
> colours in CIE L*a*b* is bogus when it comes to measuring perceptually
> uniform colors.
> Here's an example:
> Care to comment on exactly *how* bogus it is?
Well, it's not a very good idea to have these color patches overlaid
on a photograph. There's a well known effect that color vision
depends on the background / context.
cf. e.g. for brightness:
e.g. for color (have a look at pictures 3 and 4 counted from the bottom):
In addition color perception depends on the ambient light. Have you
ever tried to take a photograph in an aquarium? Wandering through the
roooms you will have natural color perception, but if you do not
adjust the white point of your camera you will get really, really
tinted photographs (blueish --- off-color).
A friend of mine has a small agency for promotions, and when he's looking
at color proofs in the print shop, they will use a room with special
(sun-like) illumination, so that they can really compare the original
and the print-outs. There they have some really expensive color
correct monitors, too...
IANA Color Expert but I presume the dE* measure has been developed in
a carefully controlled environment, which apperently was such, that
the environment allowed to spot minimal differences.
> Also, do you know of any colour-picker program suitable for choosing
> colors with high color-difference but a low cataracts factor that you
> could recommend for the job at hand?
Unfortunately, no. All color pickers I have found try to give
visually pleasing color schemes for websites. If you have a look at
color picker controls, those mostly use the RGB or HSV schemes, so
are not of any use in our problem.
I even tried to use some of those CIELAB -> RGB formulas once. But
there you run into the problem, that values evenly spaced in CIELAB
all convert to the same RGB triple for large areas of the color space
(and I didn't find any information on what the allowed domain is for
CIELAB on the net).
From the website you mention above it seems that Photoshop is able to
use a Lab color space, maybe it comes with a color picker for cielab?
> > BTW, if you have a look at
> > http://en.wikipedia.org/wiki/Lab_color_space you'll see that green is
> > not a good choice, because the eye does not differentiate bright green
> > colors well. (See http://www.cs.rit.edu/~ncs/color/t_convert.html for
> > conversion formulas to RGB)
> Well, I like the "use shades of gray for fore-color" idea just as
> well, seeing as it would keep the background color for unchanged lines
> the same as is used in the diff windows (white).
> How's that? Better or worse?
> > But maybe the problem does not lie in the colors but in the algorithm:
> > The algorithm has no notion of "relevant changes", it treats
> > all substrings the same. Maybe something like a visualization of
> > minimum edit distance (Levenshtein distance) would be more
> > appropriate, but I haven't found good references on that yet (only the
> > calculation of minimum edit distance).
> Brilliant idea.
> Let's spare ourselves the trouble and not do that and just say we did :o).
> Oh, and a last question, out of pure interest.
> You seem to do some professional diff'ing - are you actually using
> TMerge for it, or do you use some other solution?
Well, as I am mostly concerned with character files, I haven't delved
into the evaluation of the zillion of different diff programs
available on the net (or commercially).
As there also is a fuzziness in the diff algorithm applied (regarding
the allowed operations and in addition heuristics are used to gain
performance, with respect to the minimum length diff, see
http://c2.com/cgi/wiki?DiffAlgorithm ), I sticked with the tool which
uses the algorithm of svn.
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Wed Mar 23 11:35:03 2005