I seldom post, being content mainly to lurk:) I wouldn't know how to move it to the chat forum but please feel free to do so...
I see it as being more about perceptual linearization. The geographical images simply *illustrate* how 'people' (i.e. we) naively colourise images and (a) make them ugly and (b) lose information. Every day I see poorly colourised radiographic images, scientific illustrations (e.g. in neurophysiology), and indeed maths that impair the "readability". I think the PRAVDAcolour chaps nailed the solution. It's less about maths and more about communication, but then what is maths if the communication is inadequate? Just IMO, you understand. Jo. On 15 April 2017 at 07:09, Linda A Alvord <[email protected]> wrote: > Andrew, These are fascinating images. > > Several years ago I tried to convince people that GRB would be a better > palette than RGB. I used Cliff's image, your first pattern, and rotated it > 90 degrees. If you think of this as the face of a clock, then in an hour > the GRB palette goes from white to black in a continuous value from light > to dark when seen as a grayscale. Green ns brighter than Red, which is > brighter than blue. > > require 'viewmat' > at2=: 13 :'([:{:"1 *.) j./"1 y' > RGB=: 255* #:i.8 > GRB=:1 0 2{"1 RGB > we=: 13 :'<.0++/"1[0.3 0.59 0.11*"1 y' > gray=: 13 :'3#"0 we y' > RGB viewmat (D=.at2"0 /~ i:500) > GRB viewmat (D=.at2"0 /~ i:500) > (gray RGB) viewmat (D=.at2"0 /~ i:500) > (gray GRB) viewmat (D=.at2"0 /~ i:500) > > (Another little exercise for JHS) > > Maybe you'll make some interesting discoveries with your other patterns. > > Linda > > -----Original Message----- > From: Programming [mailto:[email protected]] On > Behalf Of Jo van Schalkwyk > Sent: Friday, April 14, 2017 6:18 AM > To: [email protected] > Subject: Re: [Jprogramming] Complex functions visualization > > Hi All > > I think there's room for further improvement of your colorization of the > images. Years down the line I still think the approach taken with > PRAVDAcolor makes a lot of sense. Check out: > > https://www.research.ibm.com/people/l/lloydt/color/color.HTM > > Bergman's paper is here: > http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1. > 452.1807&rep=rep1&type=pdf > > > Is this of any value to you? > > My 2c, Jo. > > On 14 April 2017 at 17:27, Linda A Alvord <[email protected]> wrote: > > > Andrew, These are fascinating images. > > > > Several years ago I tried to convince people that GRB would be a > > better palette than RGB. I used Cliff's image, your first pattern, > > and rotated it > > 90 degrees. If you think of this as the face of a clock, then in an > > hour the GRB palette goes from white to black in a continuous value > > from light to dark when seen as a grayscale. Green ns brighter than > > Red, which is brighter than blue. > > > > require 'viewmat' > > at2=: 13 :'([:{:"1 *.) j./"1 y' > > RGB=: 255* #:i.8 > > GRB=:1 0 2{"1 RGB > > we=: 13 :'<.0++/"1[0.3 0.59 0.11*"1 y' > > gray=: 13 :'3#"0 we y' > > RGB viewmat (D=.at2"0 /~ i:500) > > GRB viewmat (D=.at2"0 /~ i:500) > > (gray RGB) viewmat (D=.at2"0 /~ i:500) (gray GRB) viewmat (D=.at2"0 /~ > > i:500) > > > > (Another little exercise for JHS) > > > > Maybe you'll make some interesting discoveries with your other patterns. > > > > Linda > > > > -----Original Message----- > > From: Programming [mailto:[email protected]] On > > Behalf Of Andrew Nikitin > > Sent: Tuesday, April 11, 2017 11:39 AM > > To: J programming > > Subject: [Jprogramming] Complex functions visualization > > > > I read a book "Visual Complex Functions" by Elias Wegert. > > > > In it autor argues for so called phase portraits as a good way to > > visualize complex function of one complex variable. Each function > > value is represented by a pixel with a hue (which is an angular > > quantity) equal to function value's phase angle. He argues that the > > result provides a lot of information about function and even allows to > > restore analytic functions (up to a constant). > > > > The other 2 components of the colorspace, saturation and light, can be > > used to show lines of equal phase and equal magnitude. Author calls it > > "enhanced phase portrait". Interesting, that there is no level tracing. > > Lines appear as a byproduct of using a modified hue palette. > > > > I put up a script and couple of sample images on wiki. 'sq' utility > > generates unit square of complex numbers. Evaluate your choice of > > function on it and color each pixel with ccEnhPh and you have yourself > > a phase portrait to view with viewmat or save with writebmp. > > > > http://code.jsoftware.com/wiki/User:Andrew_Nikitin/Phase_portraits > > > > I think that if you like pretty pictures (and want to get some insight > > on complex function behavior), this technique provides a lot of bang > > for a very little buck. > > > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
