Moving color overhaul forward

Development matplotlib-devel
1

Hey all,

Today we had a phone call with myself, Eric Firing, Micheal Droettboom, Stéfan van der Walt, and Nathaniel Smith to discuss the path forward for the changes to the default color map / style. The notes are below:

  • re-order feature release/style change if needed

  • can focus sprint effort at scipy on new features

  • release order may be 2.0 -> 2.1 or 1.5 -> 2.0

  • keep style change-only release plan

  • start adding color maps as named maps on master

  • color map

  • D seems to be leader, maybe variation on theme

  • stefan is working on circular color maps

  • other style changes

  • adopt most of seaborn as defaults ?

  • start putting in style sheets as soon as possible

  • may not be worth big drawn out decision, just change them

  • line color cycle

  • need to do something, maybe related to circular color maps

  • use something from seaborn

  • get time for style BoF at scipy

Any feed back is welcome.

Tom

2

It’s good to hear Stefan is working on circular colormaps as the hsv colormap is way too saturated, ignoring any perceptual aspects. For this, it might be possible to provide some sort of saturation parameter in a function along the lines of the parameters in the cubehelix function? Also, sympy/mpmath’s cplot function allows you to modulate the nominally-perceptually-flat hsv map with a value-related function, which they take as the absolute value of the function, but ideally it would be any function or array that could ideally be passed in as a parameter. I have plotted images like this in the past, where I was plotting the phase of a complex number field but only wanted to use the envelope of the overall function to control the saturation so you could see the structure near the singularities but the brightness would fall off in the outer parts of the image.

Gary R

···

On 10 June 2015 at 09:17, Thomas Caswell <tcaswell@…714…> wrote:

Hey all,

Today we had a phone call with myself, Eric Firing, Micheal Droettboom, Stéfan van der Walt, and Nathaniel Smith to discuss the path forward for the changes to the default color map / style. The notes are below:

  • re-order feature release/style change if needed
  • can focus sprint effort at scipy on new features
  • release order may be 2.0 → 2.1 or 1.5 → 2.0
  • keep style change-only release plan
  • start adding color maps as named maps on master
  • color map
  • D seems to be leader, maybe variation on theme
  • stefan is working on circular color maps
  • other style changes
  • adopt most of seaborn as defaults ?
  • start putting in style sheets as soon as possible
  • may not be worth big drawn out decision, just change them
  • line color cycle
  • need to do something, maybe related to circular color maps
  • use something from seaborn
  • get time for style BoF at scipy

Any feed back is welcome.

Tom

Matplotlib-devel mailing list

Matplotlib-devel@lists.sourceforge.net

https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

3

Great to hear this!

- re-order feature release/style change if needed
   - can focus sprint effort at scipy on new features
   - release order may be 2.0 -> 2.1 or 1.5 -> 2.0
- keep style change-only release plan
- start adding color maps as named maps on master
- color map
   - D seems to be leader, maybe variation on theme
   - stefan is working on circular color maps
- other style changes
   - adopt most of seaborn as defaults ?

I think it would be good to have a set of core stylesheets in mpl that include

* Slight modifications of the existing mpl theme (outward ticks?,
different default cmap and color cycle)
* The default seaborn style that closely follows that of ggplot2
* Stylesheets that mimic the "contexts" of seaborn

I would also like to see a few more style helper methods such as
despine, ticks_out/in, etc.

   - start putting in style sheets as soon as possible
   - may not be worth big drawn out decision, just change them
   - line color cycle
     - need to do something, maybe related to circular color maps
     - use something from seaborn
- get time for style BoF at scipy

I would love to participate in any mpl BoFs at SciPy. Also, I have a
student working for me this summer and one of his focus areas will be
visualization. I can likely have him work on some of the
stylesheet+traitlets+nbagg stuff as part of this work. He will also be
at SciPy.

Cheers,

Brian

···

Any feed back is welcome.

Tom

------------------------------------------------------------------------------

_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel@lists.sourceforge.net
matplotlib-devel List Signup and Options

--
Brian E. Granger
Cal Poly State University, San Luis Obispo
@ellisonbg on Twitter and GitHub
bgranger@...882... and ellisonbg@...149...