K-Means palette

Comment

Comment by David Rutten on July 14, 2016 at 1:15pm

Have you seen Ivy?

I have not. 

Comment by David Rutten on July 14, 2016 at 1:13pm

More easy with parallel view !

Yup, that's why I used parallel :)

Comment by Laurent DELRIEU on July 14, 2016 at 11:04am

... Yes more simple than I thought but  black spheres must be behind exactly from colored ones. So in perpective view not easily done. More easy with parallel view !

Comment by taz on July 14, 2016 at 10:26am

...Looks like [Dot Display] with 2 dots overlayed on eachother.

Comment by Laurent DELRIEU on July 14, 2016 at 10:20am

How is done the black contouring/silhouette on colored disk ? Make2d in Rino 6 ?

Comment by taz on July 14, 2016 at 10:05am

Ah, actually I could have answered my own question if I just looked at your definition, but I can now see how you translated the RGB values into a unique set.

Have you seen Ivy? I noticed it when it came out fairly recently, but haven't played with it.

http://www.grasshopper3d.com/video/k-means-clustering-with-ivy

Comment by David Rutten on July 14, 2016 at 9:46am

Data are stored as n-dimensional points. Whenever you want to put data into a sampling set you must provide a delegate which converts your data type into a PointN structure. At present, straight up euclidean distance is used to compare n-points and build k-d trees, k-means clusters, and other types of groups/clusters/maps.

The aim here is to create as generic a framework as possible to analyse data sets.

Comment by taz on July 14, 2016 at 9:35am

K-means-Kool!

What did you use as unique identifier for the combined RGB parameters? Isn't that the tricky bit?

Comment by David Rutten on July 14, 2016 at 5:33am

And another run, this time using all three colour channels.

Comment by David Rutten on July 14, 2016 at 5:07am

Creating palettes containing between 1 and 32 colours sampled from the original image.