CONT vs CONT4

ROOT
1

Dear all, but specially Olivier!

I know this CONT discussion is an old one - but I just stumbled on yet another example where the CONT4 algorithm is far superior to the CONT algorithm when it comes to showing data of a 2D histogram (see images). So I would much prefer replacing CONT with CONT4 everywhere. But this still comes at a price: since CONT4 is at its core a 3D algorithm, the canvas then has this awkward 3D behaviour, which makes it impossible, for instance, to zoom into the axis.

So I have a choice: either give the user the superior CONT4 algorithm and deprive them of the zoom and other functionality, or give him/her the inferior CONT(0) algorithm while providing the full canvas functionality…

Isn’t there a way to turn CONT4 into a real 2D algorithm? Once you have the contours, you can just draw them in regular 2D? Olivier, can you perhaps reconsider this?

Many thanks in advance!!

Thomas
testCONT.cxx (97.4 KB)

COL.png796×606 7.45 KB


CONT4.png796×606 12 KB


CONT0.png796×606 12 KB

2

Hi Thomas,

Yes CONT4 is better we know that. And it need to be tune is a way to be better integrating in 2D. It is not a trivial job and up to now I had always something more urgent to do. In any case my feeling is that for the data representation point of view the COLZ option is much better because it represent the data "as they are’ without any interpolation between bins. Any way you are right I will "try’ to find time to make it better …

Olivier

3

Note: to zoom an histogram plotted with CONT 4 you can open the editor for TH2 click on “binning” tab and change the axis range manually … but may be that’s not so convenient in your application.

4

With the latest version, in the svn trunk, it is now possible to do interactive zooming with option CONT4. The modified files are:

hist/histpainter/src/THistPainter.cxx 
graf2d/gpad/src/TPad.cxx