Hi Aldo,
several developments in THistPainter have been made in the recent versions,
including the CVS head, by Olivier Couet.
Using non equidistant contours and option "colz" or others, you should be able
to do what you describe in your mail without implementing a Histpainter.
Rene Brun
"Aldo F. Saavedra" wrote:
>
> Hi Rene, Axel,
>
> Thanks for the tip. Once I created the dictionary for my
> derived class everything worked fine. I was just not
> sure.
>
> >
> > I am just curious to know why you want to create your own HistPainter?
> > -missing functionality in the default painter?
> > -a feature you do not like?
>
> The only reason I created a new painting object for the
> histogram was because the COLZ option with TH2.
> I wanted the bin of the histogram coloured but the colz
> uses a colour proportional to its value.
> I think this works fine with a big distribution
> of numbers where you can see a trend.
>
> One of my datasets only had 11 values and I wanted the colour
> of the bin be equal to a specific value. The z palette
> would show this relationship.
>
> So I modified the PaintColorLevels function so that
> it would find to which Contourlevel the bin value belonged.
> The colour would then be obtained from GetColorPalette(Contourlevel)
> and use the colour to paint the bin.
>
> The second change was with the z palette. My problem with
> the z palette was that it showed the levels proportional to
> their value. So if you had a big interval and a few small ones.
> only the big interval was the one really visible. This
> was not very helpful when
> for example looking at currents of the chip.
> if I had the following contourlevels
> 0 4 5 5.2 5.4 5.6 5.8 6.0 10 200 . The 10 - 200 colour would
> use most of the palette. But the interesting region is the 4-6mA
> region and you wanted to see which bins in the histogram belonged
> to the 4-6mA. You wanted to see how the 4-6mA changed within
> the histogram.
>
> So I changed PaintPalette so that all the colours would be equally
> spaced no matter their value. But their actual contour value would
> be printed. I include an example.
>
> Cheers,
> Aldo
>
> >
> > Rene Brun
> >
> > "Aldo F. Saavedra" wrote:
> > >
> > > Hi,
> > >
> > > I was wondering if someone could help me with the
> > > following. I found you could set your own
> > > painting class using SetPainter function.
> > >
> > > So in my code I have included the following:
> > > TVirtualHistPainter::SetPainter("TMyHistoPainter");
> > >
> > > The TMyHistoPainter class is derived from
> > > THistPainter as shown in the code below.
> > >
> > > The problem is that I had not being able
> > > to set the name to TMyHistoPainter class
> > > to "TMyHistoPainter" .
> > >
> > > One tactic that I tried was
> > > to let TMyHistoPainter
> > > inherit TNamed which contains functions
> > > that allow you to set the classname.
> > > But during the compilation I had
> > > the message that the delete function
> > > was ambiguous. I think both functions
> > > had the implemented the function.
> > >
> > > So my question is how to name TMyHistoPainter
> > > class "TMyHistoPainter".
> > >
> > > At the moment I'm using 3.05/05 in windows.
> > >
> > > Any ideas would be greatly appreciated.
> > >
> > > // a bit of the code
> > > // definition of the class
> > > #include "THistPainter.h"
> > > #include "Rtypes.h"
> > >
> > > class TMyHistoPainter: public THistPainter
> > > {
> > >
> > > public:
> > > TMyHistoPainter();
> > >
> > > ClassDef(TMyHistoPainter,0)
> > >
> > > };
> > >
> > > // implementation of the class
> > >
> > > ClassImp(TMyHistoPainter)
> > >
> > > TMyHistoPainter::TMyHistoPainter():THistPainter()
> > > {// the constructor
> > >
> > > }
> > >
> > > ***********************************************************
> > > Aldo F. Saavedra
> > > ATLAS Collaboration - Pixel Group (Physics Division)
> > > Lawrence Berkeley National Laboratory
> > >
> > > Email: AFSaavedra@lbl.gov
> > > Phone: 1 510 486 4690
> > > Fax: 1 510 486 5101
> > > Postal Address: Lawrence Berkeley National Laboratory
> > > 1 Cyclotron Rd Mailstop 50B-62222
> > > Berkeley, CA 94720
> > > USA
> > > ***********************************************************
> >
>
> ***********************************************************
> Aldo F. Saavedra
> ATLAS Collaboration - Pixel Group (Physics Division)
> Lawrence Berkeley National Laboratory
>
> Email: AFSaavedra@lbl.gov
> Phone: 1 510 486 4690
> Fax: 1 510 486 5101
> Postal Address: Lawrence Berkeley National Laboratory
> 1 Cyclotron Rd Mailstop 50B-62222
> Berkeley, CA 94720
> USA
> ***********************************************************
This archive was generated by hypermail 2b29 : Thu Jan 01 2004 - 17:50:12 MET