TH2D: cont0 vs cont1

At root.cern.ch/root/html/THistPainter.html it is stated that
"CONT0" Draw a contour plot using surface colors to distinguish contours.
“CONT1” Draw a contour plot using line styles to distinguish contours.
I have tested both (as well as CONT2,3,4,5). Actually I found them very similar (not identical) in the graphic outcome, although the behaviour is quite different. First my piece of code, below the question:

int xChannels=4096;
int yChannels=1024;
Int_t n_col=10;
TH2D *hpha = new TH2D("Sep",pha_char,
               xChannels,0,xChannels-1,yChannels,0,yChannels-1);

  for (Int_t jx = 0; jx < xChannels; jx++)
    {
    for(Int_t jy = 0; jy < yChannels; jy++)
      {
      hpha->SetBinContent(jx,jy, SepMatrix[jy*xChannels + jx]);
      }
    }

  gStyle->SetPalette(1);
  gStyle->SetOptStat(0);

  hpha->SetMarkerStyle(2);
  hpha->SetMarkerSize(5);
  hpha->SetContour(n_col, contours);
//  hpha->Draw("CONT0 LOGZ"); // test case: 42 s for Canvas update, 870 kB eps
  hpha->Draw("CONT1 LOGZ");    // test case: 2s for Canvas update, 5.2 MB eps

Please note the matrix size, 4096x1024.
If I use CONT0 on a test case, I get a very slow updating of the canvas (~ 42 s), but storing an EPS file from the canvas is efficient (small size, quick process). If I use CONT1, the screen output is ready in (~ 2 s), but the EPS storage becomes lengthy and large in size. Is that a common experience? Should I improve my code in some way to combine both beneficial effects? My idea is to insert a checkbutton “EPS output” in the main GUI frame.

Please note the matrix size, 4096x1024.

what is your canvas size ? if you are using the standard one ~500x500 you will have 8 cells per pixels !
not point to use the CONT option. Better use The COL option in that case.

600x600
and yes, COL does look at least as good as CONT, while being way faster! The EPS file of the test case is 2 MBytes, but that’s not a big deal.
Thanks, that’s solved!

Yes … so your histogram being 4096x1024 you have more than 8 bins per pixel.
There is no point using a smoothing representation like CONT when the
definition of your histogram is 8 bigger than the plot resolution.
The COL representation will do the job more efficiently.