Opacity issues with EPS

Hi all,

I've recently been working on a script in which I'm attempting to fill plot
data overlaid above a couple of contour plots (one plain contours, one
stipple-filled) and some polymarkers while also below some other
polymarkers and several thousand polylines. However, it seems that the
successful completion of this script depends on the cnFillType and one of
the opacity levels chosen for the fill, and further the EPS output is
sometimes corrupt/unreadable.

In order to test which part of the resources/plotting was causing issues, I
tried a number of scenarios, mainly changing the cnFillType and what kind
of opacity settings were present. All of these test were output to EPS, as
direct output to PNG seems to work fine. What I deduced is that RasterFill
does not seem to handle any partial opacities when output to an EPS
(RasterFill works fine with partial opacities when output to PNG), usually
resulting in a seg fault at the 'draw' stage later in the script. CellFill
seems to work fine with some opacity, and AreaFill, despite being time
consuming with my data, plots things fine with partial opacities but NOT
when plotting an item of 0 opacity to an EPS. The 0-level opacity seems to
create something that image software ImageMagick and Ghostview can't
process, despite the file size generally indicating that all the plotted
information is within.

For the record, I have been able to work around this with the
higher-resolution PNG driver, but it would still be nice to have the
complete transparency abilities with EPS as well.

Although I'm not providing the complete script for now, I'll include the
resources used in this troublesome plot, and highlight the ones that seem
to play a role in creating the errors. In case the bolded plain text does
not come through, those lines are re-pasted at the top of the text below.

If my descriptions have been too vague, I can work up a sample script that
will hopefully reproduce the problems. I may also be able to try and share
the relevant data and script for those who would like to see it. Thanks,

Kyle

* **irres@cnFillMode = "AreaFill"*

* colors(:,3) = 0.6** colors(0,3) = 0*

  irres = True
  irres@gsnDraw = False
  irres@gsnFrame = False
  irres@gsnLeftString = " "
  irres@gsnRightString = " "
  irres@cnInfoLabelOn = False
  irres@lbTitleFontHeightF = .009
  irres@lbLabelFontHeightF = .009* irres@cnFillMode = "AreaFill"* ;
otherwise will take forever
  irres@gsnAddCyclic = False ; data already has cyclic point

  ;***********************************
  ; IR contour/fill irresources
  irres@lbBoxLinesOn = True
  irres@lbLabelBarOn = False
  irres@lbLabelStride = 4
  irres@pmLabelBarHeightF = 0.08
  irres@pmLabelBarOrthogonalPosF = 0.07
  irres@cnFillOn = True
  irres@gsnSpreadColors = True
 ; irres@gsnSpreadColorStart = 1
 ; irres@gsnSpreadColorEnd = 32
  irres@cnLinesOn = False
  irres@cnLineLabelsOn = False
  irres@cnLevelSelectionMode = "ManualLevels"
  irres@cnMinLevelValF = 190.
  irres@cnMaxLevelValF = 270.
  irres@cnLevelSpacingF = 2.5

  colors = spread_colors_rgba("WhiteBlueGreenYellowRed",0,252,irres)
; colors = spread_colors_rgba("grayscale_rgba",0,253,irres)*
colors(:,3) = 0.6** colors(0,3) = 0*
  irres@cnFillColors = colors(::-1,:)

----------------------------------------
Kyle S. Griffin
Department of Atmospheric and Oceanic Sciences
University of Wisconsin - Madison
1225 W Dayton St, Madison, WI 53706
Room 1421 Email: ksgriffin2@wisc.edu

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Sep 12 12:20:58 2012