Re: Plotting high resolution data

Hi Matthew,

I was just reminded of the fact that in many cases, plotting with the
triangular mesh algorithm can speed up the plotting significantly with
no adverse effect on the quality of the plot. This is just a matter of
setting trGridType to "TriangularMesh". You do not need to have data
on a triangular mesh or 1D scattered data to use it. It can be
particularly useful when using 2D (topologically rectangular)
coordinate arrays.
  -dave

On Mar 17, 2010, at 11:52 AM, Matthew Janiga wrote:

> David,
>
> Thank you very much for you help cnRasterSampleFactorF seems to make
> the biggest difference in the quality of the plots.
>
> Cheers,
> Matt
>
> On Wed, Mar 17, 2010 at 5:47 PM, David Brown <dbrown@ucar.edu> wrote:
>> Hi Matthew,
>>
>> It is true that AreaFill can be very slow with high-resolution data
>> and/or
>> many contour levels. However, unless you need part of the gridded
>> data area
>> to be transparent you can make RasterFill look almost identical by
>> setting
>> cnRasterSmoothingOn to True. Of course, a RasterFill plot is
>> essentially a
>> pixel-based plot rather than a vector plot so the pixelation will
>> be evident
>> if you enlarge it or zoom into a section. Should you need a very high
>> resolution plot there are several resources that control the
>> ultimate pixel
>> size, including cnRasterMinCellSizeF , cnRasterCellSizeF,
>> cnRasterSampleFactorF, and depending on the output workstation,
>> wkWidth,
>> wkHeight, wkPSResolution, or wkPDFResolution.
>>
>> All that said, we are aware that there are faster vector contouring
>> algorithms than the routines used by NCL. A lot of the reason for
>> that is
>> the generality of the code, especially its ability to mask out
>> specific
>> areas for labels, etc. We have discussed the possibility of
>> incorporating
>> new and faster contouring methods, and hopefully that will be
>> possible
>> before too long. To be honest, however, there are a number of
>> things on the
>> to-do list that have to be accomplished before we can turn our
>> attention to
>> this issue.
>> -dave
>>
>>
>> On Mar 16, 2010, at 9:23 AM, Matthew Janiga wrote:
>>
>>> NCL-Talk,
>>>
>>> This is a fairly general question. I've noticed that when plotting
>>> high resolution data (for instance several km resolution satellite
>>> data) that NCL takes a very long time to produce a color-filled
>>> contour plot (cnFillMode = AreaFill); a raster plot will be made
>>> fairly quickly (cnFillMode = RasterFill) but don't look as nice. I
>>> ask
>>> this question because ncview can draw the color-filled contour plots
>>> quickly enough to animate the dataset like a movie. Is this simply a
>>> limitation of NCL or postscript output? Has anyone else ran into a
>>> simillar problem and/or is this an issue that may be addressed in
>>> future versions of NCL.
>>>
>>> Cheers,
>>> Matt
>>>
>>> --
>>> **********************************
>>> Matthew A. Janiga
>>> Department of Earth and Atmospheric Sciences, ES 330
>>> University at Albany / State University of New York
>>> 1400 Washington Avenue
>>> Albany, NY 12222
>>> PH: (847) 962-8397
>>> Email: janiga@atmos.albany.edu
>>> **********************************
>>> _______________________________________________
>>> ncl-talk mailing list
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>
>>
>
>
>
> --
> **********************************
> Matthew A. Janiga
> Department of Earth and Atmospheric Sciences, ES 330
> University at Albany / State University of New York
> 1400 Washington Avenue
> Albany, NY 12222
> PH: (847) 962-8397
> Email: janiga@atmos.albany.edu
> **********************************

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Mar 17 12:04:24 2010