[R] graphic output file format

Jeffrey Horner jeff.horner at vanderbilt.edu
Fri Aug 11 17:10:21 CEST 2006


Luiz Rodrigo Tozzi wrote:
> Hi
> 
> I had some problems using GDD, especially with colors and with some advanced
> plottings.
> 
> In my case I didnt try the Cairo, I just got back to the png() using Xvfb
> (its almost 3 or 4 times slower than my first tries qith GDD, but using GDD
> would cause me to rewirte some parts of my script)
> 
> Since my Xvfb "came back to normal" in cron mode, I didnt use the Cairo, but
> mr urbanek really suggested it to me.

I've had a little experience with both GDD and Cairo:

GDD has served my needs well when drawing thin lines and pixels, 
although, the current implementation has problems with the background 
color and the line width (lwd) not being honored. I've hacked around a 
bit, and I can now set the background correctly, but I need to clean up 
the patch to submit to Simon. Also, libgdd rasterizes pngs very well. If 
you're interested in examples, check out the png images here:

http://biostat.mc.vanderbilt.edu/DiffGraphics

For the same needs, Cairo does very poorly rasterizing for fine pixel 
detail; it's not so much libcairo's fault as much as the package Cairo 
not being able to handle this, if it can at all. A better explanation of 
this issue is here: http://cairographics.org/FAQ. Read the first 
question: "Why does my 1-pixel wide horizontal/vertical line come out 
fat and blurry (eg. 2 pixels wide and half-intensity)?"

I'm not sure why you're having trouble with colors in GDD; I've never 
run into this issue.


> I dont know if you have something similar to Xvfb in Mac OSX
> 
> Tozzi
> 
> 2006/8/10, erick at nospammail.net <erick at nospammail.net>:
>>
>> I received the following communication from the package maintainer
>> for GDD (Simon Urbanek) when I was having compile problems on linux...
>>
>>>> gd.x86_64                                2.0.28-4.4E.1
>>>> installed       Matched from:
>>> ^^
>>>  +--- your GD is too old. You'll need at least gd-2.0.29 (see README
>>> in GDD) and 2.0.33 is recommended.
>>> You may want to try Cairo (see the nightly page) instead of GDD,
>>> because cairographics has some features that GD is lacking and it is
>>> actively maintained (unfortunately libgd is pretty much dead).
>>>
>>> Cheers,
>>> Simon
>> Disclaimer: I actually cannot get the Cairo to work either!
>> Eric
>>
>>
>>
>> On Thu, 10 Aug 2006 19:00:08 +0200, "Stefan Grosse"
>> <stefan.grosse at uni-erfurt.de> said:
>>> Please mail always also to the list, since it may be that there is
>>> someone who could reply better and/or faster.
>> [Case in point.]
>>> Especially in this case since I am only experienced in M$ Windows and
>>> Linux so I am unable to guess whats wrong with OSX. does not even the
>>> postscript device work? With postscript I get the best quality for
>>> printing.
>>>
>>> e.g.:
>>> postscript("test.eps", width = 8.0, height = 6.0, horizontal = FALSE,
>>> onefile =FALSE, paper = "special")
>>> curve(x^2)
>>> dev.off()
>>>
>>>
>>> Conan Phelan schrieb:
>>>> Thanks for the feedback and link Stefan,
>>>>
>>>> Perhaps if I provide a little more background. (Pls
>>>> forgive both my ignorance and my long windedness.)
>>>>
>>>>  I am on Mac OSX and am only able to load the Quartz
>>>> graphic device (is that normal?). From quartz, I've
>>>> only been able to generate pdf files, which do not
>>>> match the quartz display to my satisfaction. I've been
>>>> attempting to find a way to generate other output
>>>> files for comparison. (I've managed to generate a bmp
>>>> file that looked awful, after acquiring Gostscript).
>>>>
>>>> Anyway, I figure R must able to produce high quality
>>>> graphic outputs, so I'm trying different things. To
>>>> this end I was hoping that the GDD package would allow
>>>> me to load the x11 module but it seems to be a no-go
>>>> (is there some system requirement I'm missing?).
>>>>
>>>> Conan

-- 
Jeffrey Horner       Computer Systems Analyst         School of Medicine
615-322-8606         Department of Biostatistics   Vanderbilt University



More information about the R-help mailing list