[R] Unwanted white borders on semi-transparent polygons?

[Prof Brian Ripley]

What version of R, what OS, what version of Acrobat?

I don't see this in 2.5.0 on Windows (using Acrobat 7: Acrobat does not 
exist on Linux, AFAIK).  And reading the PDF produced shows no sign of an 
extra object for the border.

On Fri, 27 Apr 2007, Matthew Neilson wrote:

> Hey all,
>
> I'm trying to create a plot of two semi-transparent regions. The reason they need to be partially transparent is so that I can see if there's any overlap. Here's some example code:
>
> # BEGIN
>
> pdf(file="test.pdf",version="1.4")
> plot(0,0,type="l",ylim=range(-3,3),xlim=range(-1,5))
> polygon(c(0,1,2,3,4,3,2,1,0), c(0,1,2,1,0,-1,-2,-1,0), col=rgb(1,0,0,0.5),
> border=NA)
> polygon(c(1,2,3,4,5,4,3,2,1), c(0,1,2,1,0,-1,-2,-1,0), col=rgb(0,0,1,0.5),
> border=NA)
> dev.off()
>
> # END
>
> The problem with this is that, despite setting "border = NA", I get a 
> big white border surrounding each polygon!! Funnily enough, setting the 
> alpha channel equal to 1 (as opposed to 0.5) *doesn't* give the border, 
> but an alpha channel of 1 produces an opaque polygon! :S
>
> I have read the FAQ, and (unfortunately) turning off line-art smoothing 
> does not give the desired effect. Furthermore, my pdfs print with a 
> white border surrounding each transparent polygon.
>
> Now, here comes the really bizarre part. Whilst Adobe Acrobat displays 
> the unwanted white border, Apple Preview respects the "border=NA" 
> argument and shows the two diamonds as they are intended. However, 
> opening up the pdf in Illustrator CS reveals that there is in fact a 
> transparent (according to Illustrator) border *on top* of each diamond. 
> Deleting these two borders (one for each polygon) and re-saving the pdf 
> appears to correct the issue. So the obvious question is: how did the 
> surrounding borders get there in the first place? A bug in the polygon 
> function, perhaps?
>
> Does anyone have any ideas for preventing these unwanted borders around 
> semi-transparent polygons (without having to resort to Illustrator)? Has 
> anyone else even come across this problem?
>
> Many thanks,
>
>
> -Matt
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595