Skip to content

png(type='cairo'): point symbols without boarders are not anti-aliased?

6 messages · Yihui Xie, Paul Murrell, Simon Urbanek

Original
Yihui Xie
# 
Hi,

It seems that anti-aliasing in png(type = 'cairo') is not well
supported for the point symbols without boarders, e.g. pch = 16. The
Cairo package works well, though. You can compare png() with
CairoPNG():

png(): http://i.imgur.com/8niB3jX.png
CairoPNG(): http://i.imgur.com/FZBJOxm.png

f = function(dev, ..., main = '') {
  dev(...)
  plot(c(1, 2, 1, 2), c(1, 1, 2, 2), pch=c(16, 19), cex=c(2, 2, 15, 15),
       xlim=c(0.5, 2.5), ylim=c(0.5, 3), main = deparse(substitute(dev)))
  dev.off()
}
f(grDevices::png, 'png-base.png', type = 'cairo')
f(Cairo::CairoPNG, 'png-Cairo.png')

If I remove the border for pch=19 (i.e. lwd=0), the point shows rough
edges as well.

I'm not sure if that is expected, or it is due to my misconfiguration
somewhere. I installed R via `apt-get install r-base-dev` under Ubuntu
using the CRAN repository.
R version 3.0.2 (2013-09-25)
Platform: x86_64-pc-linux-gnu (64-bit)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8
LC_PAPER=en_US.UTF-8       LC_NAME=C
 [9] LC_ADDRESS=C               LC_TELEPHONE=C
LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

loaded via a namespace (and not attached):
[1] Cairo_1.5-2 tools_3.0.2
jpeg      png     tiff    tcltk      X11     aqua http/ftp  sockets
    TRUE     TRUE     TRUE     TRUE     TRUE    FALSE     TRUE     TRUE
  libxml     fifo   cledit    iconv      NLS  profmem    cairo
    TRUE     TRUE     TRUE     TRUE     TRUE     TRUE     TRUE


Regards,
Yihui
--
Yihui Xie <xieyihui at gmail.com>
Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA
Yihui Xie
# 
Sorry, typo in the subject: I mean "borders".

Regards,
Yihui
--
Yihui Xie <xieyihui at gmail.com>
Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA
On Mon, Oct 21, 2013 at 4:43 PM, Yihui Xie <xie at yihui.name> wrote:
Paul Murrell
# 
Hi

Is this the same as "Bug 15462" ?
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=15462

Paul
On 10/22/13 10:43, Yihui Xie wrote:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Yihui Xie
    

    

      
      #
    

  


  

      
Perhaps yes. Sorry I did not check the bug reports. Can someone
elaborate on the "undesirable artefacts"? I made two heatmaps using
png() and CairoPNG(), respectively. I can see the difference, but it
is not very clear to me what the artefacts are, or what the facts
should be.

png(): http://i.imgur.com/lKrFG9i.png
CairoPNG(): http://i.imgur.com/Dv0rsKK.png

f = function(dev, ...) {
  dev(...)
  x = y <- seq(-4*pi, 4*pi, len = 27)
  r = sqrt(outer(x^2, y^2, "+"))
  z = cos(r^2)*exp(-r/6)
  image(z, main = deparse(substitute(dev)))
  dev.off()
}
f(grDevices::png, 'png-base.png', type = 'cairo')
f(Cairo::CairoPNG, 'png-Cairo.png')

Thanks!

Regards,
Yihui
--
Yihui Xie <xieyihui at gmail.com>
Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA

        
On Mon, Oct 21, 2013 at 5:28 PM, Paul Murrell <paul at stat.auckland.ac.nz> wrote:

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Simon Urbanek
    

    

      
      #
    

  


  

      
On Oct 21, 2013, at 9:07 PM, Yihui Xie wrote:

        

            

      
      
      
    

        
The Cairo package does "smart" anti-aliasing - it aligns lines that are perpendicular to the axes such that they centered at pixels. That avoids the anti-aliasing effects that Brian was talking about for the heatmap example. This enables Cairo to have full anti-aliasing support and still render heatmaps without artifacts.

However, there is no way around the anti-aliasing artifacts if you use arbitrary polygons without borders. For example:

library(deldir)
plot(c(-1,1),c(-1,1),ty='n')
for(p in tile.list(deldir(rnorm(200),rnorm(200)))) polygon(p$x,p$y,col=heat.colors(15)[runif(1,1,15)], border=NA)

That said, in our experience the Cairo approach works very well in practice.

Cheers,
Simon

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Yihui Xie
    

    

      
      #
    

  


  

      
Thanks for the hints! I was originally wondering the difference
between grDevices::png(type='cairo') and Cairo::CairoPNG() for the
case pch=16, a solid point without border. Cairo does a nice job for
such point symbols, but png() renders them very poorly.

Regards,
Yihui
--
Yihui Xie <xieyihui at gmail.com>
Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA


On Tue, Oct 22, 2013 at 9:09 AM, Simon Urbanek

        
<simon.urbanek at r-project.org> wrote: