[R] rgl.postscript() failure when saving a scene

[Duncan Murdoch]

On 09/12/2011 9:49 AM, ppiras at uniroma3.it wrote:
> Dear all,
>
> I write to the community to know if other people
> expierenced the same problem. I want to save a rgl
> scene where I plotted a 3d cloud of spheres whose
> radius is set proportional to a given variable and
> coloured by another group variable and with 95%
> ellipsoids for distribution of any group.
> I had no problem at all for coding but I had two
> problem when saving the scene (in any format) using
> rgl.postscript()
> 1-  using plot3d(), if the points are set to sphere
> whose radius is proportional to a given variable the
> rgl.postscript() causes R crash
>
> 2- using simple points the scene is saved but the
> transaprencies of ellipsoids are lost.
>
> HERE below I copy a script with simulated data. I use
> R.2.14 but even with R2.12 I got the same problems
> Thankyou in advance for your help.
>
>
> datii<-data.frame(cbind(rnorm(30,1,1),rnorm(30,2,1),rnorm(30,3,1),abs(rnorm(30,4,1)),c(rep(1:2,15))))
>
> axis3D(axis.col="black",axis.scale=max(datii[,1:3]),axis.len=0.03)
> arrows3D(c(0,0,0),c(-max(datii[,1:3]),0,0),length=0.03)
> arrows3D(c(0,0,0),c(0,-max(datii[,1:3]),0),length=0.03)
> arrows3D(c(0,0,0),c(0,0,-max(datii[,1:3])),length=0.03)
>
>
> #########  with this command, i.e. "type="p", I cannot
> set the points dimension proportional to a given
> variable #######
> plot3d(datii[,1:3],box=F,col=datii[,5],type="p",cex=datii[,4]/10,add=T,bbox=F)
>
>
> ##### MAJOR PROBLEM: here (type="s") I can do that
> but...when I go to save the scene using
> rgl.postscript() R crasches .###########
> plot3d(datii[,1:3],box=F,col=datii[,5],type="s",radius=datii[,4]/10,add=T,bbox=F)
>
>
>
> #### MINOR PROBLEM: these are the 95% ellipses for any
> group; when saving using rgl.postscript() IN ANY
> FORMAT the transaparencies are lost ####
> plot3d(ellipse3d(cov(subset(datii[,1:3],datii[,5]==1)),centre=c(mean(subset(datii[,1],datii[,5]==1)),mean(subset(datii[,2],datii[,5]==1)),mean(subset(datii[,3],datii[,5]==1)))),col=1,alpha=0.3,add=T)
> plot3d(ellipse3d(cov(subset(datii[,1:3],datii[,5]==2)),centre=c(mean(subset(datii[,1],datii[,5]==2)),mean(subset(datii[,2],datii[,5]==2)),mean(subset(datii[,3],datii[,5]==2)))),col=2,alpha=0.3,add=T)
>
> rgl.postscript(file="test.pdf",fmt="pdf") #### THIS

I'd suggest using rgl.snapshot instead.  As the documentation says, 
rgl.postscript doesn't always work.  You could also try a newer version 
of the gl2ps library that rgl uses for Postscript conversions.

Duncan Murdoch
> COMMAND CAUSES R CRASCH WHEN THE RGL SECENE IS LOADED
> WITH POINTS SET AS "SPHERES"  ###  AND THEIR  RADIUS
> IS PROPORTIONAL TO A GIVEN VARIABLE.
>
>
>
>
>
>
>
>