Re: [Rd] raster support in graphics devices

From: Martin Maechler <>
Date: Tue, 01 Dec 2009 10:47:23 +0100

>>>>> "CM" == Charlotte Maia <> >>>>> on Tue, 1 Dec 2009 17:48:33 +1300 writes:

    CM> Hi,
    CM> I consider raster graphics highly problematic in statistics.
    CM> People get caught up in the idea of creating pretty pictures, rather
    CM> than effectively visualising information.

    CM> Plus a lot of people (who should know better) needlessly put raster
    CM> images inside reports and articles (even books), which either makes
    CM> the files large (and very difficult to view), or creates blurry     CM> images, or both.

    CM> It's always good to have more functionality.

Yes indeed, and Paul has to be thanked loudly and heartfully and very much for providing it in the realm of graphics !!!

    CM> However, I certainly hope that most of the R community stick to vector
    CM> graphics (with conservative colour use), unless it is absolutely
    CM> necessary to do otherwise

I agree with your reasoning *pro* using vector graphics as much as possible, and warning against the overuse of raster graphics in scientific work !

    CM> necessary to do otherwise (and the only example I can think of, is     CM> modelling images themselves).

Did you look at the interesting page Paul sent the URL for, ?

Yes, modelling images is probably *the* most important application area, but *fast* direct plotting of a large matrix, mapping one matrix entry A_{i,j} to one pixel is also quite attractive, and the Matrix package authors (me being one) are quite interested to eventually see a version of   lattice::levelplot() that builds on Paul's new   grid::grid.raster() or grid::rasterGrob() functions.

    CM> On a side issue, I have found that R plots tend to be getting slower     CM> over the years.

    CM> I think cairo was a bad move, however that's just my opinion...

Dear me...
using Cairo features for X11 graphics is just the new default, so users notice the nice new possibilities ! If you want the very fast (but not so very nice looking) previous default X11() graphics, please do read the help pages:
--> ?X11
--> ?X11.options

and then either X11.options(type = "Xlib") or X11( type = "Xlib")

will give you the very fast (non-aliased, no alpha blending) X11 graphics that used to be the only one in R originally.

Martin Maechler, ETH Zurich

    CM> regards
    CM> -- 
    CM> Charlotte Maia

______________________________________________ mailing list Received on Tue 01 Dec 2009 - 09:49:21 GMT

This archive was generated by hypermail 2.2.0 : Tue 01 Dec 2009 - 11:00:54 GMT