From: Philip Nash (philip.nash_at_[hidden])
Date: 2001-11-15 16:05:33
> > Thus, the fixed distinction between color and binary images seems to be
> > appropriate even if a policy-based encoding scheme is supported within
> > kind of image.
> > IOW, I don't think that a complete ortogonality is at all
> useful; that is,
> > don't think that it would make any sense to use RLE with a color image.
> Why not? When dealing with cartoon-style graphics, large areas
> of a single color are common and can benefit from RLE. With less than 8
> per pixel,
> it might even sense to have packed color images.
I agree. I currently have a requirement I'm looking into, to pack 24bit
colour images up in a fast (to encode) and efficient algorithm that assumes
that most of the images are made up of only two or three colours, and are in
any case almost entirely block colours.
My only reservation is that there are potentially true colour photographic
images embedded in the images at times... but lets not get side-tracked by
My motivation for encoding the images is to marshall them cross-process
(with CORBA) so I want something that will reduce the size payload fast
enough to offset any time penalty involved in the encoding process (on a
modest desktop system).
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk