Jose Gomez-Dans <jgomezdans@...> writes:
> I find this problem when generating a PDF and viewing it in Linux,but the
> on-screen version seems to work fine.
While the PDF format has advanced support for different color spaces and
rendering intents, the current PDF backend just uses DeviceRGB and
whatever the default rendering intent is. I wonder if some kind of
different color space or such setting would help - but this is a subject
that I know almost nothing about, and the PDF support is so complex that
I don't even know what the next reasonable step would be above just
Does the bluemarble image come with a specification or documentation
that mentions a color space, or what the pixel values are supposed to
mean, or how they are recommended to be rendered?
> Another thing you can do is to modify the bluemarble that
> comes with matplotlib using the gimp, as it is just an image file you can
> edit easily. Starts looking like data cooking, tho' ;-)
The fact is that different devices (displays, printers, projectors) have
different gamuts, and unless there is a specified color space, a set of
pixel values has no "right" mapping to the colors of the gamut (and even
if the space is known, mapping out-of-gamut colors can be done in
several ways). So I wouldn't call it "data cooking" if you are just
trying to get a reasonable contrast in your visualization of some data
that consists of values in some arbitrary space, although of course it
will not be any kind of true-color image either.
Jouni K. Seppänen