Re: Manipulating large graphics

[Havoc Pennington <hp redhat com>]

Scott Gifford <sgifford suspectclass com> writes:
> I'm working on an application that deals with extremely large graphics
> --- it loads in images of forms and lets you type on them.  Right now
> I'm just using a GdkPixbuf, and have run into some of its limitations.
> A scanned 300dpi form that's 8.5x11 inches is 8.4 million pixels, and
> with 24bit color, over 25MB of memory!  The viewed image is much
> larger than the screen, and flickers when it is manipulated.  And I
> have to convert PostScript and PDF files into PNGs to edit them.
>
> I'm trying to find a straightforward way to reduce memory usage, to
> allow zooming in and out, and to support PostScript and PDF image
> formats.
>
> Anybody got any pointers?  
>
> The code and more information about the project is at:
>
>     http://www.suspectclass.com/~sgifford/proforma/

Well, for flicker you can just use a backing pixmap - see the
"scribble" example in the GTK tutorial.

For memory, if you want 8.4 million pixels at 24 bits each, the size
of that is pretty much fixed. Not a lot you can do about it...  there
are elaborate things you can do, like swap portions of the image to
disk; I think the Gimp tries to do some of that, but it's extremely
nontrivial.

PostScript and PDF are even more nontrivial; those aren't really image
formats, they are vector formats. So you need a vector rendering
engine. Again, extremely nontrivial. You could however maybe base your
code on xpdf or ghostscript and get somewhere.

Havoc