ANN: FreeImagePy 1.2.2

[Terry Hancock]

On Thu, 16 Feb 2006 09:34:18 -0500
Andrew Gwozdziewycz <apgwoz at gmail.com> wrote:
> > Knowing some details about PIL and as good as no details
> > about FreeImage, I would like in this context to become
> > enlightened by the answer to the question, when does it
> > make sense to use FreeImage instead of PIL?
> >  From what I know up to now I can't see any use for
> >  FreeImage :-( .
> 
> both freeimagepy and freeimage are released under the GPL,
> PIL is not.

This is misleading! The PIL license is GPL-compatible (i.e.
it could be incorporated into a GPL work), as it is a
non-copyleft free license. You may be confused by Lundh's
proprietary-then-free release strategy: he always releases a
newer version under a proprietary license, at which point
the old version is free-licensed.  But the old version is
completely without strings -- you could, if you wanted, try
to give Mr. Lundh "a run for his money" by extending PIL
faster than he can.

Not exactly "cricket", but legal. ;-)

However, while PIL is very strong at image-manipulation, it
is weak on file-format compatibility: As long as you stick
to PNG format, you can do just about anything you want (and
there is decent JPG and GIF support). But if you have
compelling reasons to output data in other formats, you'll
find PIL disappointing -- there are many formats it can't
deal with at all, and most of the rest it can only import
(in the wild, there are *dozens* of obscure image formats to
be found).

IMHO, this isn't too serious an issue -- it's just
specialization: PIL is for *image manipulation* not *image
conversion*, and it provides enough of the latter for the
most common applications.

Other choices include ImageMagick, but API stability,
particularly of the Python bindings, is a real problem (in
fact, I think there's more than one python binding for
ImageMagick, but I'm not sure -- which goes some way to
showing why it's a frustrating package to use).  Things
may have improved lately, I haven't checked in awhile.

So, IMHO, there's still PLENTY of room for innovation in
the application area of image-handling tools for
Python.  PIL is not a "category-killer". At least not yet.

Cheers,
Terry

 -- 
Terry Hancock (hancock at AnansiSpaceworks.com)
Anansi Spaceworks http://www.AnansiSpaceworks.com