[Image-SIG] PIL Consuming All System Resources

[Fredrik Lundh]

The file has mode=RGB size=20400x28079, so you'd need about 2291246400
bytes to load it all into memory at once, and twice that to do e.g.
color conversion (which creates a second image memory), so it's a bit
on the big side, at least for a 32-bit environment.

Note that PIL doesn't actually read the pixels until you do something
that needs them, so you can do a sanity check before you proceed; e.g.

    im = Image.open(...)
    bytes = im.size[0] * im.size[1]
    if im.mode not in ("1", "L", "P"):
        bytes = bytes * 4  # 32-bit storage
    if bytes > some threshold:
        raise IOError("sorry, to big for us")
    # do operation that actually needs the pixels

</F>

On Sun, May 30, 2010 at 2:55 AM, Andy McCurdy <sedrik at gmail.com> wrote:
> Hi,
> We've been using PIL for the last 2 years to resize our users' uploaded
> images and have been extremely pleased. Thanks for all your effort.
> I noticed earlier today that we were experiencing issues of servers running
> out of memory. It looks like the problem might be within PIL. I've put
> together a few lines of code below that produce the same behavior. This is
> the first time we've seen anything like this, and it seems to be a problem
> with a specific image file (URL included in the code below) being converted
> to RGB or resized. This is the only image file I've encountered that
> produces this behavior. And no, I'm not quite sure why a user felt the need
> to upload a 17M JPG... :)
> I've tried both PIL 1.1.6 and 1.1.7, and the same behavior occurs on both
> Ubuntu 8.04 and OS X 10.5.
> A bug fix would be great, but given the infrequency of seeing this problem,
> I'd settle for a way to detect whether the image I'm working with will cause
> PIL issues so that I can avoid it.
> Thanks!
> -andy
> ##### Example Code #####
> import StringIO
> import urllib2
> from PIL import Image
> image_data =
> urllib2.urlopen('http://media.giantbomb.com/uploads/8/84310/1291762-img016.jpg').read()
> io = StringIO.StringIO(image_data)
> img = Image.open(io)
> # either of the following two lines causes Python to consume > 2G of memory
> and not return.
> converted_img = img.convert('RGB')
> # or
> resized_img = img.resize((200, 150), Image.ANTIALIAS)
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>
>