I'm trying to create a set of thumbnails, each one separately downscaled from the original image.
image = Image.open(path)
image = image.crop((left, upper, right, lower))
for size in sizes:
temp = copy.copy(image)
temp.thumbnail((size, height), Image.ANTIALIAS)
temp.save('%s%s%s.%s' % (path, name, size, format), quality=95)
The above code seemed to work fine but while testing I discovered that some images (I can't tell what's special about them, maybe only for PNG) raise this error:
/usr/local/lib/python2.6/site-packages/PIL/PngImagePlugin.py in read(self=<PIL.PngImagePlugin.PngStream instance>)
line: s = self.fp.read(8)
<type 'exceptions.AttributeError'>: 'NoneType' object has no attribute 'read'
Without the copy()
these images work just fine.
I could just open and crop the image anew for every thumbnail, but I'd rather have a bette开发者_JS百科r solution.
I guess copy.copy()
does not work for the PIL Image
class. Try using Image.copy()
instead, since it is there for a reason:
image = Image.open(path)
image = image.crop((left, upper, right, lower))
for size in sizes:
temp = image.copy() # <-- Instead of copy.copy(image)
temp.thumbnail((size, height), Image.ANTIALIAS)
temp.save('%s%s%s.%s' % (path, name, size, format), quality=95)
精彩评论