From charlie at begeistert.org  Mon Jan  5 14:50:53 2009
From: charlie at begeistert.org (Charlie Clark)
Date: Mon, 5 Jan 2009 14:50:53 +0100
Subject: [Image-SIG] Problem with rendering accented capitals in text
Message-ID: <0D6D0090-1348-4A5A-9E4F-53C26E75AC64@begeistert.org>

Hiya,

this is a simple and I would have thought a common problem but I  
haven't been able to find it raised elsewhere. Characters such as ???  
are all missing their diaresises when rendered as capitals and this is  
independent of the typeface used. I've played around a bit and have  
not been able to fix it - all the magic seems to happen in the  
font.render() method in _imagingft and this is probably dependent upon  
Freetype code. But maybe someone out there knows of a workaround?

Thanks very much and a happy new year!

Charlie
--
Charlie Clark
Helmholtzstr. 20
D?sseldorf
D- 40215
Tel: +49-211-938-5360
GSM: +49-178-782-6226




From fredrik at pythonware.com  Mon Jan  5 17:23:21 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Mon, 5 Jan 2009 17:23:21 +0100
Subject: [Image-SIG] Problem with rendering accented capitals in text
In-Reply-To: <0D6D0090-1348-4A5A-9E4F-53C26E75AC64@begeistert.org>
References: <0D6D0090-1348-4A5A-9E4F-53C26E75AC64@begeistert.org>
Message-ID: <368a5cd50901050823p4db8e176qf72d2cffe614c34@mail.gmail.com>

This has been reported by some, but I do use the truetype bindings to
render swedish text myself, and it's also tested by our internal
tests.  My guess is that it's either depends on the font or the
freetype version.  Does it happen for all fonts for you?

(if you want, you can mail me a copy of the font file and a small
snippet that illustrates the problem via private mail, so I can take a
look at your specific case).

</F>

On Mon, Jan 5, 2009 at 2:50 PM, Charlie Clark <charlie at begeistert.org> wrote:
> Hiya,
>
> this is a simple and I would have thought a common problem but I haven't
> been able to find it raised elsewhere. Characters such as ??? are all
> missing their diaresises when rendered as capitals and this is independent
> of the typeface used. I've played around a bit and have not been able to fix
> it - all the magic seems to happen in the font.render() method in _imagingft
> and this is probably dependent upon Freetype code. But maybe someone out
> there knows of a workaround?
>
> Thanks very much and a happy new year!
>
> Charlie
> --
> Charlie Clark
> Helmholtzstr. 20
> D?sseldorf
> D- 40215
> Tel: +49-211-938-5360
> GSM: +49-178-782-6226
>
>
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From nick.pye at gmail.com  Thu Jan  1 22:12:28 2009
From: nick.pye at gmail.com (nicholas Pye)
Date: Thu, 1 Jan 2009 21:12:28 +0000
Subject: [Image-SIG] memory leak error / PIL Image
Message-ID: <F37D2EC5-F20D-4AC8-AD60-B3F3A24A2972@gmail.com>

Hello,

I have a peice of code that repeatidly draws new images (draw_image)  
and then compares the images drawn against an origional image  
(image_cost).

If I remove the PIL functions in these functions the rest of the code  
runs fine, and does not exhibit a memory leak, however when using PIL  
repeatidally in these functions I soon run out of memory. It's as if  
PIL is caching the images in memory or not cleaning up memory ... I've  
tried png, gif formats all eventually give a malloc of the period of  
the code run.

error

Python(2046) malloc: *** mmap(size=16777216) failed (error code=12)


def draw_image(popmember,path,img_tick,poly_vector_len):
     img=Image.new('RGB',(300,300),(0,0,0))
     d = aggdraw.Draw(img)
     for strand in range(len(popmember)):
          vector = split_strand(popmember[strand],"vec",poly_vector_len)
          rgb = split_strand(popmember[strand],"rgb",poly_vector_len)
          alpha =  
split_strand(popmember[strand],"alpha",poly_vector_len)
          b = aggdraw.Brush(tuple(rgb),alpha[0])
          d.polygon(vector,b)
     d.flush()
     img.save(path + str(img_tick) + '.jpg','PNG')

def image_cost(path,image2,img_tick):
  h1 = Image.open(path+image2).histogram()
  h2 = Image.open(path+str(img_tick)+'.png').histogram()
  rms = math.sqrt(reduce(operator.add,map(lambda a,b: (a-b)**2, h1,  
h2))/len(h1))
  return rms
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090101/fd1b48f2/attachment-0001.htm>

From bing.jian at gmail.com  Fri Jan  2 05:09:21 2009
From: bing.jian at gmail.com (Bing Jian)
Date: Thu, 1 Jan 2009 23:09:21 -0500
Subject: [Image-SIG] Specify line width in draw.rectangle
Message-ID: <e288380b0901012009o3bf51286jfbaa74fbbbd2e9b7@mail.gmail.com>

Hi all,
   Does anyone know how to specify the line width in
draw.rectangle(box,options)?
   Many thanks in advance and happy new year!


Bing
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090101/4f253ad2/attachment.htm>

From spinboy_1999 at yahoo.com  Sat Jan  3 00:58:45 2009
From: spinboy_1999 at yahoo.com (Marcello Lioy)
Date: Fri, 2 Jan 2009 15:58:45 -0800 (PST)
Subject: [Image-SIG] selftest failing
Message-ID: <916494.43024.qm@web34408.mail.mud.yahoo.com>

I have downloaded the JPG library and built it.  I also downloaded and built 1.1.6 and build it sucessfully - the outfput from the build process is:

--------------------------------------------------------------------
PIL 1.1.6 BUILD SUMMARY
--------------------------------------------------------------------
version       1.1.6
platform      linux2 2.4.1 (#2, Oct 18 2006, 22:35:03)
              [GCC 3.3.5 (Debian 1:3.3.5-13)]
--------------------------------------------------------------------
*** TKINTER support not available
--- JPEG support ok
*** ZLIB (PNG/ZIP) support not available
*** FREETYPE2 support not available
--------------------------------------------------------------------

 
The only one I care about is jpeg, and oddly when I run the self test I get the following error:

*****************************************************************
Failure in example: _info(Image.open("Images/lena.jpg"))
from line #24 of selftest.testimage
Exception raised:
Traceback (most recent call last):
  File "./doctest.py", line 499, in _run_examples_inner
    exec compile(source, "<string>", "single") in globs
  File "<string>", line 1, in ?
  File "./selftest.py", line 22, in _info
    im.load()
  File "PIL/ImageFile.py", line 180, in load
    d = Image._getdecoder(self.mode, d, a, self.decoderconfig)
  File "PIL/Image.py", line 375, in _getdecoder
    raise IOError("decoder %s not available" % decoder_name)
IOError: decoder jpeg not available
1 items had failures:
   1 of  57 in selftest.testimage
***Test Failed*** 1 failures.
*** 1 tests of 57 failed.

I am not sure how to proceed from here as the build seems to indicate that JPEG support is there, while one of the tests is clearly failing...

-
I don't want to look away
I don't want to be the one denied
It ain't no fault of mine
Someone, somewhere told you lies
- Duncan Sheik



      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090102/56153202/attachment.htm>

From fredrik at pythonware.com  Mon Jan  5 17:41:24 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Mon, 5 Jan 2009 17:41:24 +0100
Subject: [Image-SIG] malloc error
In-Reply-To: <40509F14-8DD7-43B8-AC6B-E6D0D48D7964@gmail.com>
References: <40509F14-8DD7-43B8-AC6B-E6D0D48D7964@gmail.com>
Message-ID: <368a5cd50901050841p1b25bcdam5705d8b3512e7c4e@mail.gmail.com>

What version of aggdraw are you using?  The Draw(im) form is buggy in
some versions; it doesn't release the im object when the draw object
goes away.  Here's a workaround (cut and pasted from an old message to
this list):

    as a workaround, you can use the original constructor syntax instead,
    together with fromstring/tostring:

    d = Draw(im.mode, im.size)
    d.fromstring(im.tostring())

    # draw your stuff

    im.fromstring(d.tostring())

</F>

2008/12/31 nicholas Pye <nick.pye at gmail.com>:
> Hello,
> I have a peice of code that repeatidly draws new images (draw_image) and
> then compares the images drawn against an origional image (image_cost).
> If I remove the PIL functions in these functions the rest of the code runs
> fine, and does not exhibit a memory leak, however when using PIL
> repeatidally in these functions I soon run out of memory. It's as if PIL is
> caching the images in memory or not cleaning up memory ... I've tried png,
> gif formats all eventually give a malloc of the period of the code run.
> error
> Python(2046) malloc: *** mmap(size=16777216) failed (error code=12)
>
> def draw_image(popmember,path,img_tick,poly_vector_len):
>     img=Image.new('RGB',(300,300),(0,0,0))
>     d = aggdraw.Draw(img)
>     for strand in range(len(popmember)):
>          vector = split_strand(popmember[strand],"vec",poly_vector_len)
>          rgb = split_strand(popmember[strand],"rgb",poly_vector_len)
>          alpha = split_strand(popmember[strand],"alpha",poly_vector_len)
>          b = aggdraw.Brush(tuple(rgb),alpha[0])
>          d.polygon(vector,b)
>     d.flush()
>     img.save(path + str(img_tick) + '.jpg','PNG')
> def image_cost(path,image2,img_tick):
>  h1 = Image.open(path+image2).histogram()
>  h2 = Image.open(path+str(img_tick)+'.png').histogram()
>  rms = math.sqrt(reduce(operator.add,map(lambda a,b: (a-b)**2, h1,
> h2))/len(h1))
>  return rms
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>
>

From ggpolo at gmail.com  Mon Jan  5 18:42:33 2009
From: ggpolo at gmail.com (Guilherme Polo)
Date: Mon, 5 Jan 2009 15:42:33 -0200
Subject: [Image-SIG] Infinite loop in PIL/ImageFont.py
Message-ID: <ac2200130901050942r31b5ab13y2d39a6d5a4d4693b@mail.gmail.com>

Hi,

There is an infinite loop in PIL/ImageFont.py (PIL 1.1.6) which can be
entered by using ImageFont.load. Found this while testing the py3k
port, but the attached patch is against the standard one.

-- 
-- Guilherme H. Polo Goncalves
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ImageFont.diff
Type: text/x-patch
Size: 543 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090105/ac2bf453/attachment.bin>

From charlie at begeistert.org  Tue Jan  6 10:14:32 2009
From: charlie at begeistert.org (Charlie Clark)
Date: Tue, 6 Jan 2009 10:14:32 +0100
Subject: [Image-SIG] Problem with rendering accented capitals in text
In-Reply-To: <368a5cd50901050823p4db8e176qf72d2cffe614c34@mail.gmail.com>
References: <0D6D0090-1348-4A5A-9E4F-53C26E75AC64@begeistert.org>
	<368a5cd50901050823p4db8e176qf72d2cffe614c34@mail.gmail.com>
Message-ID: <2143DCCD-8929-487E-AC11-19C15A7F3EE5@begeistert.org>


Am 05.01.2009 um 17:23 schrieb Fredrik Lundh:

> This has been reported by some, but I do use the truetype bindings to
> render swedish text myself, and it's also tested by our internal
> tests.  My guess is that it's either depends on the font or the
> freetype version.  Does it happen for all fonts for you?

While it did seem to affect all our fonts, it may only be Helvetica.  
If this is the case I wonder whether using small caps would be a  
reasonable workaround.

> (if you want, you can mail me a copy of the font file and a small
> snippet that illustrates the problem via private mail, so I can take a
> look at your specific case).


Thanks. Font sent in a separate e-mail.

Charlie
--
Charlie Clark
Helmholtzstr. 20
D?sseldorf
D- 40215
Tel: +49-211-938-5360
GSM: +49-178-782-6226




From cab938 at mail.usask.ca  Thu Jan  8 02:18:19 2009
From: cab938 at mail.usask.ca (cab938 at mail.usask.ca)
Date: Wed,  7 Jan 2009 19:18:19 -0600
Subject: [Image-SIG] Fast cropping
In-Reply-To: <ac2200130901050942r31b5ab13y2d39a6d5a4d4693b@mail.gmail.com>
References: <ac2200130901050942r31b5ab13y2d39a6d5a4d4693b@mail.gmail.com>
Message-ID: <1231377499.4965545b40212@webmail.usask.ca>

I have the need to break a single large image up (3 colour bands) into a series
of smaller images in a grid-like fashion.  The code we have written write now
works, but it's really slow - we essentially walk around the image cropping and
loading to return new PIL Image objects.  Any suggestions on a faster way? 
Should I be using fromBuffer() and pixelaccess to create my sub-images?  I want
Image objects as I use a number of PIL methods after that, I just don't know the
fastest way to create them.

Regards,

Chris

From g.kloss at massey.ac.nz  Thu Jan  8 03:10:41 2009
From: g.kloss at massey.ac.nz (Guy K. Kloss)
Date: Thu, 8 Jan 2009 15:10:41 +1300
Subject: [Image-SIG] Fast cropping
In-Reply-To: <1231377499.4965545b40212@webmail.usask.ca>
References: <ac2200130901050942r31b5ab13y2d39a6d5a4d4693b@mail.gmail.com>
	<1231377499.4965545b40212@webmail.usask.ca>
Message-ID: <200901081510.41464.g.kloss@massey.ac.nz>

On Thu, 08 Jan 2009 14:18:19 cab938 at mail.usask.ca wrote:
> I have the need to break a single large image up (3 colour bands) into a
> series of smaller images in a grid-like fashion. ?The code we have written
> write now works, but it's really slow - we essentially walk around the
> image cropping and loading to return new PIL Image objects. ?Any
> suggestions on a faster way? Should I be using fromBuffer() and pixelaccess
> to create my sub-images? ?I want Image objects as I use a number of PIL
> methods after that, I just don't know the fastest way to create them.

Have you tried to convert the PIL image into a NumPy image, and then just use 
the slicing operations? After getting those sub-arrays as slices convert them 
back to PIL arrays. Just a thought, as NumPy is actually quite fast, but don't 
know whether it compares to what you're doing now.

Guy

-- 
Guy K. Kloss
Institute of Information and Mathematical Sciences
Te Kura P?taiao o M?hiohio me P?ngarau
Room 2.63, Quad Block A Building
Massey University, Auckland, Albany
Private Bag 102 904, North Shore Mail Centre
voice: +64 9 414-0800 ext. 9585   fax: +64 9 441-8181
eMail: G.Kloss at massey.ac.nz  http://iims.massey.ac.nz


From pluskid at gmail.com  Mon Jan 12 08:09:28 2009
From: pluskid at gmail.com (Zhang Chiyuan)
Date: Mon, 12 Jan 2009 15:09:28 +0800
Subject: [Image-SIG] Bug in the mode recognition code of Image.fromarray?
Message-ID: <b6f0e9dd0901112309m5d509491s880904d29954cc79@mail.gmail.com>

Hi all,

I have a transparent png file and load it with PIL:

from PIL import Image
from numpy import array

im = Image.open('foo.png')
a = array(im)
im2 = Image.fromarray(a)

It should automatically determine that the mode is 'RGBA' but it
failed to do so. And the result im2 is not a correct image. I looked
at the code of fromarray:

def fromarray(obj, mode=None):
    arr = obj.__array_interface__
    shape = arr['shape']
    ndim = len(shape)
    try:
        strides = arr['strides']
    except KeyError:
        strides = None
    if mode is None:
        typestr = arr['typestr']
        if not (typestr[0] == '|' or typestr[0] == _ENDIAN or
                typestr[1:] not in ['u1', 'b1', 'i4', 'f4']):
            raise TypeError("cannot handle data-type")
        typestr = typestr[:2]
        if typestr == 'i4':
            mode = 'I'
        elif typestr == 'f4':
            mode = 'F'
        elif typestr == 'b1':
            mode = '1'
        elif ndim == 2:
            mode = 'L'
        elif ndim == 3:
            mode = 'RGB'
        elif ndim == 4:
            mode = 'RGBA'
        else:
            raise TypeError("Do not understand data.")
    ndmax = 4
    bad_dims=0
    if mode in ['1','L','I','P','F']:
        ndmax = 2
    elif mode == 'RGB':
        ndmax = 3
    if ndim > ndmax:
        raise ValueError("Too many dimensions.")

    size = shape[:2][::-1]
    if strides is not None:
        obj = obj.tostring()

    return frombuffer(mode, size, obj, "raw", mode, 0, 1)

it seems that when ndim is 4, the mode is set to RGBA. I'm not familiar with
the various image modes. However, I think it should be that shape[-1] == 4
instead of ndim == 4. And the same thing holds for 'RGB' and == 3.

Can anyone confirm me and fix this bug? Or correct me and provide a solution?

BTW: I'm not on this mailing list, please CC me when replying. Thanks!

-- 
pluskid

From Chris.Barker at noaa.gov  Mon Jan 12 18:12:52 2009
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Mon, 12 Jan 2009 09:12:52 -0800
Subject: [Image-SIG] Bug in the mode recognition code of Image.fromarray?
In-Reply-To: <b6f0e9dd0901112309m5d509491s880904d29954cc79@mail.gmail.com>
References: <b6f0e9dd0901112309m5d509491s880904d29954cc79@mail.gmail.com>
Message-ID: <496B7A14.9030604@noaa.gov>

Zhang Chiyuan wrote:
> it seems that when ndim is 4, the mode is set to RGBA. I'm not familiar with
> the various image modes. However, I think it should be that shape[-1] == 4
> instead of ndim == 4. And the same thing holds for 'RGB' and == 3.

at first glance, I think you are correct. however, this sounds familiar, 
so it may have been already fixed (or a fixed posted here -- I don't 
know if it was applied)

You might want to post this question to the numpy list -- that code was 
written by the numpy folks.

-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker at noaa.gov

From pluskid at gmail.com  Tue Jan 13 02:41:09 2009
From: pluskid at gmail.com (Zhang Chiyuan)
Date: Tue, 13 Jan 2009 09:41:09 +0800
Subject: [Image-SIG] Bug in the mode recognition code of Image.fromarray?
In-Reply-To: <496B7A14.9030604@noaa.gov>
References: <b6f0e9dd0901112309m5d509491s880904d29954cc79@mail.gmail.com>
	<496B7A14.9030604@noaa.gov>
Message-ID: <b6f0e9dd0901121741y1816ac1i7762b958e7391b3a@mail.gmail.com>

After searching in the Numpy mailing list. I found a link to the Image-SIG list:
http://www.mail-archive.com/image-sig at python.org/msg01482.html
And this seems to fix the problem.

Is this patch applied to the PIL code? If it is applied, from what version? Or
maybe still not released?

Thanks again.

On Tue, Jan 13, 2009 at 1:12 AM, Christopher Barker
<Chris.Barker at noaa.gov> wrote:
> Zhang Chiyuan wrote:
>>
>> it seems that when ndim is 4, the mode is set to RGBA. I'm not familiar
>> with
>> the various image modes. However, I think it should be that shape[-1] == 4
>> instead of ndim == 4. And the same thing holds for 'RGB' and == 3.
>
> at first glance, I think you are correct. however, this sounds familiar, so
> it may have been already fixed (or a fixed posted here -- I don't know if it
> was applied)
>
> You might want to post this question to the numpy list -- that code was
> written by the numpy folks.
>
> -Chris
>
>
> --
> Christopher Barker, Ph.D.
> Oceanographer
>
> Emergency Response Division
> NOAA/NOS/OR&R            (206) 526-6959   voice
> 7600 Sand Point Way NE   (206) 526-6329   fax
> Seattle, WA  98115       (206) 526-6317   main reception
>
> Chris.Barker at noaa.gov
>



-- 
pluskid

From bruces1 at gmail.com  Tue Jan 13 16:13:54 2009
From: bruces1 at gmail.com (Bruce Sutherland)
Date: Wed, 14 Jan 2009 00:13:54 +0900
Subject: [Image-SIG] Bilinear filter: unexpected results for 50 percent
	resize
Message-ID: <858411560901130713n1910afc5y60fb37c810ad367a@mail.gmail.com>

Hi everyone.

I resized an 8x16 RGB image to 4x8 using PIL (1.1.6). I used the
bilinear filter. The results were not exactly as I had expected.

I performed the same operation using The Gimp (2.4.2), and sure
enough, the result was quite different, and fit my understanding of a
bilinear filter.

See attached images.

There seem to be two difference, one significant (for my application),
and one a minor implementation detail.

1) Each pixel in The Gimp's output image can be determined by summing
the pixels in each 2x2 block starting from the top left hand corner
and dividing by four. This fits my understanding of a bilinear filter.
I don't understand how PIL arrives at its result.
2) PIL appears to use pure integer division to average the 2x2 blocks,
resulting in truncation. By contrast, Gimp's filter seems to use
fractional math and round to the nearest value.

I examined the resize function in the Image class, but stopped when I
realised that it calls a function from the compiled imaging core
library. This is my first adventure with PIL. Please forgive the
attachments, but they are very small.

Bruce.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: original.png
Type: image/png
Size: 193 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090114/afbe51b1/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pil_resize_linear.png
Type: image/png
Size: 159 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090114/afbe51b1/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gimp_resize_linear.png
Type: image/png
Size: 141 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090114/afbe51b1/attachment-0002.png>

From bruces1 at gmail.com  Tue Jan 13 16:24:22 2009
From: bruces1 at gmail.com (Bruce Sutherland)
Date: Wed, 14 Jan 2009 00:24:22 +0900
Subject: [Image-SIG] Bilinear filter: unexpected results for 50 percent
	resize
Message-ID: <858411560901130724i3853e488geabdee2cb85860@mail.gmail.com>

I forgot the source:

import Image

size = (8, 16)
halfsize = (size[0]/2, size[1]/2)
image = Image.open('original.png')
rgb = image.convert('RGB')
linear = rgb.resize(halfsize, Image.BILINEAR)
linear.save('pil_resize_linear.png')

From asbach at ient.rwth-aachen.de  Wed Jan 14 12:17:40 2009
From: asbach at ient.rwth-aachen.de (Mark Asbach)
Date: Wed, 14 Jan 2009 12:17:40 +0100
Subject: [Image-SIG] Trying to implement the array interface
Message-ID: <E3163E7F-049F-40EE-8EC4-3A3FF4107ECC@ient.rwth-aachen.de>

Hi there,

I'm currently extending the Python wrapper for the Open Computer  
Vision Library (opencv) with the goal to interface numerical libraries  
as seemless as possible. Unfortunately, it doesn't seem to be that  
easy ;-)

What I've done so far:

- Added an __array_interface__ property to the Python side of OpenCV  
data structures (matrices and images) that uses version 3 of the  
interface definition and supplies the keys 'version', 'shape',  
'typestr', 'data' and in some cases 'strides' when we have non- 
continuos memory layout. I think, I'm compatible to http://numpy.scipy.org/array_interface.shtml 
  .
- Added parsing the __array_interface__ of Python objects passed to  
OpenCV methods. I'm a bit unsure of how to use the C/C++ side (array  
struct) and if I can expect it to be there (for example: I don't  
provide one with OpenCV). Since I intend to keep OpenCV independent of  
numpy, calling functions from numpy.h is not an option, as far as I  
can see.

The stuff described above is in the head revision of OpenCV,  
accessible via "svn co https://opencvlibrary.svn.sourceforge.net/svnroot/opencvlibrary/trunk/opencv 
".

I've tried using the following packages with OpenCV this way:
- numpy (1.0.4): everything works as expected. This is the most  
important library for OpenCV users, so this is a good sign.
- pylab/matplotlib (0.91.2): seems to use numpy / scipy-core.  
Everything okay.
- PIL (1.1.6): the array interface (Python side) doesn't adhere to the  
definition -> no 'version' key, 'data' is string, not a tuple holding  
the pointer. What to do with this?
- Numeric (24.2): I can create arrays from OpenCV datatypes and print  
them. Converting to other types however always yields 'Cannot convert  
scalar to float' or 'a float is required'. Strange :-/ Numeric.array  
instances also don't carry an __array_interface__. I can however  
convert by using numpy.arrays as intermediate step.
- Gnuplot (1.7): uses Numeric, so doesn't work as well
- pymat: didn't check. Seems to use Numeric, test results cover  
Numeric 23 and Matlab 6.5 only, so this package might be dead?
- numarray: didn't check. Is there still any relevance of this package?

Best,
Mark

-- 
Mark Asbach
Institut f?r Nachrichtentechnik, RWTH Aachen University
http://www.ient.rwth-aachen.de/cms/team/m_asbach

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4403 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090114/66ef420d/attachment.bin>

From mhagger at alum.mit.edu  Wed Jan 14 12:52:24 2009
From: mhagger at alum.mit.edu (Michael Haggerty)
Date: Wed, 14 Jan 2009 12:52:24 +0100
Subject: [Image-SIG] Trying to implement the array interface
In-Reply-To: <E3163E7F-049F-40EE-8EC4-3A3FF4107ECC@ient.rwth-aachen.de>
References: <E3163E7F-049F-40EE-8EC4-3A3FF4107ECC@ient.rwth-aachen.de>
Message-ID: <496DD1F8.2020300@alum.mit.edu>

Mark Asbach wrote:
> I'm currently extending the Python wrapper for the Open Computer Vision
> Library (opencv) with the goal to interface numerical libraries as
> seemless as possible. [...]
> - Gnuplot (1.7): uses Numeric, so doesn't work as well

Gnuplot.py version 1.8 uses numpy.  Maybe that will help?

Michael

From gael.varoquaux at normalesup.org  Wed Jan 14 14:05:18 2009
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 14 Jan 2009 14:05:18 +0100
Subject: [Image-SIG] [Numpy-discussion] Trying to implement the array
	interface
In-Reply-To: <E3163E7F-049F-40EE-8EC4-3A3FF4107ECC@ient.rwth-aachen.de>
References: <E3163E7F-049F-40EE-8EC4-3A3FF4107ECC@ient.rwth-aachen.de>
Message-ID: <20090114130518.GA21830@phare.normalesup.org>

On Wed, Jan 14, 2009 at 12:17:40PM +0100, Mark Asbach wrote:
> - Added an __array_interface__ property to the Python side of OpenCV data 
> structures (matrices and images) [...]

> - Added parsing the __array_interface__ of Python objects passed to OpenCV 
> methods. [...]

> I've tried using the following packages with OpenCV this way:
> - numpy (1.0.4): everything works as expected. This is the most important 
> library for OpenCV users, so this is a good sign. [...]

_Fantastic_! I read from this e-mail that you are having a few
difficulties moving forward, but what you have already achieved is
really fantastic and will definitely be very useful to many people.

Sorry, I can't help you with the problems you are having, but I wanted to
thank you for this effort.

Ga?l

From berankin99 at yahoo.com  Wed Jan 14 16:26:45 2009
From: berankin99 at yahoo.com (Bernard Rankin)
Date: Wed, 14 Jan 2009 07:26:45 -0800 (PST)
Subject: [Image-SIG] adding overlay & text to many files
Message-ID: <933435.46579.qm@web112213.mail.gq1.yahoo.com>

Hello,

I'm sure this must be a common question, but I could not find any docs on this.  Perhapse my google skills needs improvment.

Basically, I've got a bunch of pictures from a party/event that I want to do several things to:

1)  Resize to 4x6 @ 300 PPI 
2)  Add a slightly translucent ~ 1" white bar on top of the image (across the bottom).
3)  Over the white bar, on the left, place a small 1"x1" logo.
4)  Over the white bar, centered on the remainder, put the names of the people in the photo. (Reducing text size if needed to fit on one line.)
5)  Over the white bar, centered on the remainder, put the name of the event on the next line.
6)  Add a black border to entire image.
7)  Export as JPEG.

There is a CSV file that contains the "names of the people in each photo" to file name mapping.  (One line per image file.)

Can this be automated with the PIL?

The biggest challenge I imagine would be the step 4 that needs to auto-resize text to fit in a certain area....  That is, it's a single-line text box such that the text inside starts out at some desired size, but is
automatically shrunk (in font size) to fit the width of the box if the
supplied text is too long.

Additionally, for the overlay, it would be nice to use a GIMP file with dummy text as a template that I can supply the correct text and base image file.  Is this possible with PIL?

Thank you,
:)



      


From frederic.mantegazza at gbiloba.org  Wed Jan 14 16:49:34 2009
From: frederic.mantegazza at gbiloba.org (=?ISO-8859-1?Q?Fr=E9d=E9ric?=)
Date: Wed, 14 Jan 2009 16:49:34 +0100
Subject: [Image-SIG] adding overlay & text to many files
In-Reply-To: <933435.46579.qm@web112213.mail.gq1.yahoo.com>
Message-ID: <lTIPaszB.1231948174.9867510.fma@alezan.org>


Le 14/1/2009, "Bernard Rankin" <berankin99 at yahoo.com> a ?crit:

>I'm sure this must be a common question, but I could not find any docs on this.  Perhapse my google skills needs improvment.
>
>Basically, I've got a bunch of pictures from a party/event that I want to do several things to:
>
>1)  Resize to 4x6 @ 300 PPI
>2)  Add a slightly translucent ~ 1" white bar on top of the image (across the bottom).
>3)  Over the white bar, on the left, place a small 1"x1" logo.
>4)  Over the white bar, centered on the remainder, put the names of the people in the photo. (Reducing text size if needed to fit on one line.)
>5)  Over the white bar, centered on the remainder, put the name of the event on the next line.
>6)  Add a black border to entire image.
>7)  Export as JPEG.
>
>There is a CSV file that contains the "names of the people in each photo" to file name mapping.  (One line per image file.)
>
>Can this be automated with the PIL?
>
>The biggest challenge I imagine would be the step 4 that needs to auto-resize text to fit in a certain area....  That is, it's a single-line text box such that the text inside starts out at some desired size, but is
>automatically shrunk (in font size) to fit the width of the box if the
>supplied text is too long.
>
>Additionally, for the overlay, it would be nice to use a GIMP file with dummy text as a template that I can supply the correct text and base image file.  Is this possible with PIL?

I think that it is possible to do all you want. First, I would suggest
you have a look at Phatch, a great batch processor, build on PIL. Step
1, 6 and 7 are already done in Phatch default actions.

Then, I would write a few Phatch actions to match your needs. I would
first build an transparent image with the logo and names, then merge
this image with the original.

I agree that the tricky part is the text. Anyway, you can start to
prototype this with PIL, then turn your code into a Phatch action to
have a great batch processor.

--
   Fr?d?ric

From cannon.el at gmail.com  Wed Jan 14 17:27:55 2009
From: cannon.el at gmail.com (Laura & Edward Cannon)
Date: Wed, 14 Jan 2009 08:27:55 -0800
Subject: [Image-SIG] adding overlay & text to many file
In-Reply-To: <933435.46579.qm@web112213.mail.gq1.yahoo.com>
References: <933435.46579.qm@web112213.mail.gq1.yahoo.com>
Message-ID: <b986b290901140827j59794cecmbd0ea56b29d554f3@mail.gmail.com>

Reading gimp files is tricky. Resizing photos is not hard just
remember that PIL thinks in pixels, so just multiply your PPI by the
dimension to get the size in pixels. Look at the paste function in the
Image module for details on adding logos and bars. The ImageDraw
function might be useful for the rest of the process. There is a
method on font objects, getsize, that will tell you how large your
text will be, this would allow you to choose a smaller font if
necessary.
Happy programing

On Wed, Jan 14, 2009 at 7:26 AM, Bernard Rankin <berankin99 at yahoo.com> wrote:
> Hello,
>
> I'm sure this must be a common question, but I could not find any docs on this.  Perhapse my google skills needs improvment.
>
> Basically, I've got a bunch of pictures from a party/event that I want to do several things to:
>
> 1)  Resize to 4x6 @ 300 PPI
> 2)  Add a slightly translucent ~ 1" white bar on top of the image (across the bottom).
> 3)  Over the white bar, on the left, place a small 1"x1" logo.
> 4)  Over the white bar, centered on the remainder, put the names of the people in the photo. (Reducing text size if needed to fit on one line.)
> 5)  Over the white bar, centered on the remainder, put the name of the event on the next line.
> 6)  Add a black border to entire image.
> 7)  Export as JPEG.
>
> There is a CSV file that contains the "names of the people in each photo" to file name mapping.  (One line per image file.)
>
> Can this be automated with the PIL?
>
> The biggest challenge I imagine would be the step 4 that needs to auto-resize text to fit in a certain area....  That is, it's a single-line text box such that the text inside starts out at some desired size, but is
> automatically shrunk (in font size) to fit the width of the box if the
> supplied text is too long.
>
> Additionally, for the overlay, it would be nice to use a GIMP file with dummy text as a template that I can supply the correct text and base image file.  Is this possible with PIL?
>
> Thank you,
> :)
>
>
>
>
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From berankin99 at yahoo.com  Wed Jan 14 20:37:29 2009
From: berankin99 at yahoo.com (Bernard Rankin)
Date: Wed, 14 Jan 2009 11:37:29 -0800 (PST)
Subject: [Image-SIG] adding overlay & text to many files
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>
Message-ID: <956018.37319.qm@web112222.mail.gq1.yahoo.com>




> 
> I think that it is possible to do all you want. First, I would suggest
> you have a look at Phatch, a great batch processor, build on PIL. Step
> 1, 6 and 7 are already done in Phatch default actions.
> 
> Then, I would write a few Phatch actions to match your needs. I would
> first build an transparent image with the logo and names, then merge
> this image with the original.
> 
> I agree that the tricky part is the text. Anyway, you can start to
> prototype this with PIL, then turn your code into a Phatch action to
> have a great batch processor.
> 


Where can I find Phatch?

http://photobatch.stani.be/ seems to only have a screen shot, and no further info or links......  Am I missing something?

 I do see a version (0.1.5) in the Ubuntu repositories, is that that the latest?


      


From Chris.Barker at noaa.gov  Wed Jan 14 20:53:51 2009
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Wed, 14 Jan 2009 11:53:51 -0800
Subject: [Image-SIG] adding overlay & text to many files
In-Reply-To: <956018.37319.qm@web112222.mail.gq1.yahoo.com>
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>
	<956018.37319.qm@web112222.mail.gq1.yahoo.com>
Message-ID: <496E42CF.7090001@noaa.gov>

Bernard Rankin wrote:
> Where can I find Phatch?
> 
> http://photobatch.stani.be/ seems to only have a screen shot, and no further info or links......  Am I missing something?

look at the bottom of the page -- there are links to download and docs.

And yes, that was far more subtle than all the ads! sigh.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker at noaa.gov

From spe.stani.be at gmail.com  Wed Jan 14 22:23:46 2009
From: spe.stani.be at gmail.com (Stani)
Date: Wed, 14 Jan 2009 22:23:46 +0100
Subject: [Image-SIG] adding overlay & text to many files
In-Reply-To: <lTIPaszB.1231948174.9867510.fma@alezan.org>
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>
Message-ID: <1231968226.2882.100.camel@blue>

Op woensdag 14-01-2009 om 16:49 uur [tijdzone +0100], schreef Fr?d?ric:
> Le 14/1/2009, "Bernard Rankin" <berankin99 at yahoo.com> a ?crit:
> 
> >I'm sure this must be a common question, but I could not find any docs on this.  Perhapse my google skills needs improvment.
> >
> >Basically, I've got a bunch of pictures from a party/event that I want to do several things to:
> >
> >1)  Resize to 4x6 @ 300 PPI
> >2)  Add a slightly translucent ~ 1" white bar on top of the image (across the bottom).
> >3)  Over the white bar, on the left, place a small 1"x1" logo.
> >4)  Over the white bar, centered on the remainder, put the names of the people in the photo. (Reducing text size if needed to fit on one line.)
> >5)  Over the white bar, centered on the remainder, put the name of the event on the next line.
> >6)  Add a black border to entire image.
> >7)  Export as JPEG.
> >
> >There is a CSV file that contains the "names of the people in each photo" to file name mapping.  (One line per image file.)
> >
> >Can this be automated with the PIL?
> >
> >The biggest challenge I imagine would be the step 4 that needs to auto-resize text to fit in a certain area....  That is, it's a single-line text box such that the text inside starts out at some desired size, but is
> >automatically shrunk (in font size) to fit the width of the box if the
> >supplied text is too long.
> >
> >Additionally, for the overlay, it would be nice to use a GIMP file with dummy text as a template that I can supply the correct text and base image file.  Is this possible with PIL?
> 
> I think that it is possible to do all you want. First, I would suggest
> you have a look at Phatch, a great batch processor, build on PIL. Step
> 1, 6 and 7 are already done in Phatch default actions.
And you would get 2 and 3 also for free. For both steps you can use the
watermark action: once with a fixed translucent white bar image (.png)
and once with the logo.

> 
> Then, I would write a few Phatch actions to match your needs. I would
> first build an transparent image with the logo and names, then merge
> this image with the original.
> 
> I agree that the tricky part is the text. Anyway, you can start to
> prototype this with PIL, then turn your code into a Phatch action to
> have a great batch processor.
Phatch has already a text action which has done the hard work already as
well. Just start from that as a base and hack it to use a csv file as
input instead of a fixed text. If you have played with python and PIL
already, it should be not that hard.

BTW Phatch can be used as a desktop application but also as a server
process without a GUI, once you have your action configured.

Phatch is available in Ubuntu Intrepid, Fedora, Suse, Arch Linux, ... in
the usual repositories. The latest version is 0.1.x (where x are bug fix
releases which you probably don't need).

Stani


From frederic.mantegazza at gbiloba.org  Wed Jan 14 23:04:47 2009
From: frederic.mantegazza at gbiloba.org (=?iso-8859-15?q?Fr=E9d=E9ric?=)
Date: Wed, 14 Jan 2009 23:04:47 +0100
Subject: [Image-SIG] adding overlay & text to many files
In-Reply-To: <1231968226.2882.100.camel@blue>
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>
	<1231968226.2882.100.camel@blue>
Message-ID: <200901142304.48223.frederic.mantegazza@gbiloba.org>

On mercredi 14 janvier 2009, Stani wrote:

> And you would get 2 and 3 also for free. For both steps you can use the
> watermark action: once with a fixed translucent white bar image (.png)
> and once with the logo.

Oh, yes, you're right!

> Phatch has already a text action which has done the hard work already as
> well. Just start from that as a base and hack it to use a csv file as
> input instead of a fixed text. If you have played with python and PIL
> already, it should be not that hard.
>
> BTW Phatch can be used as a desktop application but also as a server
> process without a GUI, once you have your action configured.
>
> Phatch is available in Ubuntu Intrepid, Fedora, Suse, Arch Linux, ... in
> the usual repositories. The latest version is 0.1.x (where x are bug fix
> releases which you probably don't need).

Bernard, Phatch is definitly the tool you need to use to solve your 
problem!

And feel free to share your actions, if you write some ;o)

-- 
    Fr?d?ric

    http://www.gbiloba.org

From berankin99 at yahoo.com  Thu Jan 15 04:42:47 2009
From: berankin99 at yahoo.com (Bernard Rankin)
Date: Wed, 14 Jan 2009 19:42:47 -0800 (PST)
Subject: [Image-SIG] adding overlay & text to many files
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>
	<1231968226.2882.100.camel@blue>
Message-ID: <108947.9589.qm@web112214.mail.gq1.yahoo.com>




> Phatch has already a text action which has done the hard work already as
> well. Just start from that as a base and hack it to use a csv file as
> input instead of a fixed text. If you have played with python and PIL
> already, it should be not that hard.
> 

Thanks for the help.

I've taken a look at the tutorial and the existing text.py action.

I am new to Python, and have never used PIL, but it does look do-able..

A couple of questions on the existing text action:

1) Is the the "cache" dict  the only recommended way to maintain state between invocations?
2) What is the "values" method of the Action class in the text.py file?
3) What does the "global" statement in the init() method do, and why is it there?
4) What is line 38 trying to do, and why is it in the if clause?

Also, was this file based on the Watermark code?  Many if the code comments would indicate so. (Running out of Ubuntu Intrepid repo)

Where can I find further docs on the plug-in / action  API?

Thank you
:)



      


From hoffmann at hiskp.uni-bonn.de  Thu Jan 22 21:05:29 2009
From: hoffmann at hiskp.uni-bonn.de (Tim Hoffmann)
Date: Thu, 22 Jan 2009 21:05:29 +0100
Subject: [Image-SIG] 16bit TIFF files: error in Image.putpixel()
Message-ID: <4978D189.4030000@hiskp.uni-bonn.de>

Hello,

I think I found a bug in Image.putpixel() for 16bit images. It writes 
the given x coordinates in bytes, not pixels. i.e. x==0 is lower byte of 
pixel 0, x==1 addresses upper byte of pixel zero. y indices are ok.


The following code shows the problem (using PIL 1.1.6):


im = Image.open("text16black.tif")
     print im.format, im.size, im.mode
     dx = size[0]
	
     for i in range(0, dx):
         # reopening to restore data every time
         im = Image.open("test16black.tif")
         im.putpixel((i, 0), 2)
         print [im.getpixel((x, 0)) for x in range(0, dx)]

output (for completely black test image):

TIFF (9, 15) I;16
[2, 0, 0, 0, 0, 0, 0, 0, 0]
[512, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 2, 0, 0, 0, 0, 0, 0, 0]
[0, 512, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 2, 0, 0, 0, 0, 0, 0]
[0, 0, 512, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 2, 0, 0, 0, 0, 0]
[0, 0, 0, 512, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 2, 0, 0, 0, 0]



Because of range checking based on the pixels I cannot even work around 
supplying the data in an appropriate format. There is no way to address 
the right half of the line.

BTW the same problem occurs in putdata().

Thanks for your comments,
Tim



From fredrik at pythonware.com  Fri Jan 23 14:36:45 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Fri, 23 Jan 2009 14:36:45 +0100
Subject: [Image-SIG] 16bit TIFF files: error in Image.putpixel()
In-Reply-To: <4978D189.4030000@hiskp.uni-bonn.de>
References: <4978D189.4030000@hiskp.uni-bonn.de>
Message-ID: <368a5cd50901230536s7f3b1290l4fd78758b5dce706@mail.gmail.com>

What's the mode for this image?  If it says "I;16" or something
similar, it's not fully supported in 1.1.6, and you may have to
explicitly convert it to mode "I" to work with it.

(there should be an option to open() so you can explicitly tell PIL if
you're prepared to work with experimental modes, really, but with the
current releases, you have to look for them).

</F>

On Thu, Jan 22, 2009 at 9:05 PM, Tim Hoffmann
<hoffmann at hiskp.uni-bonn.de> wrote:
> Hello,
>
> I think I found a bug in Image.putpixel() for 16bit images. It writes the
> given x coordinates in bytes, not pixels. i.e. x==0 is lower byte of pixel
> 0, x==1 addresses upper byte of pixel zero. y indices are ok.
>
>
> The following code shows the problem (using PIL 1.1.6):
>
>
> im = Image.open("text16black.tif")
>    print im.format, im.size, im.mode
>    dx = size[0]
>
>    for i in range(0, dx):
>        # reopening to restore data every time
>        im = Image.open("test16black.tif")
>        im.putpixel((i, 0), 2)
>        print [im.getpixel((x, 0)) for x in range(0, dx)]
>
> output (for completely black test image):
>
> TIFF (9, 15) I;16
> [2, 0, 0, 0, 0, 0, 0, 0, 0]
> [512, 0, 0, 0, 0, 0, 0, 0, 0]
> [0, 2, 0, 0, 0, 0, 0, 0, 0]
> [0, 512, 0, 0, 0, 0, 0, 0, 0]
> [0, 0, 2, 0, 0, 0, 0, 0, 0]
> [0, 0, 512, 0, 0, 0, 0, 0, 0]
> [0, 0, 0, 2, 0, 0, 0, 0, 0]
> [0, 0, 0, 512, 0, 0, 0, 0, 0]
> [0, 0, 0, 0, 2, 0, 0, 0, 0]
>
>
>
> Because of range checking based on the pixels I cannot even work around
> supplying the data in an appropriate format. There is no way to address the
> right half of the line.
>
> BTW the same problem occurs in putdata().
>
> Thanks for your comments,
> Tim
>
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From tibor.arpas at infinit.sk  Fri Jan 23 16:34:41 2009
From: tibor.arpas at infinit.sk (Tibor Arpas)
Date: Fri, 23 Jan 2009 16:34:41 +0100
Subject: [Image-SIG] feature request: 8-bit alpha transparency PNG;s
In-Reply-To: <19a7c8c20901211003k3f18b2f9yd7171af4b0830288@mail.gmail.com>
References: <19a7c8c20901211003k3f18b2f9yd7171af4b0830288@mail.gmail.com>
Message-ID: <19a7c8c20901230734i15f874e9k78553c6044f30bbc@mail.gmail.com>

Hi,
I'm using little attached program to convert my RGB and RGBA PNG's to png8.
But with RGBA pngs it's limited because it uses only GIF-like 1-bit
transparency. Seems that the alpha transparency in 8-bit (and 4-bit) PNG's
is little known or supported. But the main browsers seem to support it
including IE 6.0. One opensource utility which supports writing these is:
http://www.libpng.org/pub/png/apps/pngquant.html .

It would be greate if PIL would also do that.

Have a nice day,
Tibor



import Image
import random
import optparse
import sys

def quantize_and_invert(alpha):
    if alpha <= 64: # You may change the threshold value into something
other.
        return 255
    return 0

im = Image.open(sys.argv[1])

if im.mode=='RGBA':
    print "Warning, converting alpha transparency to 1-bit transparency"
    al = im.split()[3]
    al = Image.eval(al, quantize_and_invert)
    colors=255
else:
    al = None
    colors=256
if im.mode!='RGB':
    im = im.convert('RGB')
im=im.convert('P', palette=Image.ADAPTIVE, colors=colors)
if al: im.paste(255,None,al)
im.save('out.png',optimize=1, transparency=255)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090123/2916def1/attachment-0001.htm>

From odaniil at gmail.com  Sat Jan 24 19:17:27 2009
From: odaniil at gmail.com (Daniil Osokin)
Date: Sat, 24 Jan 2009 21:17:27 +0300
Subject: [Image-SIG] PIL bug report: touble in save jpeg file with
	quality=100 and optimize=True
Message-ID: <b55e68bb0901241017g37f1d285n88b26b982bbf8dcc@mail.gmail.com>

Hello

I have try to save jpeg file with quality=100 and optimize but get this:
-----------------------------------------------------------
Traceback (most recent call last):
  File "D:\temp\Python-PIL-test\pil_test.py", line 29, in <module>
    main("image_test.jpg")
  File "D:\temp\Python-PIL-test\pil_test.py", line 23, in main
    img.save(file_out, "JPEG", quality=100, optimize=True)
  File "C:\Python25\Lib\site-packages\PIL\Image.py", line 1405, in save
    save_handler(self, fp, filename)
  File "C:\Python25\lib\site-packages\PIL\JpegImagePlugin.py", line
409, in _save
    ImageFile._save(im, fp, [("jpeg", (0,0)+im.size, 0, rawmode)])
  File "C:\Python25\Lib\site-packages\PIL\ImageFile.py", line 493, in _save
    raise IOError("encoder error %d when writing image file" % s)
IOError: encoder error -2 when writing image file
-----------------------------------------------------------

But this lines work fine:
img.save(file_out, "JPEG", quality=80, optimize=True)
or
img.save(file_out, "JPEG", quality=100)

PIL version 1.1.6
Python 2.5.4
Windows XP SP3

test files are attached

Thanx
---
Daniil
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image_test.jpg
Type: image/jpeg
Size: 2428090 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090124/3cd6de71/attachment-0001.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pil_test.py
Type: application/octet-stream
Size: 772 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090124/3cd6de71/attachment-0001.obj>

From fredrik at pythonware.com  Sun Jan 25 20:34:42 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sun, 25 Jan 2009 20:34:42 +0100
Subject: [Image-SIG] PIL bug report: touble in save jpeg file with
	quality=100 and optimize=True
In-Reply-To: <b55e68bb0901241017g37f1d285n88b26b982bbf8dcc@mail.gmail.com>
References: <b55e68bb0901241017g37f1d285n88b26b982bbf8dcc@mail.gmail.com>
Message-ID: <368a5cd50901251134q2e51f06bq8e4392bb75679156@mail.gmail.com>

It's a limitation in the libjpeg library that PIL doesn't work around;
the approach described here should still work:

    http://mail.python.org/pipermail/image-sig/1999-August/000816.html

</F>

2009/1/24 Daniil Osokin <odaniil at gmail.com>:
> Hello
>
> I have try to save jpeg file with quality=100 and optimize but get this:
> -----------------------------------------------------------
> Traceback (most recent call last):
>  File "D:\temp\Python-PIL-test\pil_test.py", line 29, in <module>
>    main("image_test.jpg")
>  File "D:\temp\Python-PIL-test\pil_test.py", line 23, in main
>    img.save(file_out, "JPEG", quality=100, optimize=True)
>  File "C:\Python25\Lib\site-packages\PIL\Image.py", line 1405, in save
>    save_handler(self, fp, filename)
>  File "C:\Python25\lib\site-packages\PIL\JpegImagePlugin.py", line
> 409, in _save
>    ImageFile._save(im, fp, [("jpeg", (0,0)+im.size, 0, rawmode)])
>  File "C:\Python25\Lib\site-packages\PIL\ImageFile.py", line 493, in _save
>    raise IOError("encoder error %d when writing image file" % s)
> IOError: encoder error -2 when writing image file
> -----------------------------------------------------------
>
> But this lines work fine:
> img.save(file_out, "JPEG", quality=80, optimize=True)
> or
> img.save(file_out, "JPEG", quality=100)
>
> PIL version 1.1.6
> Python 2.5.4
> Windows XP SP3
>
> test files are attached

From dineshbvadhia at hotmail.com  Mon Jan 26 16:04:49 2009
From: dineshbvadhia at hotmail.com (Dinesh B Vadhia)
Date: Mon, 26 Jan 2009 07:04:49 -0800
Subject: [Image-SIG] Image Mining Library
Message-ID: <COL103-DS1677F254B319D1507870A0A3CA0@phx.gbl>

Hi!  I've asked this question before but didn't get a response so let's try again!  I'm looking for an (open source) Python image mining library to extract information from images eg. to perform Gabor or Tamura or HSV filtering and others.  All help will be appreciated.  Cheers ...

Dinesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090126/eac7444f/attachment.htm>

From tim at timhatch.com  Mon Jan 26 18:29:06 2009
From: tim at timhatch.com (Tim Hatch)
Date: Mon, 26 Jan 2009 09:29:06 -0800
Subject: [Image-SIG] [PIL] ImageColor.getcolor incorrect for grayscale
Message-ID: <ED47389C-A32C-4E81-BC99-CAC23B679E4C@timhatch.com>

I think I've discovered a bug when using named colors, due to using  
integer division too early in getcolor, which makes it disagree with  
the color conversion done in ImagingConvert for a large number of  
colors (106 of the 147 named colors).

To start off, I'm assuming that for grays where r=g=b, that  
getcolor(name, 'L') should return the same as getcolor(name, 'RGB') 
[0].  Integer division done too early rounds down, so grays end up  
slightly too dark in grayscale modes.  Here's the line in question  
(PIL 1.1.6, ImageColor.py line 108):

         return r*299/1000 + g*587/1000 + b*114/1000

All the true grays except black (that is, darkgray, dimgray, gray,  
lightgray, and white) exhibit this issue.  The one that made me notice  
this is that 'white' means 254 instead of 255.  With the attached  
patch, it now behaves sanely, i.e.:

	Image.new('RGB', (1, 1), color).convert('L').getpixel((0, 0)) ==  
Image.new('L', (1, 1), color).getpixel((0, 0))

for all named colors, and

	ImageColor.getcolor(color, 'L') == ImageColor.getcolor(color, 'RGB')[0]

for the grays.

Tim

-------------- next part --------------
A non-text attachment was scrubbed...
Name: imagecolor-rounding.diff
Type: application/octet-stream
Size: 361 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090126/25f81315/attachment.obj>
-------------- next part --------------


From lukas.orsin at infinit.sk  Mon Jan 26 19:34:34 2009
From: lukas.orsin at infinit.sk (Lukas Orsin)
Date: Mon, 26 Jan 2009 19:34:34 +0100
Subject: [Image-SIG] Converting images to palette mode
Message-ID: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>

Hi,
We are converting much images (296,000 tiles) at size 256x256px from RGB to
Palette mode.
After about 200,000 converted images our simple program crashes with
quantization error:

..processing 129.png
..processing 130.png
Traceback (most recent call last):
  File "C:\tmp\to8bit_folders_upgrade.py", line 35, in <module>
    os.path.walk(cesta,start_image_crush, 'blablabla')
  File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
    walk(name, func, arg)
  File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
    walk(name, func, arg)
  File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
    walk(name, func, arg)
  File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
    walk(name, func, arg)
  File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
    walk(name, func, arg)
  File "C:\ActivePython25\lib\ntpath.py", line 328, in walk
    func(arg, top, names)
  File "C:\tmp\to8bit_folders_upgrade.py", line 29, in start_image_crush
    im=im.convert('P', palette=Image.ADAPTIVE, colors=colors)
  File "C:\ActivePython25\Lib\site-packages\PIL\Image.py", line 663, in
convert
    im = self.im.quantize(colors)
ValueError: quantization error

Could anybody know whats the matter?
We are running at Windows 2003 R2 SP2 32bit / 3.5 GB RAM
and using os.path.walk funcion.

Lukas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090126/86871626/attachment.htm>

From fredrik at pythonware.com  Tue Jan 27 11:53:33 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Tue, 27 Jan 2009 11:53:33 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>
References: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>
Message-ID: <368a5cd50901270253p1957dce1jd2bee18c9d48a6fc@mail.gmail.com>

Does it always stop at the same image, or after roughly the same
number of images?  Have you checked for memory/cpu consumption during
the conversion (in case it's a leak).

</F>

2009/1/26 Lukas Orsin <lukas.orsin at infinit.sk>:
> Hi,
> We are converting much images (296,000 tiles) at size 256x256px from RGB to
> Palette mode.
> After about 200,000 converted images our simple program crashes with
> quantization error:
>
> ..processing 129.png
> ..processing 130.png
> Traceback (most recent call last):
>   File "C:\tmp\to8bit_folders_upgrade.py", line 35, in <module>
>     os.path.walk(cesta,start_image_crush, 'blablabla')
>   File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
>     walk(name, func, arg)
>   File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
>     walk(name, func, arg)
>   File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
>     walk(name, func, arg)
>   File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
>     walk(name, func, arg)
>   File "C:\ActivePython25\lib\ntpath.py", line 334, in walk
>     walk(name, func, arg)
>   File "C:\ActivePython25\lib\ntpath.py", line 328, in walk
>     func(arg, top, names)
>   File "C:\tmp\to8bit_folders_upgrade.py", line 29, in start_image_crush
>     im=im.convert('P', palette=Image.ADAPTIVE, colors=colors)
>   File "C:\ActivePython25\Lib\site-packages\PIL\Image.py", line 663, in
> convert
>     im = self.im.quantize(colors)
> ValueError: quantization error
>
> Could anybody know whats the matter?
> We are running at Windows 2003 R2 SP2 32bit / 3.5 GB RAM
> and using os.path.walk funcion.
>
> Lukas
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>
>

From lukas.orsin at infinit.sk  Tue Jan 27 17:44:43 2009
From: lukas.orsin at infinit.sk (Lukas Orsin)
Date: Tue, 27 Jan 2009 17:44:43 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <368a5cd50901270253p1957dce1jd2bee18c9d48a6fc@mail.gmail.com>
References: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>
	<368a5cd50901270253p1957dce1jd2bee18c9d48a6fc@mail.gmail.com>
Message-ID: <6fca00fa0901270844g7a5874dcw72f14382b041cd70@mail.gmail.com>

Thanks for answers.
I am sure no image is broken. If I run my program once again, previously
stopped image is processed with no error and no exception.
I tested conversion a few times with many little images on two mashines
(both 32bit Win 2003 server) and it crashed after allocation 50 - 60 MB of
memory.
A preliminary test on 64bit Win 2008 server ended well with 200,000
converted images at memory allocation  250 MB in the end.
We think problem takes about memory allocation at mass conversion.

My script is attached.
(usage: python to8bit_folders_upgrade.py c:\temp\images)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090127/4d056e06/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: to8bit_folders_upgrade.py
Type: text/x-python
Size: 1850 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090127/4d056e06/attachment.py>

From gregor at kopka.net  Wed Jan 28 00:29:49 2009
From: gregor at kopka.net (Gregor Kopka)
Date: Wed, 28 Jan 2009 00:29:49 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <6fca00fa0901270844g7a5874dcw72f14382b041cd70@mail.gmail.com>
References: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>	<368a5cd50901270253p1957dce1jd2bee18c9d48a6fc@mail.gmail.com>
	<6fca00fa0901270844g7a5874dcw72f14382b041cd70@mail.gmail.com>
Message-ID: <497F98ED.5070203@kopka.net>

Hello Lukas,

could be that you found a memleak somewhere in PIL.

Maybe try to add some /del() /to force cleanup of the images - i'm not 
that sure at the moment but i remember i somewhen have read something 
somewhere about the need to use del(image) to force a correct cleanup in 
a special case.
In case you're curious if i'm right try this version which cleans used 
stuff explicitly via del().

also while looking at it i'm not that sure why you don't check the 
extension by checking against /a[-4:] == ".png" /which i feel like a bit 
more readable compared to your if condition.
Do you want to stop your code from processing files with a "." in the 
name (like 'mypic 02.05.2007.png') ?

Gregor

def start_image_crush(bla, path, files):
    os.chdir(path)
    print '\nACTUAL DIRECTORY: ' + path + '\n'
    for a in files:
        # directory and file filter
        if string.find(a,'.') > 0 and string.split(a,'.')[1]=='png' and len(string.split(a,'.'))==2:
            print 'Trying '+ a
            im = Image.open(a)
            if im.mode!='P':
                print '.....processing'
                if im.mode=='RGBA':
                    print "Warning, converting alpha transparency to 1-bit transparency"
                    al = im.split()[3]
                    nal = Image.eval(al, quantize_and_invert)
                    del(al)
                    colors=255
                else:
                    nal = None
                    colors=256
                if im.mode!='RGB':
                    nim = im.convert('RGB')
                else:
                    nim = im
                del(im)
                try:
                    im=nim.convert('P', palette=Image.ADAPTIVE, colors=colors)
                except ValueError, e:
                    print 'Python memusage: ' + str( memusage() ) + ' !!!'
                    raise e
                del(nim)

                if nal:
                    im.paste(255,None,nal)
                    del(nal)
                im.save(string.split(a,'.')[0]+'.png',optimize=1, transparency=255)
            else: 
                print 'Image still converted. No processing !!!'
            del(im)


Lukas Orsin schrieb:
> Thanks for answers.
> I am sure no image is broken. If I run my program once again, 
> previously stopped image is processed with no error and no exception.
> I tested conversion a few times with many little images on two 
> mashines (both 32bit Win 2003 server) and it crashed after allocation 
> 50 - 60 MB of memory.
> A preliminary test on 64bit Win 2008 server ended well with 200,000 
> converted images at memory allocation  250 MB in the end.
> We think problem takes about memory allocation at mass conversion.
>
> My script is attached.
> (usage: python to8bit_folders_upgrade.py c:\temp\images)
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090128/3cbfe5a9/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gregor.vcf
Type: text/x-vcard
Size: 314 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090128/3cbfe5a9/attachment.vcf>

From charlie at begeistert.org  Wed Jan 28 10:10:22 2009
From: charlie at begeistert.org (Charlie Clark)
Date: Wed, 28 Jan 2009 10:10:22 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <497F98ED.5070203@kopka.net>
References: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>	<368a5cd50901270253p1957dce1jd2bee18c9d48a6fc@mail.gmail.com>
	<6fca00fa0901270844g7a5874dcw72f14382b041cd70@mail.gmail.com>
	<497F98ED.5070203@kopka.net>
Message-ID: <7FC4301F-21A8-4724-88F1-9B261856E61E@begeistert.org>


Am 28.01.2009 um 00:29 schrieb Gregor Kopka:

> also while looking at it i'm not that sure why you don't check the  
> extension by checking against a[-4:] == ".png" which i feel like a  
> bit more readable compared to your if condition.
> Do you want to stop your code from processing files with a "." in  
> the name (like 'mypic 02.05.2007.png') ?


Actually why not rely on os.path to identify things correctly in the  
first place?

Which book still recommends code like this?

if string.find(a,'.') > 0 and string.split(a,'.')[1]=='png' and  
len(string.split(a,'.'))==2

Strings have had methods since Python 1.6!

But os.path.splitext() is probably all you need. Sorry that this won't  
help you find the cause of the crash.

Charlie
--
Charlie Clark
Helmholtzstr. 20
D?sseldorf
D- 40215
Tel: +49-211-938-5360
GSM: +49-178-782-6226




From luko.pn at gmail.com  Thu Jan 29 11:29:28 2009
From: luko.pn at gmail.com (=?ISO-8859-2?B?THVr4bkgT3K5aW4=?=)
Date: Thu, 29 Jan 2009 11:29:28 +0100
Subject: [Image-SIG] Converting images to palette mode
Message-ID: <89f2e6100901290229u3819b8f7x66d9a8a095b318e8@mail.gmail.com>

Thanks for syntax advices ;)
The version with del(image) clause seems to be more successful.
Unfortunately it can?t still run to the end.
Program crashed at 37 % of processing with 98 MB allocated memory.
I am using PIL in 1.1.6 version.
Unfortunately we have not the opportunity to test that batch in Linux,
just in Windows.

PS:
I am posting from another e-mail (luko.pn at gmail.com). Posting from my
main address (lukas.orsin at infinit.sk), powered by Google Apps for
Business, still returns the notification about suspicious header. The
same to my colleague @infinit.sk. Could the admin chech it please ?

From spe.stani.be at gmail.com  Thu Jan 29 17:39:24 2009
From: spe.stani.be at gmail.com (Stani)
Date: Thu, 29 Jan 2009 17:39:24 +0100
Subject: [Image-SIG] adding overlay & text to many files
In-Reply-To: <108947.9589.qm@web112214.mail.gq1.yahoo.com>
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>
	<1231968226.2882.100.camel@blue>
	<108947.9589.qm@web112214.mail.gq1.yahoo.com>
Message-ID: <1233247164.26585.20.camel@blue>

Op woensdag 14-01-2009 om 19:42 uur [tijdzone -0800], schreef Bernard
Rankin:
> 
> 
> > Phatch has already a text action which has done the hard work already as
> > well. Just start from that as a base and hack it to use a csv file as
> > input instead of a fixed text. If you have played with python and PIL
> > already, it should be not that hard.
> > 
> 
> Thanks for the help.
> 
> I've taken a look at the tutorial and the existing text.py action.
> 
> I am new to Python, and have never used PIL, but it does look do-able..
> 
> A couple of questions on the existing text action:
> 
> 1) Is the the "cache" dict  the only recommended way to maintain state between invocations?
Yes.
> 2) What is the "values" method of the Action class in the text.py file?
The values method convert relative values (given by the user) into
absolute values based on image properties (height, width, ...). For
example for a pixel value a user can enter absolute pixels, metrics (cm,
mm, ...) or a percentage. In the end we need absolute pixels to work
with PIL. So metrics are converted to pixels by multiplying with the
dpi. Percentage are calculated by multiplying with the appropriate image
property. For example for a horizontal offset of 5%, we take 5% of the
image width. The font size is based on the average of the image height
and width.

> 3) What does the "global" statement in the init() method do, and why is it there?
Phatch uses a very simple plugin mechanism. It imports all the actions
at startup. In order not to load modules which might not be used and to
keep one action limited to one file, we need a lazy import for modules
which are only necessary when executing the action. Therefore they are
grouped in an init method and global is used to make them globally
available.

> 4) What is line 38 trying to do, and why is it in the if clause?
If no (truetype) font is given, Phatch will use the default PIL font
which AFAIK can not handle unicode (e.g. chinese characters). Therefore
text needs to be converted to ascii. Phatch choses to replace a
character with a ? if it can't be encoded.
> 
> Also, was this file based on the Watermark code?  Many if the code comments would indicate so.
Indeed. Many actions are based on other ones. This is the fastest way to
write new actions.

>  (Running out of Ubuntu Intrepid repo)
If you start developing, please checkout phatch from the bazaar
repository (bzr lp:phatch)
> 
> Where can I find further docs on the plug-in / action  API?
I guess you've read already:
http://photobatch.wikidot.com/writing-actions

The best is to study and understand the source of the other actions and
see them in action.
> 
> Thank you
You're welcome!

Stani


From pander at users.sourceforge.net  Thu Jan 29 17:53:00 2009
From: pander at users.sourceforge.net (Pander)
Date: Thu, 29 Jan 2009 17:53:00 +0100
Subject: [Image-SIG] adding overlay & text to many files
In-Reply-To: <1233247164.26585.20.camel@blue>
References: <lTIPaszB.1231948174.9867510.fma@alezan.org>	<1231968226.2882.100.camel@blue>	<108947.9589.qm@web112214.mail.gq1.yahoo.com>
	<1233247164.26585.20.camel@blue>
Message-ID: <4981DEEC.4080503@users.sourceforge.net>

Hi all,

Here is a way to add text, see attachment. I've extended the test cases
with capitols with diacritics. On the list a question regarding this
arose, so it was good to double check this.

If possible, I would like to have this module included into PIL. Once I
had it revied, but would like to resubmit it again since the PIL list
has become more active again, especially regarding text overlays.

Regards,

Pander


Stani wrote:
> Op woensdag 14-01-2009 om 19:42 uur [tijdzone -0800], schreef Bernard
> Rankin:
>>
>>> Phatch has already a text action which has done the hard work already as
>>> well. Just start from that as a base and hack it to use a csv file as
>>> input instead of a fixed text. If you have played with python and PIL
>>> already, it should be not that hard.
>>>
>> Thanks for the help.
>>
>> I've taken a look at the tutorial and the existing text.py action.
>>
>> I am new to Python, and have never used PIL, but it does look do-able..
>>
>> A couple of questions on the existing text action:
>>
>> 1) Is the the "cache" dict  the only recommended way to maintain state between invocations?
> Yes.
>> 2) What is the "values" method of the Action class in the text.py file?
> The values method convert relative values (given by the user) into
> absolute values based on image properties (height, width, ...). For
> example for a pixel value a user can enter absolute pixels, metrics (cm,
> mm, ...) or a percentage. In the end we need absolute pixels to work
> with PIL. So metrics are converted to pixels by multiplying with the
> dpi. Percentage are calculated by multiplying with the appropriate image
> property. For example for a horizontal offset of 5%, we take 5% of the
> image width. The font size is based on the average of the image height
> and width.
> 
>> 3) What does the "global" statement in the init() method do, and why is it there?
> Phatch uses a very simple plugin mechanism. It imports all the actions
> at startup. In order not to load modules which might not be used and to
> keep one action limited to one file, we need a lazy import for modules
> which are only necessary when executing the action. Therefore they are
> grouped in an init method and global is used to make them globally
> available.
> 
>> 4) What is line 38 trying to do, and why is it in the if clause?
> If no (truetype) font is given, Phatch will use the default PIL font
> which AFAIK can not handle unicode (e.g. chinese characters). Therefore
> text needs to be converted to ascii. Phatch choses to replace a
> character with a ? if it can't be encoded.
>> Also, was this file based on the Watermark code?  Many if the code comments would indicate so.
> Indeed. Many actions are based on other ones. This is the fastest way to
> write new actions.
> 
>>  (Running out of Ubuntu Intrepid repo)
> If you start developing, please checkout phatch from the bazaar
> repository (bzr lp:phatch)
>> Where can I find further docs on the plug-in / action  API?
> I guess you've read already:
> http://photobatch.wikidot.com/writing-actions
> 
> The best is to study and understand the source of the other actions and
> see them in action.
>> Thank you
> You're welcome!
> 
> Stani
> 
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig

-------------- next part --------------
A non-text attachment was scrubbed...
Name: ImageText.py
Type: text/x-python
Size: 15370 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090129/12d2fb96/attachment-0001.py>

From fbuchinger at gmail.com  Wed Jan 28 20:07:16 2009
From: fbuchinger at gmail.com (Franz Buchinger)
Date: Wed, 28 Jan 2009 20:07:16 +0100
Subject: [Image-SIG] Determine byte offset of JPEG SOI marker
Message-ID: <a07b86220901281107s9c8f52cld55b7b9f76a86ae2@mail.gmail.com>

Hi,

I want to implement some sort of "post header checksum" for JPEG images,
i.e. the checksum should only change if the actual image data was altered,
EXIF/IPTC metadata modifications should have no effect. My plan to do this
is to scan for the JPEG SOI marker, read n bytes after this marker and
calcultate an MD5/SHA checksum for this.

Can PIL tell me the byte offset of this marker?

Greetings,

Franz
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090128/b6165a48/attachment.htm>

From fredrik at pythonware.com  Thu Jan 29 19:52:19 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Thu, 29 Jan 2009 19:52:19 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <89f2e6100901290229u3819b8f7x66d9a8a095b318e8@mail.gmail.com>
References: <89f2e6100901290229u3819b8f7x66d9a8a095b318e8@mail.gmail.com>
Message-ID: <368a5cd50901291052pc660e9scbb9efb2c44ac010@mail.gmail.com>

> I am posting from another e-mail (luko.pn at gmail.com). Posting from my
> main address (lukas.orsin at infinit.sk), powered by Google Apps for
> Business, still returns the notification about suspicious header. The
> same to my colleague @infinit.sk. Could the admin chech it please ?

Looks like the mailman filter thinks that your messages are to HTML-y
for the list.  Try switching off "rich formatting" and see if that
helps.

(HTML mails used to mess up the archives, iirc, but that might have been fixed).

</F>

From charlie at begeistert.org  Thu Jan 29 19:56:57 2009
From: charlie at begeistert.org (Charlie Clark)
Date: Thu, 29 Jan 2009 19:56:57 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <89f2e6100901290229u3819b8f7x66d9a8a095b318e8@mail.gmail.com>
References: <89f2e6100901290229u3819b8f7x66d9a8a095b318e8@mail.gmail.com>
Message-ID: <2210705F-1D4A-4803-AFC0-EA2C3A8430ED@begeistert.org>


Am 29.01.2009 um 11:29 schrieb Luk?? Or?in:

> Thanks for syntax advices ;)

You're welcome. I've learned to appreciate that good Python code is  
usually compact and legible and that I'm probably doing something  
wrong when it isn't!

> The version with del(image) clause seems to be more successful.
> Unfortunately it can?t still run to the end.
> Program crashed at 37 % of processing with 98 MB allocated memory.
> I am using PIL in 1.1.6 version.
> Unfortunately we have not the opportunity to test that batch in Linux,
> just in Windows.

I'd be happy to run a test on my machine (Mac OS) if you could send a  
couple of sample image files to me privately.

> PS:
> I am posting from another e-mail (luko.pn at gmail.com). Posting from my
> main address (lukas.orsin at infinit.sk), powered by Google Apps for
> Business, still returns the notification about suspicious header. The
> same to my colleague @infinit.sk. Could the admin chech it please ?


The problem is almost certainly with Google - what is the precise  
error message?

Charlie
--
Charlie Clark
Helmholtzstr. 20
D?sseldorf
D- 40215
Tel: +49-211-938-5360
GSM: +49-178-782-6226




From ben at ekran.org  Fri Jan 30 00:06:13 2009
From: ben at ekran.org (B. Bogart)
Date: Thu, 29 Jan 2009 15:06:13 -0800
Subject: [Image-SIG] Issues generating large PNGs
Message-ID: <49823665.3060905@ekran.org>

Hello all,

I tried to create a 17600x2400 pixel PNG in PIL.

The image appears wrong, and identify reports:

identify: Ignoring bad adaptive filter type `montage.png' @
png.c/PNGWarningHandler/1421.
identify: Extra compressed data `montage.png' @ png.c/PNGErrorHandler/1404.
identify: Corrupt image `montage.png' @ png.c/ReadPNGImage/2892.

If I change the output file to tif then everything works:

montage.tif TIFF 17600x2400 17600x2400+0+0 8-bit DirectClass 120.8mb

Is there an upper limit to the size of PNG files?

.b.

From tim at timhatch.com  Fri Jan 30 00:44:34 2009
From: tim at timhatch.com (Tim Hatch)
Date: Thu, 29 Jan 2009 15:44:34 -0800
Subject: [Image-SIG] Issues generating large PNGs
In-Reply-To: <49823665.3060905@ekran.org>
References: <49823665.3060905@ekran.org>
Message-ID: <84C66D0A-E11C-43CB-AF37-1314D407942B@timhatch.com>

> Is there an upper limit to the size of PNG files?

How big is yours?  I was able to create a 17600x2400 one containing  
just a single color and it works with identify and Preview.app so it's  
not a dimension issue.

Tim

From macginitie at charter.net  Sat Jan 31 00:51:01 2009
From: macginitie at charter.net (Andrew MacGinitie)
Date: Fri, 30 Jan 2009 18:51:01 -0500
Subject: [Image-SIG] installing PIL from source kit for Python 2.6.1
Message-ID: <49839265.8090707@charter.net>

...can it be done?

another question: why does the PIL installer for windows demand that 
Python be "registered"? Can someone fix that (is the installer's source 
available)?

tia,
Andrew

From Chris.Barker at noaa.gov  Sat Jan 31 01:17:34 2009
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Fri, 30 Jan 2009 16:17:34 -0800
Subject: [Image-SIG] installing PIL from source kit for Python 2.6.1
In-Reply-To: <49839265.8090707@charter.net>
References: <49839265.8090707@charter.net>
Message-ID: <4983989E.4060303@noaa.gov>

Andrew MacGinitie wrote:
> another question: why does the PIL installer for windows demand that 
> Python be "registered"?

It's Windows that's demanding that. I think you have to pay a bunch of 
money to get officially registered. It's analogous to SSL site certificates.


-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker at noaa.gov

From fredrik at pythonware.com  Sat Jan 31 21:21:21 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 31 Jan 2009 21:21:21 +0100
Subject: [Image-SIG] installing PIL from source kit for Python 2.6.1
In-Reply-To: <49839265.8090707@charter.net>
References: <49839265.8090707@charter.net>
Message-ID: <368a5cd50901311221t29367cddkc792744161ffbc17@mail.gmail.com>

PIL's using Python's standard installation framework (distutils), and
that installer requires Python to be registered (that is, linked to by
the appropriate entries in the registry) for the installer to be able
to figure out which Python versions you have on the machine, and where
they are.

If you don't want to use the standard installer, you can just unzip
the file and put the contents somewhere on the path.

</F>

On Sat, Jan 31, 2009 at 12:51 AM, Andrew MacGinitie
<macginitie at charter.net> wrote:
> ...can it be done?
>
> another question: why does the PIL installer for windows demand that Python
> be "registered"? Can someone fix that (is the installer's source available)?
>
> tia,
> Andrew
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From fredrik at pythonware.com  Sat Jan 31 21:22:35 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 31 Jan 2009 21:22:35 +0100
Subject: [Image-SIG] installing PIL from source kit for Python 2.6.1
In-Reply-To: <4983989E.4060303@noaa.gov>
References: <49839265.8090707@charter.net> <4983989E.4060303@noaa.gov>
Message-ID: <368a5cd50901311222h76dbcfe2uc49dd417add6072f@mail.gmail.com>

I suspect he's referring to the distutils installer (which is a part
of standard Python), not the signing procedure.

(the installer is signed, iirc).

</F>

On Sat, Jan 31, 2009 at 1:17 AM, Christopher Barker
<Chris.Barker at noaa.gov> wrote:
> Andrew MacGinitie wrote:
>>
>> another question: why does the PIL installer for windows demand that
>> Python be "registered"?
>
> It's Windows that's demanding that. I think you have to pay a bunch of money
> to get officially registered. It's analogous to SSL site certificates.
>
>
> -Chris
>
>
> --
> Christopher Barker, Ph.D.
> Oceanographer
>
> Emergency Response Division
> NOAA/NOS/OR&R            (206) 526-6959   voice
> 7600 Sand Point Way NE   (206) 526-6329   fax
> Seattle, WA  98115       (206) 526-6317   main reception
>
> Chris.Barker at noaa.gov
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From fredrik at pythonware.com  Sat Jan 31 21:25:08 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 31 Jan 2009 21:25:08 +0100
Subject: [Image-SIG] installing PIL from source kit for Python 2.6.1
In-Reply-To: <49839265.8090707@charter.net>
References: <49839265.8090707@charter.net>
Message-ID: <368a5cd50901311225l6ee0cc77q4cf4c9964e283fa5@mail.gmail.com>

On Sat, Jan 31, 2009 at 12:51 AM, Andrew MacGinitie
<macginitie at charter.net> wrote:

> ...can it be done?

There's an installer for Python 2.6 available from here:

    http://effbot.org/downloads/#pil

which seem to indicate that it can be done.  If you insist on building
it yourself, see the "Build instructions (all platforms)" and
"Additional notes for Windows" sections of the README.

</F>

From lukas.orsin at infinit.sk  Wed Jan 28 20:05:02 2009
From: lukas.orsin at infinit.sk (Lukas Orsin)
Date: Wed, 28 Jan 2009 20:05:02 +0100
Subject: [Image-SIG] Converting images to palette mode
In-Reply-To: <6fca00fa0901270844g7a5874dcw72f14382b041cd70@mail.gmail.com>
References: <6fca00fa0901261034h7a60f678t6f75f50c1961a285@mail.gmail.com>
	<368a5cd50901270253p1957dce1jd2bee18c9d48a6fc@mail.gmail.com>
	<6fca00fa0901270844g7a5874dcw72f14382b041cd70@mail.gmail.com>
Message-ID: <6fca00fa0901281105s54c0a99cgfaad871128c0eeef@mail.gmail.com>

Thanks for syntax advices ;)
The version with del(image) clause seems to be more successful.
Unfortunatelly it can?t still run to the end.
Program crashed at 37 % of processing with 98 MB allocated memory.
I am using PIL in 1.1.6 version.
Unfortunatelly we can't test that batch in Linux, just in Windows.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090128/5b5a5168/attachment-0001.htm>

From H.Zahiri at curtin.edu.au  Sat Jan 31 16:13:39 2009
From: H.Zahiri at curtin.edu.au (Hani Zahiri)
Date: Sun, 1 Feb 2009 00:13:39 +0900
Subject: [Image-SIG] Problem with converting GeoTIFF to Numpy array
	(ValueError: unrecognised mode)
Message-ID: <82200558F6DE2C479D381D3000D1551C01C7942B@EXMSK1.staff.ad.curtin.edu.au>

Hi

I am trying to read a GeoTIFF and convert it to a Numpy array. Though, I am able to open the image with no problem, however I am keep getting 'ValueError: unrecognised mode' anytime I try to use any methods from PIL's classes. My lines and results are like this:


>>import Numpy
>>from PIL import Image
>>im = Image.open('dem.TIF')
>>print im.format, im.size, im.mode
>> TIFF (6000, 6000) I;16S
>>dem = fromstring(im.tostring(),'int16')
....
....
ValueError: unrecognised mode

The problem must be because of mode of my data (I;16S) and I don't know how to read it.
I really appreciate any help with this.

Many Thanks
Cheers,

Hani
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090201/f3435d8c/attachment-0001.htm>

From dineshbvadhia at hotmail.com  Sat Jan 31 19:57:42 2009
From: dineshbvadhia at hotmail.com (Dinesh B Vadhia)
Date: Sat, 31 Jan 2009 10:57:42 -0800
Subject: [Image-SIG] Image Search at Xyggy
Message-ID: <COL103-DS174787A5955FF735F682ECA3C70@phx.gbl>

Hi!  This group might be interested in what we are doing at Xyggy (www.xyggy.com), a Bay Area search startup that is building item-based search services as well as an online platform for developers.  We have started to showcase the technology with 5 demos including image search which allows you to find similar images by dragging and dropping one or more images into the search oval.  Go have a play.

The image demo is based on texture (Gabor and Tamura) and color (HSV) features.  The Xyggy search engine can operate on features ranging from the simple to the complex (ie. color, texture, luminosity, invariance, edge detection and so on).  The features used for any particular image search service are in the developer's control.  Another advantage is that an organizations primary data, in this case images, reside on their site to ensure data protection and privacy.

We use Python for all prototype development and data processing and is superb for both these functions.

Cheers ...

Dinesh

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090131/e6c04262/attachment-0001.htm>

From cjlesh at gmail.com  Fri Jan 30 16:24:22 2009
From: cjlesh at gmail.com (charlie)
Date: Fri, 30 Jan 2009 10:24:22 -0500
Subject: [Image-SIG] Windows PIL installer question
Message-ID: <7e7202260901300724r524e3461m4dc5a4c7c040f50d@mail.gmail.com>

Is there any way to run the *PIL* installer from the command line on
Windows in 'silent' mode, without displaying the install screens or
requiring any user interaction?

I figured out how to do this with MSIs (the Python installer) and with
installers created with Inno setup (use innounp), but can't quite figure out
how to do this with the PIL installer, or if it is even possible.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090130/f731dd34/attachment-0001.htm>

From YHWANG1 at PARTNERS.ORG  Fri Jan 30 16:00:41 2009
From: YHWANG1 at PARTNERS.ORG (Hwang, Youngha,Ph.D.)
Date: Fri, 30 Jan 2009 10:00:41 -0500
Subject: [Image-SIG] installation error of PIL1.1.6
Message-ID: <52368789F9A31D43B7D335E33F45868201B1FE95@PHSXMB24.partners.org>


Hi, could someone explain to me what the following error means?

Thanks, Youngha.

 

building '_imagingtk' extension
creating build/temp.linux-i686-2.5/Tk
gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O3 -Wall
-Wstrict-prototypes -fPIC -I/usr/include/freetype2 -IlibImaging
-I/programs/l/python/2.5.2/include -I/usr/local/include -I/usr/include
-I/programs/l/python/2.5.2/include/python2.5 -c _imagingtk.c -o
build/temp.linux-i686-2.5/_imagingtk.o
_imagingtk.c:20:16: error: tk.h: No such file or directory
_imagingtk.c:23: error: expected ?)? before ?*? token
_imagingtk.c:31: error: expected specifier-qualifier-list before ?Tcl_Interp?
_imagingtk.c: In function ?_tkinit?:
_imagingtk.c:37: error: ?Tcl_Interp? undeclared (first use in this function)
_imagingtk.c:37: error: (Each undeclared identifier is reported only once
_imagingtk.c:37: error: for each function it appears in.)
_imagingtk.c:37: error: ?interp? undeclared (first use in this function)
_imagingtk.c:45: error: expected expression before ?)? token
_imagingtk.c:51: error: ?TkappObject? has no member named ?interp?
_imagingtk.c:55: warning: implicit declaration of function ?TkImaging_Init?
error: command 'gcc' failed with exit status 1


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.


From fredrik at pythonware.com  Sat Jan 31 23:20:44 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 31 Jan 2009 23:20:44 +0100
Subject: [Image-SIG] Problem with converting GeoTIFF to Numpy array
	(ValueError: unrecognised mode)
In-Reply-To: <82200558F6DE2C479D381D3000D1551C01C7942B@EXMSK1.staff.ad.curtin.edu.au>
References: <82200558F6DE2C479D381D3000D1551C01C7942B@EXMSK1.staff.ad.curtin.edu.au>
Message-ID: <368a5cd50901311420q7d29be90r18ddbac773c441f1@mail.gmail.com>

PIL have only limited support for the "I;16" modules; try converting
the image to mode "I" before you pass it on to NumPy.

</F>

2009/1/31 Hani Zahiri <H.Zahiri at curtin.edu.au>:
> Hi
>
> I am trying to read a GeoTIFF and convert it to a Numpy array. Though, I am
> able to open the image with no problem, however I am keep getting
> 'ValueError: unrecognised mode' anytime I try to use any methods from PIL's
> classes. My lines and results are like this:
>
>
>>>import Numpy
>>>from PIL import Image
>>>im = Image.open('dem.TIF')
>>>print im.format, im.size, im.mode
>>> TIFF (6000, 6000) I;16S
>>>dem = fromstring(im.tostring(),'int16')
> ....
> ....
> ValueError: unrecognised mode
>
> The problem must be because of mode of my data (I;16S) and I don't know how
> to read it.
> I really appreciate any help with this.
>
> Many Thanks
> Cheers,
>
> Hani
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>
>

From fredrik at pythonware.com  Sat Jan 31 23:26:38 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 31 Jan 2009 23:26:38 +0100
Subject: [Image-SIG] installation error of PIL1.1.6
In-Reply-To: <52368789F9A31D43B7D335E33F45868201B1FE95@PHSXMB24.partners.org>
References: <52368789F9A31D43B7D335E33F45868201B1FE95@PHSXMB24.partners.org>
Message-ID: <368a5cd50901311426s1675348ctcf8c806a119e3501@mail.gmail.com>

You're missing the Tcl/Tk developer files; see e.g.

  http://mail.python.org/pipermail/image-sig/2008-December/005308.html

  "This probably means that you have enough of Tcl/Tk installed for the
  installation process to think it's worth building the Tkinter support,
  but, as Karsten points out, not enough of it to actually be able to
  build that support.  To fix this, either install tcl-dev/tk-dev (or
  whatever the packages are called on your machine), or tweak setup.py
  to skip that part of the process."

</F>

PS. Next time you get an unexpected error message, try googling for a
representative part of the message.  Chances are that you're not the
first one that's seen the same message.

On Fri, Jan 30, 2009 at 4:00 PM, Hwang, Youngha,Ph.D.
<YHWANG1 at partners.org> wrote:
>
> Hi, could someone explain to me what the following error means?
>
> Thanks, Youngha.
>
>
>
> building '_imagingtk' extension
> creating build/temp.linux-i686-2.5/Tk
> gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O3 -Wall
> -Wstrict-prototypes -fPIC -I/usr/include/freetype2 -IlibImaging
> -I/programs/l/python/2.5.2/include -I/usr/local/include -I/usr/include
> -I/programs/l/python/2.5.2/include/python2.5 -c _imagingtk.c -o
> build/temp.linux-i686-2.5/_imagingtk.o
> _imagingtk.c:20:16: error: tk.h: No such file or directory
> _imagingtk.c:23: error: expected ')' before '*' token
> _imagingtk.c:31: error: expected specifier-qualifier-list before 'Tcl_Interp'
> _imagingtk.c: In function '_tkinit':
> _imagingtk.c:37: error: 'Tcl_Interp' undeclared (first use in this function)
> _imagingtk.c:37: error: (Each undeclared identifier is reported only once
> _imagingtk.c:37: error: for each function it appears in.)
> _imagingtk.c:37: error: 'interp' undeclared (first use in this function)
> _imagingtk.c:45: error: expected expression before ')' token
> _imagingtk.c:51: error: 'TkappObject' has no member named 'interp'
> _imagingtk.c:55: warning: implicit declaration of function 'TkImaging_Init'
> error: command 'gcc' failed with exit status 1
>
>
> The information in this e-mail is intended only for the person to whom it is
> addressed. If you believe this e-mail was sent to you in error and the e-mail
> contains patient information, please contact the Partners Compliance HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in error
> but does not contain patient information, please contact the sender and properly
> dispose of the e-mail.
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From fredrik at pythonware.com  Sat Jan 31 23:30:56 2009
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 31 Jan 2009 23:30:56 +0100
Subject: [Image-SIG] Windows PIL installer question
In-Reply-To: <7e7202260901300724r524e3461m4dc5a4c7c040f50d@mail.gmail.com>
References: <7e7202260901300724r524e3461m4dc5a4c7c040f50d@mail.gmail.com>
Message-ID: <368a5cd50901311430n9541521r321acb82d7073ce1@mail.gmail.com>

Hint:

C:\>unzip -v \effbot.org\downloads\PIL-1.1.5.win32-py2.5.exe
Archive: /effbot.org/downloads/PIL-1.1.5.win32-py2.5.exe
warning [/effbot.org/downloads/PIL-1.1.5.win32-py2.5.exe]:  62010
extra bytes at beginning or within zipfile
  (attempting to process anyway)
 Length   Method    Size  Ratio   Date   Time   CRC-32    Name
--------  ------  ------- -----   ----   ----   ------    ----
       5  Defl:N        7 -40%  06-04-24 14:17  34d6cc96  PLATLIB/PIL.pth
   12658  Defl:N     3006  76%  05-04-06 10:13  3b4dbaad
PLATLIB/PIL/ArgImagePlugin.py
    3470  Defl:N     1362  61%  05-04-06 10:13  94cb46df
PLATLIB/PIL/BdfFontFile.py
    6914  Defl:N     2182  68%  05-04-06 10:13  c2710661  PLATLIB/PIL/BmpImag

(etc)

This is true for all package installers built w. distutils.

(You can either use your favourite command-line unzip utility, or even
write a simple installation script using Python's zipfile module).

</F>

2009/1/30 charlie <cjlesh at gmail.com>:
> Is there any way to run the PIL installer from the command line on
> Windows in 'silent' mode, without displaying the install screens or
> requiring any user interaction?
>
> I figured out how to do this with MSIs (the Python installer) and with
> installers created with Inno setup (use innounp), but can't quite figure out
> how to do this with the PIL installer, or if it is even possible.
>
>
>
>
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>
>

From evg.salmin at gmail.com  Sat Jan 31 23:39:47 2009
From: evg.salmin at gmail.com (Evgeny Salmin)
Date: Sun, 1 Feb 2009 04:39:47 +0600
Subject: [Image-SIG]  Converting images to palette mode
Message-ID: <d7f793bd0901311439i689fefaao19b3055376c4f255@mail.gmail.com>

Hello,

I made some basic checks in this problem (quantize error) and have following
results now:

I've added gc debugging to a script, tried to force collection, but at the
time when quantize error happens gc says that there is no leaks.
So I've decided to check C implementation of quantize() function and wrote
small utility that uses C part of PIL for image quantization. Library is too
old, so there is no leaks in quantize() of course :)
Debugging of script shows that after processing some files malloc returns 0
(usually it in Quant.c : 1265). Sometimes I get python MemoryError exception
after receiving "quantization error", sometimes not. As for memory
allocation process it looks like: 7 MB (at start) -->  processing --> 50 MB
--> seems gc begins working here --> 7 MB -> repeat N times -> quantization
error.
Now I have some assumptions about it:

a) Possible memory leak somewhere in split/eval/paste functions. I can check
them, but I don't think it will give any results.
b) Leak somewhere in a glue code. Cycle references/memmove ?

I have similiar subjects at other mailing lists, maybe they will give some
ideas:
http://mail.python.org/pipermail/image-sig/2007-December/004720.html
http://mail.python.org/pipermail/image-sig/1999-April/000713.html
http://www.daniweb.com/forums/thread133133.html

How will be better to track an allocation/deallocation of Imaging objects?
Anyone have C samples of using C part library of PIL? I'd like to rewrite
python script that causes an error in C to make sure that there is no leaks
at C part, but have to work with png loading/decoding manually, so an
example will be great and help to save time which I spend on examination C
API.

Any ideas?


Regards, Evgeny.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/image-sig/attachments/20090201/fac02e57/attachment.htm>