From fredrik at pythonware.com  Sun Dec  2 14:30:11 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sun, 02 Dec 2007 14:30:11 +0100
Subject: [Image-SIG] Is getdata() or load() more efficient for accessing
	pixels?
In-Reply-To: <bda7ac6d0711301042m70bad8afh3e3a54b4e31b78d2@mail.gmail.com>
References: <bda7ac6d0711301042m70bad8afh3e3a54b4e31b78d2@mail.gmail.com>
Message-ID: <fiuc12$h1r$1@ger.gmane.org>

jorma kala wrote:

> I read that the getpixel function of the PIL library is relatively slow.
> Which alternative is therefore faster and more efficient, accesing the 
> result of getdata() as a flat list or
> using  the function load(), and accessing pixels through the resulting  
> 2-dimensional array?

I suggest benchmarking this for your specific use case.

</F>


From fredrik at pythonware.com  Sun Dec  2 14:37:13 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sun, 02 Dec 2007 14:37:13 +0100
Subject: [Image-SIG] Find camera information from JPG image
In-Reply-To: <5b6eb1490711300658w2b073a61r7639f830f54de4fa@mail.gmail.com>
References: <5b6eb1490711300658w2b073a61r7639f830f54de4fa@mail.gmail.com>
Message-ID: <fiuce8$i6o$1@ger.gmane.org>

George LeCompte wrote:

> How do I recover camera information such as shtter speed, focal length
> from the header of a JPG image?

Assuming a recent version of PIL, you can use the _getexif() method on 
JPEG images to get the raw EXIF information, and use the dictionaries in 
the ExifTags module to map from EXIT codes and back (alternatively, look 
up the keys in the EXIF specification).  Example:

 >>> import Image, ExifTags
 >>> im = Image.open("im001.jpg")
 >>> im._getexif().keys()
[36864, 37121, 37122, 36867, 36868, 37381, 37510, 37383, ...
 >>> for key, value in sorted(im._getexif().items()):
...     print key, ExifTags.TAGS.get(key), value
...
271 Make Canon
272 Model Canon PowerShot A85
274 Orientation 1
282 XResolution (180, 1)
283 YResolution (180, 1)
296 ResolutionUnit 2
306 DateTime 2005:08:14 10:24:05
531 YCbCrPositioning 1
33434 ExposureTime (1, 160)
33437 FNumber (48, 10)
34665 ExifOffset 196
36864 ExifVersion 0220
36867 DateTimeOriginal 2005:08:14 10:24:05
36868 DateTimeDigitized 2005:08:14 10:24:05
...

</F>


From scarfboy at gmail.com  Mon Dec  3 16:33:33 2007
From: scarfboy at gmail.com (Bart)
Date: Mon, 3 Dec 2007 16:33:33 +0100
Subject: [Image-SIG] Image-SIG Digest, Vol 56, Issue 2
In-Reply-To: <mailman.22.1196679610.26280.image-sig@python.org>
References: <mailman.22.1196679610.26280.image-sig@python.org>
Message-ID: <b71082d80712030733v67525ba2t3a22e35ad4f25864@mail.gmail.com>

I've noticed that the lens type can be fetched too, but the method
depends on the camera manufacturer (so basically on the make field).

For example, for canon, the lenses are enumerated, see:
http://www.gvsoft.homedns.org/exif/lens-type.html
http://www.sno.phy.queensu.ca/~phil/exiftool/TagNames/Canon.html#CameraInfo

...not that I've tried this (who knows, maybe such non-core tags are
stripped by some software so are only guaranteed on camera raws), or
know of camera/lense mappings that are authoritative, or include other
companies.

--Bart

From Jim.Vickroy at noaa.gov  Mon Dec  3 18:31:20 2007
From: Jim.Vickroy at noaa.gov (jv)
Date: Mon, 03 Dec 2007 10:31:20 -0700
Subject: [Image-SIG] How do i import text file which has NaNs ?
In-Reply-To: <2b125de60711212130s4c99ede9jcc3775cbb7f45bbf@mail.gmail.com>
References: <2b125de60711212130s4c99ede9jcc3775cbb7f45bbf@mail.gmail.com>
Message-ID: <004101c835d2$51ae4980$6500a8c0@questar>

 

 

  _____  

From: image-sig-bounces at python.org [mailto:image-sig-bounces at python.org] On
Behalf Of raghuram narasimhan
Sent: Wednesday, November 21, 2007 10:31 PM
To: image-sig at python.org
Subject: [Image-SIG] How do i import text file which has NaNs ?

 

Hi all,

 

I have converted an ENVI image file into ASCII.

 

The ASCII file has NaNs or NoData in them.

 

I am not able to read the file into python since it is not able to recognize
NaN.

 

I'm not an experience python user, and hence i guess i'm not able to figure
it out.

 

Does anyone have any idea as to how i can import an ASCII file with NaNs in
them ?

[jv] I'm not sure this is applicable, but have you looked at the numpy
extension package which has support for NaN?

 

Hoping to hear from you,

 

 Raghu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/image-sig/attachments/20071203/7cf4aebd/attachment.htm 

From bbaxter at wadsworth.org  Mon Dec  3 16:28:13 2007
From: bbaxter at wadsworth.org (William Baxter)
Date: Mon, 03 Dec 2007 10:28:13 -0500
Subject: [Image-SIG] GifImagePlugin
Message-ID: <4754208D.3133523A@wadsworth.org>

Does anyone have a copy of GifImagePlugin.py that can handle GIF89a?


IMPORTANT NOTICE: This e-mail and any attachments may contain
confidential or sensitive information which is, or may be, legally
privileged or otherwise protected by law from further disclosure.  It
is intended only for the addressee.  If you received this in error or
from someone who was not authorized to send it to you, please do not
distribute, copy or use it or any attachments.  Please notify the
sender immediately by reply e-mail and delete this from your
system. Thank you for your cooperation.



From Chris.Barker at noaa.gov  Wed Dec  5 18:37:55 2007
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Wed, 05 Dec 2007 09:37:55 -0800
Subject: [Image-SIG] How do i import text file which has NaNs ?
In-Reply-To: <004101c835d2$51ae4980$6500a8c0@questar>
References: <2b125de60711212130s4c99ede9jcc3775cbb7f45bbf@mail.gmail.com>
	<004101c835d2$51ae4980$6500a8c0@questar>
Message-ID: <4756E1F3.1040101@noaa.gov>


> The ASCII file has NaNs or NoData in them.

> I am not able to read the file into python since it is not able to 
> recognize NaN.

> */[jv] I?m not sure this is applicable, but have you looked at the numpy 
> extension package which has support for NaN?/*

numpy does support NaN, and it's an excellent choice for this kind of 
data anyway. It's also easy to turn a numpy array into a PIL image.

However, I'm not sure it supports NaN text literals, and, if it does, 
it's probably not platform independent, as it uses the system library 
scanf().

I would check out numpy, but you may have to write the file scanning 
code yourself, and translate the NaN.

An example might help us get other ideas. For instance, what is the 
ascii representation of NaN? "NaN", or something else?

What system are you running on? For instance, on OS-X (Python 2.5, 
compiled with gcc4.0.1), this works:

 >>> float("NaN")
nan

But I don't think that works on the stock Windows Python. However, there 
may be another literal that does. Try:

 >>> import numpy as N
 >>> str(N.nan)
'nan'

and it might tell you the literal on your system.

-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 Chris.Barker at noaa.gov  Wed Dec  5 18:50:45 2007
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Wed, 05 Dec 2007 09:50:45 -0800
Subject: [Image-SIG] Is getdata() or load() more efficient for accessing
 pixels?
In-Reply-To: <bda7ac6d0711301042m70bad8afh3e3a54b4e31b78d2@mail.gmail.com>
References: <bda7ac6d0711301042m70bad8afh3e3a54b4e31b78d2@mail.gmail.com>
Message-ID: <4756E4F5.3040607@noaa.gov>

jorma kala wrote:
> I read that the getpixel function of the PIL library is relatively slow.

Depending on your needs, numpy may be a good bet:

http://jehiah.cz/archive/creating-images-with-numpy

Be sure to look at the comment below, demonstrating the new 
Image.fromarray() function.

And you can convert a PIL image to a numpy array with no data copying:

arr = numpy.asarray(PIL_image)

-CHB

-- 
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  Wed Dec  5 19:55:55 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Wed, 05 Dec 2007 19:55:55 +0100
Subject: [Image-SIG] GifImagePlugin
In-Reply-To: <4754208D.3133523A@wadsworth.org>
References: <4754208D.3133523A@wadsworth.org>
Message-ID: <fj6s7q$4mi$1@ger.gmane.org>

William Baxter wrote:

> Does anyone have a copy of GifImagePlugin.py that can handle GIF89a?

define "handle".  the standard module always writes GIF87a files, but 
has no problems reading GIF89 files.

</F>


From cfinley at u.washington.edu  Wed Dec  5 22:38:38 2007
From: cfinley at u.washington.edu (Chris Finley)
Date: Wed, 05 Dec 2007 13:38:38 -0800
Subject: [Image-SIG] GUI Toolkit recommendations
Message-ID: <1196890718.5093.15.camel@sdf-workstation>

Greetings,

I'm new to Python and GUI programming (from the Perl/Web world) and
would like some recommendations for GUI toolkits or reference to toolkit
reviews. I'm presently looking at Glade, but that is more by
happenstance than purpose.

The application is a CAD program for testing and manipulating
signed-distance field models. The SDF models will be created from
functions or converted from polygon meshes.

I'm programming more of a research tool than end-product, so the desired
traits are:
1. cross-platform compatibility
2. easy to use and quick development
3. professional looking

Your advise is greatly appreciated,

Chris



From steve at holdenweb.com  Wed Dec  5 22:01:50 2007
From: steve at holdenweb.com (Steve Holden)
Date: Wed, 05 Dec 2007 16:01:50 -0500
Subject: [Image-SIG] How do i import text file which has NaNs ?
In-Reply-To: <4756E1F3.1040101@noaa.gov>
References: <2b125de60711212130s4c99ede9jcc3775cbb7f45bbf@mail.gmail.com>	<004101c835d2$51ae4980$6500a8c0@questar>
	<4756E1F3.1040101@noaa.gov>
Message-ID: <475711BE.8090004@holdenweb.com>

Christopher Barker wrote:
>> The ASCII file has NaNs or NoData in them.
> 
>> I am not able to read the file into python since it is not able to 
>> recognize NaN.
> 
>> */[jv] I?m not sure this is applicable, but have you looked at the numpy 
>> extension package which has support for NaN?/*
> 
> numpy does support NaN, and it's an excellent choice for this kind of 
> data anyway. It's also easy to turn a numpy array into a PIL image.
> 
> However, I'm not sure it supports NaN text literals, and, if it does, 
> it's probably not platform independent, as it uses the system library 
> scanf().
> 
> I would check out numpy, but you may have to write the file scanning 
> code yourself, and translate the NaN.
> 
> An example might help us get other ideas. For instance, what is the 
> ascii representation of NaN? "NaN", or something else?
> 
> What system are you running on? For instance, on OS-X (Python 2.5, 
> compiled with gcc4.0.1), this works:
> 
>  >>> float("NaN")
> nan
> 
> But I don't think that works on the stock Windows Python. However, there 
> may be another literal that does. Try:
> 
>  >>> import numpy as N
>  >>> str(N.nan)
> 'nan'
> 
> and it might tell you the literal on your system.
> 
> -Chris
> 
Sadly not:

 >>> import numpy
 >>> str(numpy.nan)
'-1.#IND'
 >>> float(str(numpy.nan))
Traceback (most recent call last):
   File "<stdin>", line 1, in <module>
ValueError: invalid literal for float(): -1.#IND
 >>> repr(numpy.nan)
'-1.#IND'
 >>>

regards
  Steve
-- 
Steve Holden        +1 571 484 6266   +1 800 494 3119
Holden Web LLC              http://www.holdenweb.com/

From leknarf at pacbell.net  Thu Dec  6 00:21:21 2007
From: leknarf at pacbell.net (Scott Frankel)
Date: Wed, 5 Dec 2007 15:21:21 -0800
Subject: [Image-SIG] GUI Toolkit recommendations
In-Reply-To: <1196890718.5093.15.camel@sdf-workstation>
References: <1196890718.5093.15.camel@sdf-workstation>
Message-ID: <F41B7486-23C2-43EA-BEC1-A6BA2532F606@pacbell.net>


I recently put out a not-too-dissimilar question.  Responses were  
overwhelmingly in favor of WxPython.  You could check the pythonmac- 
sig archives for the thread, "Eclipse + SWT + Python."

Good luck!
Scott




On Dec 5, 2007, at 1:38 PM, Chris Finley wrote:

> Greetings,
>
> I'm new to Python and GUI programming (from the Perl/Web world) and
> would like some recommendations for GUI toolkits or reference to  
> toolkit
> reviews. I'm presently looking at Glade, but that is more by
> happenstance than purpose.
>
> The application is a CAD program for testing and manipulating
> signed-distance field models. The SDF models will be created from
> functions or converted from polygon meshes.
>
> I'm programming more of a research tool than end-product, so the  
> desired
> traits are:
> 1. cross-platform compatibility
> 2. easy to use and quick development
> 3. professional looking
>
> Your advise is greatly appreciated,
>
> Chris
>
>
> _______________________________________________
> 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/20071205/56e57b90/attachment.htm 

From nadavh at visionsense.com  Thu Dec  6 12:01:42 2007
From: nadavh at visionsense.com (Nadav Horesh)
Date: Thu, 06 Dec 2007 13:01:42 +0200
Subject: [Image-SIG] GUI Toolkit recommendations
In-Reply-To: <1196890718.5093.15.camel@sdf-workstation>
References: <1196890718.5093.15.camel@sdf-workstation>
Message-ID: <1196938902.27197.0.camel@nadav.envision.co.il>

wxPython.

  Nadav.

On Wed, 2007-12-05 at 13:38 -0800, Chris Finley wrote:

> Greetings,
> 
> I'm new to Python and GUI programming (from the Perl/Web world) and
> would like some recommendations for GUI toolkits or reference to toolkit
> reviews. I'm presently looking at Glade, but that is more by
> happenstance than purpose.
> 
> The application is a CAD program for testing and manipulating
> signed-distance field models. The SDF models will be created from
> functions or converted from polygon meshes.
> 
> I'm programming more of a research tool than end-product, so the desired
> traits are:
> 1. cross-platform compatibility
> 2. easy to use and quick development
> 3. professional looking
> 
> Your advise is greatly appreciated,
> 
> Chris
> 
> 
> _______________________________________________
> 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/20071206/789c16d7/attachment.htm 

From xiao at blueforge.net  Fri Dec 14 02:47:38 2007
From: xiao at blueforge.net (Xiao)
Date: Fri, 14 Dec 2007 09:47:38 +0800
Subject: [Image-SIG] reading part of an image
Message-ID: <dfede30b0712131747l694975a9h176adfe998243894@mail.gmail.com>

Hi, is it possible to read only a part of an image (png or gif) by PIL?

For example, an image is 20000x20000 in pixels. Is it possible for python to
read only 100x100 of the picture (not loading the whole picture) and save
the small part to a new png file?

Any ideas would be appreciated. Thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/image-sig/attachments/20071214/c0c68b7b/attachment.htm 

From gwidion at mpc.com.br  Sun Dec 16 12:51:21 2007
From: gwidion at mpc.com.br (Joao S. O. Bueno)
Date: Sun, 16 Dec 2007 08:51:21 -0300
Subject: [Image-SIG] reading part of an image
In-Reply-To: <dfede30b0712131747l694975a9h176adfe998243894@mail.gmail.com>
References: <dfede30b0712131747l694975a9h176adfe998243894@mail.gmail.com>
Message-ID: <200712160851.21279.gwidion@mpc.com.br>

On Thursday 13 December 2007 22:47, Xiao wrote:
> Hi, is it possible to read only a part of an image (png or gif) by
> PIL?
>
> For example, an image is 20000x20000 in pixels. Is it possible for
> python to read only 100x100 of the picture (not loading the whole
> picture) and save the small part to a new png file?
>
> Any ideas would be appreciated. Thanks.

Not with PIL straight. It could be done interfacing straight with 
libpng - you can check the PIL source code to check how PNG's are 
loaded and work your own implementation from there.  Maybe it is evne 
possible to make "low level" calls to the classes and methods defined 
in PIL/PngImagePlugin.py (in the source tree) - if not, you certainly 
can just override some classes from that file and work your 
implementation.


	js
	-><_

From fredrik at pythonware.com  Mon Dec 17 21:00:11 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Mon, 17 Dec 2007 21:00:11 +0100
Subject: [Image-SIG] reading part of an image
In-Reply-To: <dfede30b0712131747l694975a9h176adfe998243894@mail.gmail.com>
References: <dfede30b0712131747l694975a9h176adfe998243894@mail.gmail.com>
Message-ID: <fk6kga$nmv$1@ger.gmane.org>

Xiao wrote:

> Hi, is it possible to read only a part of an image (png or gif) by PIL?
> 
> For example, an image is 20000x20000 in pixels. Is it possible for 
> python to read only 100x100 of the picture (not loading the whole 
> picture) and save the small part to a new png file?

PNG and GIF files are encoded streams, so there's no way to "seek" in 
the actual image data.

For other formats, you can check the "tile" attribute after you've 
opened the image, but before you've loaded it; if it contains a list of 
descriptors, or if the type for the descriptors is "raw", you can 
manipulate the "tile" list so it contains only the part you want to 
read, and manipulate the "size" accordingly.

For an example, see:

    http://effbot.org/zone/pil-large-images.htm

</F>


From epui at sjri.com  Tue Dec 18 13:10:24 2007
From: epui at sjri.com (Gross)
Date: Tue, 18 Dec 2007 12:10:24 +0000
Subject: [Image-SIG] I've had many people tell me they liked it more than
	JavaOne.
Message-ID: <4767B8B0.6060507@carolinaturkeys.com>

An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/image-sig/attachments/20071218/a8eeb81f/attachment.htm 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: steering.gif
Type: image/gif
Size: 14657 bytes
Desc: not available
Url : http://mail.python.org/pipermail/image-sig/attachments/20071218/a8eeb81f/attachment.gif 

From rowen at cesmail.net  Tue Dec 18 23:09:50 2007
From: rowen at cesmail.net (Russell E. Owen)
Date: Tue, 18 Dec 2007 14:09:50 -0800
Subject: [Image-SIG] Get PIL version number?
Message-ID: <rowen-8E46B0.14095018122007@news.gmane.org>

How can I query the version number of PIL from Python? Most packages 
have a __version__ attribute but neither the Image nor PIL package seem 
to have this.

-- Russell


From Chris.Barker at noaa.gov  Wed Dec 19 00:10:49 2007
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Tue, 18 Dec 2007 15:10:49 -0800
Subject: [Image-SIG] Get PIL version number?
In-Reply-To: <rowen-8E46B0.14095018122007@news.gmane.org>
References: <rowen-8E46B0.14095018122007@news.gmane.org>
Message-ID: <47685379.4050909@noaa.gov>

Image.VERSION

It's too bad it doesn't follow the now-common convention of __version__. 
This is how I found it:

 >>> for s in dir(Image):
...   if "version" in s.lower(): print s
...
VERSION
 >>> Image.VERSION
'1.1.6'



-- 
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 danielsforummail at terra.com.br  Wed Dec 19 05:05:52 2007
From: danielsforummail at terra.com.br (Daniel Felix Ferber)
Date: Wed, 19 Dec 2007 02:05:52 -0200
Subject: [Image-SIG] Remove noise from high ISO photos
Message-ID: <476898A0.5070708@terra.com.br>

Hi!

I was wondering if PIL has a filter or enhancement in order to remove 
noise from photos that were taked with high ISO. Gimp and ImageMagic 
both have such a feature and it work very good.
I tried to use SMOOTH or SMOOTH_MORE filter. But there are better 
strategies to remove high ISO noise that try to preserve sharpness of 
the image.
Currently I am calling ImageMagic from within my python script, but I 
would prefer if PIL would handle it. Is this possible in current version 
of PIL?

Best regards,
Daniel Ferber

From carlos.s.santos at gmail.com  Wed Dec 19 13:18:13 2007
From: carlos.s.santos at gmail.com (Carlos da Silva Santos)
Date: Wed, 19 Dec 2007 10:18:13 -0200
Subject: [Image-SIG] Remove noise from high ISO photos
In-Reply-To: <476898A0.5070708@terra.com.br>
References: <476898A0.5070708@terra.com.br>
Message-ID: <1dc6ddb60712190418r259e7477l5468008113097b53@mail.gmail.com>

Daniel,

I don't know much about high ISO noise, but a standard way of removing
noise while preserving edges is using a median filter:
http://www.pythonware.com/library/pil/handbook/imagefilter.htm

If your goal is only running the whole pipeline from python, you might
install the python bindings for Gimp and try to use the Gimp filter
from python.

I hope this helps.

[]s

Carlos



On Dec 19, 2007 2:05 AM, Daniel Felix Ferber
<danielsforummail at terra.com.br> wrote:
> Hi!
>
> I was wondering if PIL has a filter or enhancement in order to remove
> noise from photos that were taked with high ISO. Gimp and ImageMagic
> both have such a feature and it work very good.
> I tried to use SMOOTH or SMOOTH_MORE filter. But there are better
> strategies to remove high ISO noise that try to preserve sharpness of
> the image.
> Currently I am calling ImageMagic from within my python script, but I
> would prefer if PIL would handle it. Is this possible in current version
> of PIL?
>
> Best regards,
> Daniel Ferber
> _______________________________________________
> Image-SIG maillist  -  Image-SIG at python.org
> http://mail.python.org/mailman/listinfo/image-sig
>

From danielsforummail at terra.com.br  Wed Dec 19 13:57:33 2007
From: danielsforummail at terra.com.br (Daniel Felix Ferber)
Date: Wed, 19 Dec 2007 10:57:33 -0200
Subject: [Image-SIG] Remove noise from high ISO photos
In-Reply-To: <1dc6ddb60712190418r259e7477l5468008113097b53@mail.gmail.com>
References: <476898A0.5070708@terra.com.br>
	<1dc6ddb60712190418r259e7477l5468008113097b53@mail.gmail.com>
Message-ID: <4769153D.903@terra.com.br>

Hi Carlos,

Thanks for the advice about median filter. I will do some experiments.

When taking a photo with high ISO on a digital camera, there will be 
many pixels that have a random color that is completely different color 
from the expected one. But the photo still has a good sharpness 
ImageMagic uses a stategy that is similar to the median filter: it looks 
at the neighborhood o a pixel. If the color of the pixel is considerably 
different from the color of adjacent pixels, then the color of the pixel 
is changed to a color close to the neighborhood. If not, the pixel is 
preserved.

It seems that the median filter always changes the color of the pixel, 
and I am afraid that the image might loose sharpness.

I already considered using bindings for gimp, but only found information 
about writing gimp plugins in python. Does the binding allow me to 
control gimp from a python script that I run on a bash console?

Best regards,
Daniel Felix Ferber

Carlos da Silva Santos escreveu:
> Daniel,
>
> I don't know much about high ISO noise, but a standard way of removing
> noise while preserving edges is using a median filter:
> http://www.pythonware.com/library/pil/handbook/imagefilter.htm
>
> If your goal is only running the whole pipeline from python, you might
> install the python bindings for Gimp and try to use the Gimp filter
> from python.
>
> I hope this helps.
>
> []s
>
> Carlos
>
>
>
> On Dec 19, 2007 2:05 AM, Daniel Felix Ferber
> <danielsforummail at terra.com.br> wrote:
>   
>> Hi!
>>
>> I was wondering if PIL has a filter or enhancement in order to remove
>> noise from photos that were taked with high ISO. Gimp and ImageMagic
>> both have such a feature and it work very good.
>> I tried to use SMOOTH or SMOOTH_MORE filter. But there are better
>> strategies to remove high ISO noise that try to preserve sharpness of
>> the image.
>> Currently I am calling ImageMagic from within my python script, but I
>> would prefer if PIL would handle it. Is this possible in current version
>> of PIL?
>>
>> Best regards,
>> Daniel Ferber
>> _______________________________________________
>> Image-SIG maillist  -  Image-SIG at python.org
>> http://mail.python.org/mailman/listinfo/image-sig
>>
>>     
>
> Esta mensagem foi verificada pelo E-mail Protegido Terra.
> Scan engine: McAfee VirusScan / Atualizado em 18/12/2007 / Vers?o: 5.1.00/5188
> Proteja o seu e-mail Terra: http://mail.terra.com.br/
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/image-sig/attachments/20071219/c8cd59af/attachment.htm 

From douglas at paradise.net.nz  Wed Dec 19 23:38:03 2007
From: douglas at paradise.net.nz (Douglas Bagnall)
Date: Thu, 20 Dec 2007 11:38:03 +1300
Subject: [Image-SIG] Remove noise from high ISO photos
In-Reply-To: <4769153D.903@terra.com.br>
References: <476898A0.5070708@terra.com.br>
	<1dc6ddb60712190418r259e7477l5468008113097b53@mail.gmail.com>
	<4769153D.903@terra.com.br>
Message-ID: <47699D4B.9030902@paradise.net.nz>

Daniel Felix Ferber wrote:

> When taking a photo with high ISO on a digital camera, there will be
> many pixels that have a random color that is completely different color
> from the expected one. But the photo still has a good sharpness
> ImageMagic uses a stategy that is similar to the median filter: it looks
> at the neighborhood o a pixel. If the color of the pixel is considerably
> different from the color of adjacent pixels, then the color of the pixel
> is changed to a color close to the neighborhood. If not, the pixel is
> preserved.
> 
> It seems that the median filter always changes the color of the pixel,
> and I am afraid that the image might loose sharpness.


What you are describing could be achieved thus:

1. Take a 3x3 median image. Most pixels will hardly be changed, but the
noisy ones will be.

2. find the difference between the original and the median (using
ImageChops.difference). Good pixels will be black or nearly so, but the
noisy ones won't be.

3. Use the difference image as a mask for pasting the median over the
original.  You will need to convert the difference image to grey-scale
in some way, or split it into bands and do each band of the image
separately. image.convert('L') is probably not the best way. You might
want to make it into a threshold image (rather than have it blending
proportionately to the difference).

But I'm not sure how much better than a simple median it will be--after
all, the pixels that this process doesn't change are exactly the pixels
that the median hardly changes.

In any case, the median will be much better than SMOOTH and SMOOTH_MORE.


Douglas


From fredrik at pythonware.com  Wed Dec 19 23:56:23 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Wed, 19 Dec 2007 23:56:23 +0100
Subject: [Image-SIG] Get PIL version number?
In-Reply-To: <47685379.4050909@noaa.gov>
References: <rowen-8E46B0.14095018122007@news.gmane.org>
	<47685379.4050909@noaa.gov>
Message-ID: <fkc7il$tdd$1@ger.gmane.org>

Christopher Barker wrote:

> Image.VERSION
> 
> It's too bad it doesn't follow the now-common convention of __version__. 

PIL is older than most Python conventions, and PIL's programming API 
tends to be *extremely* stable (on purpose).

(Note that the style guide says that "__version__" should contain the 
VCS revision number, which isn't that helpful.  It also says that the 
__name__ form is for magic stuff.  And for completeness, Python's own 
version number is of course found in "sys.version".  Consistency...)

</F>


From Chris.Barker at noaa.gov  Thu Dec 20 01:24:22 2007
From: Chris.Barker at noaa.gov (Christopher Barker)
Date: Wed, 19 Dec 2007 16:24:22 -0800
Subject: [Image-SIG] Get PIL version number?
In-Reply-To: <fkc7il$tdd$1@ger.gmane.org>
References: <rowen-8E46B0.14095018122007@news.gmane.org>
	<47685379.4050909@noaa.gov> <fkc7il$tdd$1@ger.gmane.org>
Message-ID: <4769B636.4040204@noaa.gov>

Fredrik Lundh wrote:
> Christopher Barker wrote:
>> It's too bad it doesn't follow the now-common convention of __version__. 
> 
> PIL is older than most Python conventions, 

yes, it sure is.

> and PIL's programming API  tends to be *extremely* stable (on purpose).

Which is much appreciated.

> (Note that the style guide says that "__version__" should contain the 
> VCS revision number, which isn't that helpful.  It also says that the 
> __name__ form is for magic stuff.  And for completeness, Python's own 
> version number is of course found in "sys.version".  Consistency...)

All quite true, but like it or not, __version__ has become pretty 
common. Would it be so bad to have:

__version__ = VERSION

in there somewhere?

One less question to the mailing list is a good thing....

-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 danielsforummail at terra.com.br  Thu Dec 20 13:34:40 2007
From: danielsforummail at terra.com.br (danielsforummail)
Date: Thu, 20 Dec 2007 10:34:40 -0200
Subject: [Image-SIG] Remove noise from high ISO photos
Message-ID: <JTCLLS$22A37736B4C14062B9156B01237E0DFF@multidominios>

Hi Douglas, thanks for the advices. I will give them a try.
Best regards,
Daniel Ferber

---------- Cabe?alho original -----------

De: "Douglas Bagnall" douglas at paradise.net.nz
Para: "Daniel Felix Ferber" danielsforummail at terra.com.br, image-sig at python.org
C?pia: 
Data: Thu, 20 Dec 2007 11:38:03 +1300
Assunto: Re: [Image-SIG] Remove noise from high ISO photos

> Daniel Felix Ferber wrote:
> 
> > When taking a photo with high ISO on a digital camera, there will be
> > many pixels that have a random color that is completely different color
> > from the expected one. But the photo still has a good sharpness
> > ImageMagic uses a stategy that is similar to the median filter: it looks
> > at the neighborhood o a pixel. If the color of the pixel is considerably
> > different from the color of adjacent pixels, then the color of the pixel
> > is changed to a color close to the neighborhood. If not, the pixel is
> > preserved.
> > 
> > It seems that the median filter always changes the color of the pixel,
> > and I am afraid that the image might loose sharpness.
> 
> 
> What you are describing could be achieved thus:
> 
> 1. Take a 3x3 median image. Most pixels will hardly be changed, but the
> noisy ones will be.
> 
> 2. find the difference between the original and the median (using
> ImageChops.difference). Good pixels will be black or nearly so, but the
> noisy ones won't be.
> 
> 3. Use the difference image as a mask for pasting the median over the
> original.  You will need to convert the difference image to grey-scale
> in some way, or split it into bands and do each band of the image
> separately. image.convert('L') is probably not the best way. You might
> want to make it into a threshold image (rather than have it blending
> proportionately to the difference).
> 
> But I'm not sure how much better than a simple median it will be--after
> all, the pixels that this process doesn't change are exactly the pixels
> that the median hardly changes.
> 
> In any case, the median will be much better than SMOOTH and SMOOTH_MORE.
> 
> 
> Douglas
> 
> 
> Esta mensagem foi verificada pelo E-mail Protegido Terra.
> Scan engine: McAfee VirusScan / Atualizado em 19/12/2007 / Vers?o: 5.1.00/5189
> Proteja o seu e-mail Terra: http://mail.terra.com.br/
> 


From donn.ingle at gmail.com  Thu Dec 20 15:09:20 2007
From: donn.ingle at gmail.com (Donn)
Date: Thu, 20 Dec 2007 16:09:20 +0200
Subject: [Image-SIG] PIL ImageFont segfault mystery
Message-ID: <200712201609.20665.donn.ingle@gmail.com>

Hello,
 Here's the story:

* python-imaging Version: 1.1.6-1
* Python 2.4 and 2.5
* Kubuntu Gnu/Linux 7.10

This is the code snippet that causes a segfault:

font = ImageFont.truetype("path/to/Anatevka Caps.ttf", points, index=0) 
print "getname:", font.getname() # <- App dies here.

To reproduce:
It does not happen often, but there are some oddball fonts from my deep murky 
past that I have kept around for testing and they kill it quick. I don't know 
if I can attach items to this list, so will await interest.

I cannot really narrow this down much more. It looks like the family name is 
missing (see info at end), which might be the bug. 

When I try [font = ImageFont.truetype(paf, points,index=0, encoding="armn") ] 
or any of the other encodings (which I can't really find a good list of) the 
effect is still the same.

Naturally, having an segfault means there ain't no way to catch it within 
Python! 

Regards,
\d

Font info:
crashme2:$ ftinfo -la Anatevka\ Caps.ttf
Anatevka Caps.ttf: 0 face
Anatevka Caps.ttf: 0: 0 charmap: Apple, Unicode defaut semantics
Anatevka Caps.ttf: 0: 1 charmap: Macintosh, Roman
Anatevka Caps.ttf: 0: 2 charmap: Microsoft, Unicode
Anatevka Caps.ttf: 0: header: HASH(0x8350614)
Anatevka Caps.ttf: 0: horizontal: HASH(0x8350740)
Anatevka Caps.ttf: 0: max_Contours: 11
Anatevka Caps.ttf: 0: max_Points: 117
Anatevka Caps.ttf: 0: num_Faces: 1
Anatevka Caps.ttf: 0: num_Glyphs: 251
Anatevka Caps.ttf: 0: os2: HASH(0x8350824)
Anatevka Caps.ttf: 0: postscript: HASH(0x83509bc)
Anatevka Caps.ttf: 0: header: CheckSum_Adjust: -1410471583
Anatevka Caps.ttf: 0: header: Created: ARRAY(0x8350668)
Anatevka Caps.ttf: 0: header: Flags: 0
Anatevka Caps.ttf: 0: header: Font_Direction: 2
Anatevka Caps.ttf: 0: header: Font_Revision: 65536
Anatevka Caps.ttf: 0: header: Glyph_Data_Format: 0
Anatevka Caps.ttf: 0: header: Index_To_Loc_Format: 1
Anatevka Caps.ttf: 0: header: Lowest_Rec_PPEM: 3
Anatevka Caps.ttf: 0: header: Mac_Style: 0
Anatevka Caps.ttf: 0: header: Magic_Number: 1594834165
Anatevka Caps.ttf: 0: header: Modified: ARRAY(0x8350698)
Anatevka Caps.ttf: 0: header: Table_Version: 65536
Anatevka Caps.ttf: 0: header: Units_Per_EM: 4096
Anatevka Caps.ttf: 0: header: xMax: 3424
Anatevka Caps.ttf: 0: header: xMin: -574
Anatevka Caps.ttf: 0: header: yMax: 2693
Anatevka Caps.ttf: 0: header: yMin: -573
Anatevka Caps.ttf: 0: horizontal: Ascender: 2693
Anatevka Caps.ttf: 0: horizontal: Descender: -573
Anatevka Caps.ttf: 0: horizontal: Line_Gap: 300
Anatevka Caps.ttf: 0: horizontal: Reserved0: 0
Anatevka Caps.ttf: 0: horizontal: Reserved1: 0
Anatevka Caps.ttf: 0: horizontal: Reserved2: 0
Anatevka Caps.ttf: 0: horizontal: Reserved3: 0
Anatevka Caps.ttf: 0: horizontal: Reserved4: 0
Anatevka Caps.ttf: 0: horizontal: Version: 65536
Anatevka Caps.ttf: 0: horizontal: advance_Width_Max: 3809
Anatevka Caps.ttf: 0: horizontal: caret_Slope_Rise: 1
Anatevka Caps.ttf: 0: horizontal: caret_Slope_Run: 0
Anatevka Caps.ttf: 0: horizontal: metric_Data_Format: 0
Anatevka Caps.ttf: 0: horizontal: min_Left_Side_Bearing: -574
Anatevka Caps.ttf: 0: horizontal: min_Right_Side_Bearing: -971
Anatevka Caps.ttf: 0: horizontal: number_Of_HMetrics: 248
Anatevka Caps.ttf: 0: horizontal: xMax_Extent: 3424
Anatevka Caps.ttf: 0: 0 name: Apple, Unicode defaut semantics, unknown 0, 
Copyright 
notice: ............................................................
Anatevka Caps.ttf: 0: 1 name: Apple, Unicode defaut semantics, unknown 0, Font 
Family name: ............
Anatevka Caps.ttf: 0: 2 name: Apple, Unicode defaut semantics, unknown 0, Font 
Subfamily name:
Anatevka Caps.ttf: 0: 3 name: Apple, Unicode defaut semantics, unknown 0, 
Unique font identifier: ................
Anatevka Caps.ttf: 0: 4 name: Apple, Unicode defaut semantics, unknown 0, Full 
font name: ............
Anatevka Caps.ttf: 0: 5 name: Apple, Unicode defaut semantics, unknown 0, 
Version string: ............
Anatevka Caps.ttf: 0: 6 name: Apple, Unicode defaut semantics, unknown 0, 
Postscript name for the font: ............
Anatevka Caps.ttf: 0: 7 name: Apple, Unicode defaut semantics, unknown 0, 
Trademark: ..........
Anatevka Caps.ttf: 0: 8 name: Macintosh, Roman, English, Copyright notice: 
Copyright 1999 DATA BECKER GmbH & Co. KG. All Rights Reserved
Anatevka Caps.ttf: 0: 9 name: Macintosh, Roman, English, Font Family name: 
Anatevka Caps
Anatevka Caps.ttf: 0: 10 name: Macintosh, Roman, English, Font Subfamily name:
Anatevka Caps.ttf: 0: 11 name: Macintosh, Roman, English, Unique font 
identifier: PAW-Anatevka Caps
Anatevka Caps.ttf: 0: 12 name: Macintosh, Roman, English, Full font name: 
Anatevka Caps
Anatevka Caps.ttf: 0: 13 name: Macintosh, Roman, English, Version string: 
Version 1.50
Anatevka Caps.ttf: 0: 14 name: Macintosh, Roman, English, Postscript name for 
the font: Anatevka Caps
Anatevka Caps.ttf: 0: 15 name: Macintosh, Roman, English, Trademark: DATA 
BECKER
Anatevka Caps.ttf: 0: 16 name: Microsoft, Unicode, English - United States, 
Copyright notice: Copyright 1999 DATA BECKER GmbH & Co. KG. All Rights 
Reserved
Anatevka Caps.ttf: 0: 17 name: Microsoft, Unicode, English - United States, 
Font Family name: Anatevka Caps
Anatevka Caps.ttf: 0: 18 name: Microsoft, Unicode, English - United States, 
Font Subfamily name:
Anatevka Caps.ttf: 0: 19 name: Microsoft, Unicode, English - United States, 
Unique font identifier: PAW-Anatevka Caps
Anatevka Caps.ttf: 0: 20 name: Microsoft, Unicode, English - United States, 
Full font name: Anatevka Caps
Anatevka Caps.ttf: 0: 21 name: Microsoft, Unicode, English - United States, 
Version string: Version 1.50
Anatevka Caps.ttf: 0: 22 name: Microsoft, Unicode, English - United States, 
Postscript name for the font: Anatevka Caps
Anatevka Caps.ttf: 0: 23 name: Microsoft, Unicode, English - United States, 
Trademark: DATA BECKER
Anatevka Caps.ttf: 0: os2: code page range:
Anatevka Caps.ttf: 0: os2: unicode range:
Anatevka Caps.ttf: 0: os2: xAvgCharWidth = 1369
Anatevka Caps.ttf: 0: os2: usWeightClass: Normal (Regular)
Anatevka Caps.ttf: 0: os2: usWidthClass : Medium (normal)
Anatevka Caps.ttf: 0: os2: fsSelection : REGULAR
Anatevka Caps.ttf: 0: os2: achVendID: Alts
Anatevka Caps.ttf: 0: os2: fsSelection: 64
Anatevka Caps.ttf: 0: os2: fsType: 1
Anatevka Caps.ttf: 0: os2: panose: 0x00 0x00 0x04 0x00 0x00 0x00 0x00 0x00 
0x00 0x00
Anatevka Caps.ttf: 0: os2: sFamilyClass: 0
Anatevka Caps.ttf: 0: os2: sTypoAscender: 2307
Anatevka Caps.ttf: 0: os2: sTypoDescender: -1789
Anatevka Caps.ttf: 0: os2: sTypoLineGap: 300
Anatevka Caps.ttf: 0: os2: ulCodePageRange1: 0
Anatevka Caps.ttf: 0: os2: ulCodePageRange2: 0
Anatevka Caps.ttf: 0: os2: ulUnicodeRange1: 0
Anatevka Caps.ttf: 0: os2: ulUnicodeRange2: 0
Anatevka Caps.ttf: 0: os2: ulUnicodeRange3: 0
Anatevka Caps.ttf: 0: os2: ulUnicodeRange4: 0
Anatevka Caps.ttf: 0: os2: usFirstCharIndex: 32
Anatevka Caps.ttf: 0: os2: usLastCharIndex: 64258
Anatevka Caps.ttf: 0: os2: usWeightClass: 400
Anatevka Caps.ttf: 0: os2: usWidthClass: 5
Anatevka Caps.ttf: 0: os2: usWinAscent: 2693
Anatevka Caps.ttf: 0: os2: usWinDescent: 573
Anatevka Caps.ttf: 0: os2: version: 0
Anatevka Caps.ttf: 0: os2: xAvgCharWidth: 1369
Anatevka Caps.ttf: 0: os2: yStrikeoutPosition: 1120
Anatevka Caps.ttf: 0: os2: yStrikeoutSize: 204
Anatevka Caps.ttf: 0: os2: ySubscriptXOffset: 0
Anatevka Caps.ttf: 0: os2: ySubscriptXSize: 512
Anatevka Caps.ttf: 0: os2: ySubscriptYOffset: 586
Anatevka Caps.ttf: 0: os2: ySubscriptYSize: 512
Anatevka Caps.ttf: 0: os2: ySuperscriptXOffset: 0
Anatevka Caps.ttf: 0: os2: ySuperscriptXSize: 512
Anatevka Caps.ttf: 0: os2: ySuperscriptYOffset: 2300
Anatevka Caps.ttf: 0: os2: ySuperscriptYSize: 512
Anatevka Caps.ttf: 0: postscript: FormatType: 131072
Anatevka Caps.ttf: 0: postscript: isFixedPitch: 0
Anatevka Caps.ttf: 0: postscript: italicAngle: 0
Anatevka Caps.ttf: 0: postscript: maxMemType1: 0
Anatevka Caps.ttf: 0: postscript: maxMemType42: 0
Anatevka Caps.ttf: 0: postscript: minMemType1: 0
Anatevka Caps.ttf: 0: postscript: minMemType42: 0
Anatevka Caps.ttf: 0: postscript: underlinePosition: -400
Anatevka Caps.ttf: 0: postscript: underlineThickness: 160


From donn.ingle at gmail.com  Sat Dec 22 14:30:14 2007
From: donn.ingle at gmail.com (Donn)
Date: Sat, 22 Dec 2007 15:30:14 +0200
Subject: [Image-SIG] PIL ImageFont segfault mystery
In-Reply-To: <200712201609.20665.donn.ingle@gmail.com>
References: <200712201609.20665.donn.ingle@gmail.com>
Message-ID: <200712221530.14941.donn.ingle@gmail.com>

No takers? I'll have a look for the bug report link online.

\d

From fredrik at pythonware.com  Sat Dec 22 15:23:23 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sat, 22 Dec 2007 15:23:23 +0100
Subject: [Image-SIG] PIL ImageFont segfault mystery
In-Reply-To: <200712221530.14941.donn.ingle@gmail.com>
References: <200712201609.20665.donn.ingle@gmail.com>
	<200712221530.14941.donn.ingle@gmail.com>
Message-ID: <fkj6kp$voa$1@ger.gmane.org>

Donn wrote:

> No takers? I'll have a look for the bug report link online.

seeing the bug report isn't necessarily the same thing as having time to 
work on a fix ;-)

my guess is that the file uses some less common file format feature, 
which causes either freetype or the _imagingft binding to choke, but I 
need a copy of the font file to be sure (either mail it to me, or post 
it somewhere and mail me a link).

cheers /F


From fredrik at pythonware.com  Sun Dec 23 20:38:08 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Sun, 23 Dec 2007 20:38:08 +0100
Subject: [Image-SIG] PIL ImageFont segfault mystery
In-Reply-To: <fkj6kp$voa$1@ger.gmane.org>
References: <200712201609.20665.donn.ingle@gmail.com>	<200712221530.14941.donn.ingle@gmail.com>
	<fkj6kp$voa$1@ger.gmane.org>
Message-ID: <fkmdf0$p54$1@ger.gmane.org>

Fredrik Lundh wrote:

> seeing the bug report isn't necessarily the same thing as having time to 
> work on a fix ;-)

as I suspected, the offending files was missing some optional fields; 
freetype2 handled that nicely, but _imagingft module didn't expect the 
fields to be NULL.

here's a tentative patch:

$ svn diff _imagingft.c
Index: _imagingft.c
===================================================================
--- _imagingft.c        (revision 3345)
+++ _imagingft.c        (working copy)
@@ -422,10 +422,11 @@

      /* attributes */
      if (!strcmp(name, "family"))
-        return PyString_FromString(self->face->family_name);
+        return PyString_FromString(self->face->family_name ?
+                                   self->face->family_name : "");
      if (!strcmp(name, "style"))
-        return PyString_FromString(self->face->style_name);
-
+        return PyString_FromString(self->face->style_name ?
+                                   self->face->style_name : "");
      if (!strcmp(name, "ascent"))
          return PyInt_FromLong(PIXEL(self->face->size->metrics.ascender));
      if (!strcmp(name, "descent"))

(might be a better idea to return None in this case, but the above will 
at least prevent the crash).

</F>


From donn.ingle at gmail.com  Mon Dec 24 09:24:31 2007
From: donn.ingle at gmail.com (Donn)
Date: Mon, 24 Dec 2007 10:24:31 +0200
Subject: [Image-SIG] PIL ImageFont segfault mystery
In-Reply-To: <fkmdf0$p54$1@ger.gmane.org>
References: <200712201609.20665.donn.ingle@gmail.com>
	<fkj6kp$voa$1@ger.gmane.org> <fkmdf0$p54$1@ger.gmane.org>
Message-ID: <200712241024.31700.donn.ingle@gmail.com>

Fredrik,
 I dropped off the Internet (I have a fragile ISP :( ) and did not get your 
mail until a moment ago. Thanks for looking at this. 

 I am not at all informed about how patches get used and, more importantly, 
how they make their way into the final repos used by distros. Is there any 
way to get _imagingft.c fixed so that old users worldwide get an update, or 
will this fix only come out in new versions of PIL from this point on?

 I ask because I am facing a choice of 1) just waiting and using my short time 
to code my app then relying on the fix being "out there", or 2) Trying to 
grok and compile PIL and then house it in a sub-directory of my app and 
distribute it as a "module" within my app.

Thanks for the help!
\d

PS: Not to muddy the waters, but PIL is the only effective FT2 interface for 
Python and I was wondering if it would be possible to expose more meta 
information about fonts than family and style? There are other snippets of 
info in there like number of faces, copyright, foundry and so forth. Would it 
be possible to add these too?

From donn.ingle at gmail.com  Tue Dec 25 18:56:10 2007
From: donn.ingle at gmail.com (Donn)
Date: Tue, 25 Dec 2007 19:56:10 +0200
Subject: [Image-SIG] Python MemoryError on a ttf font
In-Reply-To: <200712251936.32314.donn.ingle@gmail.com>
References: <200712251936.32314.donn.ingle@gmail.com>
Message-ID: <200712251956.10318.donn.ingle@gmail.com>

More info,
 I noticed that if you render only one character, like "A", it works. 
It crashes on "TE", but not "T". I tried various combinations, but the picture 
is not clear. It sure doesn't like the capital "T" :)

\d

> Hi again,
>  I thought I'd let you know about another font that's causing an error. The
> font is attached. Here's the code that I used:



From donn.ingle at gmail.com  Wed Dec 26 15:40:23 2007
From: donn.ingle at gmail.com (Donn)
Date: Wed, 26 Dec 2007 16:40:23 +0200
Subject: [Image-SIG] ImageFont Unicode Decode error
Message-ID: <200712261640.23347.donn.ingle@gmail.com>

Hello,
 Trying to open a ttf file with a non-standard character in it (u'\xe5') 
causes an error, no matter what I pass to encoding:

enc="unic"
enc="unicode"
enc="utf8"
etc.
font = ImageFont.truetype( paf, 64, index=i , encoding=enc )

The error:
  File "/usr/lib/python2.5/site-packages/PIL/ImageFont.py", line 205, in 
truetype
    return FreeTypeFont(filename, size, index, encoding)
  File "/usr/lib/python2.5/site-packages/PIL/ImageFont.py", line 121, in 
__init__
    self.font = _imagingft.getfont(file, size, index, encoding)
UnicodeEncodeError: 'ascii' codec can't encode character u'\xe5' in position 
66: ordinal not in range(128)


Is there something I'm doing wrong, or is this a bug?

\d

From donn.ingle at gmail.com  Tue Dec 25 18:36:32 2007
From: donn.ingle at gmail.com (Donn)
Date: Tue, 25 Dec 2007 19:36:32 +0200
Subject: [Image-SIG] Python MemoryError on a ttf font
Message-ID: <200712251936.32314.donn.ingle@gmail.com>

Hi again,
 I thought I'd let you know about another font that's causing an error. The 
font is attached. Here's the code that I used:

paf = "/home/donn/06.FontStore/TTFS/O/Onsoku Seinen Plane.ttf"  
text = "Test Text"
font = ImageFont.truetype( paf, 64, index=0, encoding="unicode" )
pilheight = int( font.getsize( text )[1] )
pilimage = Image.new("RGB", (500,pilheight), (255,255,255)) 
drawnFont = ImageDraw.Draw( pilimage )
# This line causes a MemoryError:
drawnFont.text((0,0) , text, font=font, fill=0 )

Is there any way to tell ahead of time (i.e. without having to render it 
with.text) that it's a candidate for error?

Regards,
Donn.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: Onsoku Seinen Plane.ttf
Type: application/x-font-ttf
Size: 67244 bytes
Desc: not available
Url : http://mail.python.org/pipermail/image-sig/attachments/20071225/055c7253/attachment-0001.bin 

From junk at ysengrin.com  Thu Dec 27 01:47:10 2007
From: junk at ysengrin.com (junk at ysengrin.com)
Date: Wed, 26 Dec 2007 16:47:10 -0800
Subject: [Image-SIG] [PIL 1.1.6] bug with python 2.51 but not with python
	2.44
Message-ID: <4772F60E.3000000@ysengrin.com>

Hi,

the same code runs fine in python 2.44 but fails with python 2.51

File "/usr/local/lib/python2.5/site-packages/PIL/Image.py", line 783, in filter
 filter = filter()
AttributeError: Kernel instance has no __call__ method

contact me if you need more info

yomgui




From fredrik at pythonware.com  Thu Dec 27 11:19:07 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Thu, 27 Dec 2007 11:19:07 +0100
Subject: [Image-SIG] [PIL 1.1.6] bug with python 2.51 but not with
	python 2.44
In-Reply-To: <4772F60E.3000000@ysengrin.com>
References: <4772F60E.3000000@ysengrin.com>
Message-ID: <fkvu6r$i2l$1@ger.gmane.org>

junk at ysengrin.com wrote:

> the same code runs fine in python 2.44 but fails with python 2.51
> 
> File "/usr/local/lib/python2.5/site-packages/PIL/Image.py", line 783, in filter
>  filter = filter()
> AttributeError: Kernel instance has no __call__ method
> 
> contact me if you need more info

are you perhaps importing things both from "PIL" and from the path 
itself?  (e.g. mixing "from PIL import Image" with "import Image"?). 
does PIL's installation directory appear in sys.path?

if you cannot make this work, try changing

         if not isinstance(filter, Filter):
             filter = filter()

to

	if type(filter) is type(Filter):
	    filter = filter()

in Image.py.

</F>


From fredrik at pythonware.com  Thu Dec 27 12:52:37 2007
From: fredrik at pythonware.com (Fredrik Lundh)
Date: Thu, 27 Dec 2007 12:52:37 +0100
Subject: [Image-SIG] [PIL 1.1.6] bug with python 2.51 but not with
	python 2.44
In-Reply-To: <fkvu6r$i2l$1@ger.gmane.org>
References: <4772F60E.3000000@ysengrin.com> <fkvu6r$i2l$1@ger.gmane.org>
Message-ID: <fl03m5$19r$1@ger.gmane.org>

Fredrik Lundh wrote:

> if you cannot make this work, try changing
> 
>          if not isinstance(filter, Filter):
>              filter = filter()
> 
> to
> 
> 	if type(filter) is type(Filter):
> 	    filter = filter()
> 
> in Image.py.

or, better:

  	if callable(filter):
  	    filter = filter()


From donn.ingle at gmail.com  Thu Dec 27 16:46:35 2007
From: donn.ingle at gmail.com (Donn)
Date: Thu, 27 Dec 2007 17:46:35 +0200
Subject: [Image-SIG] ImageFont Unicode Decode error
In-Reply-To: <194b7da00712260710v6cab83afv82f67948a18c97b3@mail.gmail.com>
References: <200712261640.23347.donn.ingle@gmail.com>
	<194b7da00712260710v6cab83afv82f67948a18c97b3@mail.gmail.com>
Message-ID: <200712271746.36038.donn.ingle@gmail.com>

> I don't know, but may be PIL needs to have the Byte Order Mark for handling
> unicode.
Not 100% sure what you are saying.

That character is ( I don't know if it will travel ) "?" and I can open files 
that contain it with my other python stuff like codes.open etc.

Are you suggesting a solution to me or thinking aloud :)

I'm on Kubuntu Gnu/Linux 7.10 with Python 2.5


\d

> Hello,
>  Trying to open a ttf file with a non-standard character in it (u'\xe5')
> causes an error, no matter what I pass to encoding:
>
> enc="unic"
> enc="unicode"
> enc="utf8"
> etc.
> font = ImageFont.truetype( paf, 64, index=i , encoding=enc )
>
> The error:
>  File "/usr/lib/python2.5/site-packages/PIL/ImageFont.py", line 205, in
> truetype
>    return FreeTypeFont(filename, size, index, encoding)
>  File "/usr/lib/python2.5/site-packages/PIL/ImageFont.py", line 121, in
> __init__
>    self.font = _imagingft.getfont(file, size, index, encoding)
> UnicodeEncodeError: 'ascii' codec can't encode character u'\xe5' in
> position
> 66: ordinal not in range(128)
>
>
> Is there something I'm doing wrong, or is this a bug?
>
> \d