[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: problem with imwrite and underlying __magic_write__
From: |
Ian Journeaux |
Subject: |
RE: problem with imwrite and underlying __magic_write__ |
Date: |
Tue, 3 Aug 2010 18:49:57 -0500 |
David
Found the source code. I see in the code that there is an attempt to
correctly encode the 8bit type images. Unfortunately under Windows it
doesn't seem to work.
I don't mind doing some troubleshooting legwork but I need to figure out how
to compile up the required code so I can test out the code and determine
what the problem might be.
I will try and build an environment that will allow me to compile Octave.
Any tips?
Take Care
-----Original Message-----
From: David Grundberg [mailto:address@hidden
Sent: Tuesday, August 03, 2010 12:33 PM
To: Ian Journeaux
Cc: address@hidden
Subject: Re: problem with imwrite and underlying __magic_write__
Ian Journeaux skrev:
>
>
> I am trying to troubleshoot a problem with the imwrite.m function. Under
> Windows, if I read in an 8-bit image with imread.m and write it out with
> imwrite.m, the image saved by imwrite is not exactly the same as what was
> read in.
>
>
>
> This can be demonstrated with code as simple as
>
>
>
> I = imread('pickoutcentralin 32bit.png');
>
> imwrite(I,'pickoutcentralout 32bit.png');
>
>
>
> then compare the histograms of the ingoing and outgoing images. Has anyone
> else experienced this problem? Is there a workaround?
>
>
>
> imwrite.m pretty much passes through its parameters to __magic_write__ so
> the problem is likely in __magic_write__. Where is the sourcecode for
> __magic_write__? I would like to take a stab at troubleshooting if a
> solution doesn't already exist.
>
The builtin is defined in
src/DLD-FUNCTIONS/__magick_read__.cc
Grundberg