Looking at the diffs between 6.4.5-7 and 6.4.5-8, I think this patch
may be the problem:
@@ -1203,9 +1203,9 @@
exception);
if (image->mask != (Image *) NULL)
clone_image->mask=CloneImage(image->mask,0,0,MagickTrue,exception);
+ clone_image->cache=ReferenceCache(image->cache);
return(clone_image);
}
- (void) SetImageExtent(clone_image,columns,rows);
scale=(MagickRealType) clone_image->columns/(MagickRealType) image->columns;
clone_image->page.width=(unsigned long) (scale*image->page.width+0.5);
clone_image->page.x=(long) (scale*image->page.x+0.5);
The old code set the columns and rows in the clone image_then calculates
the scale as 150.0/750.0 or 0.20. The new code removes the call
to SetImageExtent(), so the scale value used is 1.0.
On Wed, Nov 19, 2008 at 02:54:47PM -0600, Peter Fales wrote:
> In moving from 6.4.5-7 to 6.4.5-8, we've noticed a change in the behavior
> of the convert command. I can't find anything in the mailing list or
> ChangeLogs that would lead me to believe that this was intentional.
>
> Start with a 640x480 png image, then run:
>
> convert foo.png -thumbnail 150x150 foo.gif
>
> With the 6.4.5-7 version of convert, running identify on foo.gif would yield:
>
> foo.gif GIF 150x113 150x113+0+0 8-bit PseudoClass 64c 3.89kb
>
> but with 6.4.5-8, the result is:
>
> foo.gif GIF 150x113 640x480+0+0 8-bit PseudoClass 64c 3.21kb
> ^^^^^^^
>
> Since the "screen descriptor" is still set to 640x480, the web
> browser is treating it as a 640x480 image, though only the 150x113
> pixels in the corner are actually used.
>
> --
> Peter Fales
> Alcatel-Lucent
> Member of Technical Staff
> 2000 Lucent Lane
> Room: 1C-436
> Naperville, IL 60566-7033
> Email: [EMAIL PROTECTED]
> Phone: 630 979 8031
--
Peter Fales
Alcatel-Lucent
Member of Technical Staff
2000 Lucent Lane
Room: 1C-436
Naperville, IL 60566-7033
Email: [EMAIL PROTECTED]
Phone: 630 979 8031
_______________________________________________
Magick-bugs mailing list
[email protected]
http://studio.imagemagick.org/mailman/listinfo/magick-bugs
