From: Thomas Rauscher <pano@at...> - 2005-08-21 00:27:52
thx to Fulvio i have improved the filter routines, so now the code
doesn't disturb other applications. I just added some (16!) new
filters. The filter functions are "inspired" by GraphicsMagick and a
paper of Ken Turkowski.
I also improved the adaption for the filter size. I interpolate the
size from an half pixel, because this is the size, which gives the most
"power" to the filter function.
For now i only support full color and only 8bit per sample, but it should
be easy to add this. I think this is not necessary for the testing phase,
it just messes up the source code.
Also the current version has serious problems in the zenit and nadir
The source code of the patch and a patched "release" DLL can be found at:
Please give me your comments and ideas to these modifications.
From: Thomas Rauscher <pano@at...> - 2005-08-22 16:47:16
on Sunday, August 21, 2005, 02:27:46 I wrote:
TR> Also the current version has serious problems in the zenit and nadir
This is much more serious (and fundamental) problem.
I should have thought about it BEFORE i started to implement, but the
results in the "normal" regions were so impressive...
To solve this issues i would need the derivation of the function,
which i don't have. Last night i had an idea with the inverse function,
which is available, so now i am working on a hopefully better
Please just ignore the current patch. ;-)