> On Thu, 29 Mar 2007 22:31:50 +0200, Jim Sabatke <[EMAIL PROTECTED]> wrote:
>> I've just compiled a plugin based on a denoising program that attracted
>> slashdot's attention a short while ago.
>> http://rss.slashdot.org/~r/Slashdot/slashdot/~3/99611463/article.pl
>> It gets very good reviews compared to Noise Ninja and other commercial
>> products.
>> I was wondering if it is good enough that you might be willing to
>> include it in the standard distro?  The only holdback I can see is that
>> it is written in C++.
>> I've got it running on SuSE 10.0 and there is a windows version
>> available online.
>> The author has no objections to including it.
>> Jim
>> _________________________________
> very interesting filter.
> I just did a quick comparison rescaling some of thier sample images with 
> lanczos interpolation in gimp.
> Generally, lanczos was inbetween cubic and the denoising filter. The 
> latter usually produces images that are easier on the eye than both 
> cubic and lanczos but on closer inspection this is not without a price.
> Denoising grossly simplifies the image in cleaning it up. The girl in 
> hat image, res_lana.png becomes badly distorted and the result is worse 
> than both lanczos and cubic. (This is going by thier images on the demo 
> page).
> Ringing, while present, is less than gimp lanczos.
> However, quite a bit of detail is lost and the biggest defect seems to 
> be constast takes a bit hit.
> The overall effect is pleasing and probably would be good for a lot of 
> applications.
> Like any image processing it's a case of horses for courses and what 
> defects in the result are acceptable in a specific application with a 
> specific image.
> Thanks for bringing this up. A useful filter to have.
> gg.

Thanks for the analysis.  This code is still undergoing active 
development, with students scheduled to keep it up.  I'm passing your 
comments to the development team.

Gimp-developer mailing list

Reply via email to