Bug 224293

Summary: wavelets NR tool confuses red and blue
Product: [Applications] digikam Reporter: S. Burmeister <sven.burmeister>
Component: Plugin-Editor-NoiseReductionAssignee: Digikam Developers <digikam-bugs-null>
Status: RESOLVED FIXED    
Severity: normal CC: caulier.gilles
Priority: NOR    
Version: 1.1.0   
Target Milestone: ---   
Platform: openSUSE   
OS: Linux   
Latest Commit: Version Fixed In: 1.1.0

Description S. Burmeister 2010-01-26 09:27:50 UTC 
Version:           1.1.0 svn (using KDE 4.3.4)
OS:                Linux
Installed from:    openSUSE RPMs

svn Rev: 1079999


I noticed a lot of blue spots in the black areas and thus set luminance and Chrominance red to 0, i.e. only used Chrominance blue, yet the blue spots did not go away. Then I set blue to 0 and red to some value >0 and the blue spots disappeared.

So here is my dumb question, am I wrong assuming that red should get rid of the red spots and blue of the blue ones and that there is a bug that those two are interchanged?

Then I thought that maybe a treshold of 0 means that it will erase all kinds of blue. Yet if I set both red and blue to 0 I get more blue noise compared to blue set to 0 and red to some value >0. Which seems odd as I did not change any value for blue thus blue noise should stay at the same level.
Comment 1 caulier.gilles 2010-01-26 09:34:17 UTC 
I agree, sound like something is inverted somewhere.

To be sure, can you test with this gimp tool if problem is the same, because it's the same algorithm used (in YCbCr color space)

http://registry.gimp.org/node/4235

Thanks in advance

Gilles Caulier
Comment 2 S. Burmeister 2010-01-26 10:44:37 UTC 
Ok, I think I got it.

Gimp shopws some tooltips and explains that the treshold is the value below
which everything is considered noise. I think it might be useful to add this to
digikam as well.

Further. If I set Y and Cb to 0 I get the full blue noise. Increasing the
treshold for Cr does not change anything about that so it seems that digikam
has a bug.

Increasing Cb gets rid of the blue noise.
Comment 3 caulier.gilles 2010-01-26 11:39:01 UTC 
SVN commit 1080460 by cgilles:

CRed and CBlue parameters are inverted
BUGS: 224293


 M  +20 -20    noisereductionsettings.cpp  


WebSVN link: http://websvn.kde.org/?view=rev&revision=1080460