SVN commit 518579 by cgilles:
digikam from trunk : New powerfull Noise Reduction tool based on experimental dcamnoise2 gimp plugin algorithm. You can forget the old implementation based on despeckle algorithm. This plugin support 16 bits image. I have fixed a lot of problems from the original implementation. The plugin have been tested indeep into valgrind to prevent memory leak. The setting limits have been adjusted to prevent crash in any case (for example divide by zero). TODO : - We have the capabilty to apply a noise curve correction to the image. A new curve widget need to be created for that (for the future, not 0.9.0 release). - Add save/load settings to a text file. - Polish settings area (to group by action/type) - Add settings description using QWhatsThis CCMAIL: [hidden email] M +229 -96 imageeffect_noisereduction.cpp M +12 -11 imageeffect_noisereduction.h M +663 -359 noisereduction.cpp M +47 -14 noisereduction.h _______________________________________________ Digikam-devel mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-devel |
On Tuesday 14 March 2006 16:16, Gilles Caulier wrote:
> SVN commit 518579 by cgilles: > > digikam from trunk : New powerfull Noise Reduction tool based on > experimental dcamnoise2 gimp plugin algorithm. You can forget the old > implementation based on despeckle algorithm. > > This plugin support 16 bits image. I have fixed a lot of problems from the > original implementation. The plugin have been tested indeep into valgrind > to prevent memory leak. The setting limits have been adjusted to prevent > crash in any case (for example divide by zero). > > TODO : > > - We have the capabilty to apply a noise curve correction to the image. A > new curve widget need to be created for that (for the future, not 0.9.0 > release). - Add save/load settings to a text file. > - Polish settings area (to group by action/type) > - Add settings description using QWhatsThis > > CCMAIL: [hidden email] > > M +229 -96 imageeffect_noisereduction.cpp > M +12 -11 imageeffect_noisereduction.h > M +663 -359 noisereduction.cpp > M +47 -14 noisereduction.h > > > _______________________________________ I've tested the new plugin for some time now, and I think it is really good, and certainly better than the old algo. I've not been able to study the various parameters and to identify a best default setting. This is a tedious task with so many parameters at hand (for example, after using Refocus routinely, I have found my optimum default values only now after one year (which I might suggest for svn, by the way)). On the other hand I was able to achieve the same quality with the Restoration plugin, except that it need much more computing time. Gerhard _______________________________________________ Digikam-devel mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-devel |
On Friday 24 March 2006 10:11 am, Gerhard Kulzer wrote:
> On Tuesday 14 March 2006 16:16, Gilles Caulier wrote: > > SVN commit 518579 by cgilles: > > > > digikam from trunk : New powerfull Noise Reduction tool based on > > experimental dcamnoise2 gimp plugin algorithm. You can forget the old > > implementation based on despeckle algorithm. > > > > This plugin support 16 bits image. I have fixed a lot of problems from > > the original implementation. The plugin have been tested indeep into > > valgrind to prevent memory leak. The setting limits have been adjusted to > > prevent crash in any case (for example divide by zero). > > > > TODO : > > > > - We have the capabilty to apply a noise curve correction to the image. A > > new curve widget need to be created for that (for the future, not 0.9.0 > > release). - Add save/load settings to a text file. > > - Polish settings area (to group by action/type) > > - Add settings description using QWhatsThis > > > > CCMAIL: [hidden email] > > > > M +229 -96 imageeffect_noisereduction.cpp > > M +12 -11 imageeffect_noisereduction.h > > M +663 -359 noisereduction.cpp > > M +47 -14 noisereduction.h > > > > > > _______________________________________ > > Hello Gilles, > I've tested the new plugin for some time now, and I think it is really > good, and certainly better than the old algo. I've not been able to study > the various parameters and to identify a best default setting. This is a > tedious task with so many parameters at hand (for example, after using > Refocus routinely, I have found my optimum default values only now after > one year (which I might suggest for svn, by the way)). > On the other hand I was able to achieve the same quality with the > Restoration plugin, except that it need much more computing time. > I'm currenlty in discution with the new NR algorithm author to improve it. I will CC to you all messages for informations. It's very interressing !!! Gilles _______________________________________________ Digikam-devel mailing list [hidden email] https://mail.kde.org/mailman/listinfo/digikam-devel |
Free forum by Nabble | Edit this page |