Saturday, September 24, 2016

digiKam 5.2.0 and new Red-Eye Effect Reduction

digiKam developers has just pushed digiKam 5.2.0 for public and it contains fixes for over 42 bugs listed in their bug tracker. Another good news about this new release is that red-eyes effect reduction process:
it introduced a new red eyes tool which automatize the
Faces detection is processed on whole image and a new algorithm written by a Google Summer of Code 2016 student named Omar Amin is dedicated to recognize shapes and try to found eyes with direct flash reflection on retina.

This new tool is available in Image Editor and also in Batch Queue Manager to be able to process a set of photos at the same time. The algorithm have been very optimized for speed efficiency and for small memory fingerprint. End user can adjust just a single parameter about the reducing level of retina red color to the average of blue and green channels. The default threshold have been tested successfully with a trial and error basis from a large data-set of images. So typically, no user interaction are need with this new version of this tool compared to older one, where no red eyes detection was performed, and a lots of manual settings was required to process this kind of correction.
I have pushed the change to my SlackHacks repository in order to build digiKam 5.2.0 along with the arch switch for x86. It will default to i586 instead of i486.

Feel free to test the new digiKam 5.2.0 :)