![]() Registered Member ![]()
|
I'm an old fashioned pixel pusher so want to use the "sRGB built-in" colourspace for a traditional, naive colour intensity curve, however using posterize with "sRGB built-in" result in unequal segmentation.
I can get equal segmentation using the default colourspace, but then don't get the colour intensity curve I desire. It appears to me as though the posterize is operating in a linearized colourspace, regardless of image colourspace. Is there any way I can make it operate in my chosen colourspace? ![]() |
![]() KDE Developer ![]()
|
Oh, that's actually kind of interesting... Posterize works in 16 bits sRGB, and was written before 16 bits sRGB used a linear TRC by default, so, yes, it works in linear space, unintentionally.
|
![]() KDE Developer ![]()
|
Should be fixed in https://invent.kde.org/kde/krita/commit ... b871d02706
|
![]() Registered Member ![]()
|
Cheers, that's working as I was wanting.
![]() |
![]() Registered Member ![]()
|
Actually, I'm getting some serious slowdown with a posterize filter layer now.
![]() Speed is back to normal if I revert this commit. |
![]() KDE Developer ![]()
|
That's quite possible, because we're now not reusing the global color transformations, but recreate them especially for the posterize filter.
|
![]() Registered Member ![]()
|
I hackily moved the colour transformation creation out of the transform function into the constructor and filter layers are working at interactive speeds again.
|
![]() KDE Developer ![]()
|
please share the patch? I might just be prepared to apply it
![]() |
![]() Registered Member ![]()
|
Never created a patch before so might be screwy.
|
![]() Registered Member ![]()
|
Added patch to phabricator here: https://phabricator.kde.org/D21529
|
Registered users: Baidu [Spider], Bing [Bot], Google [Bot]