https://bugs.kde.org/show_bug.cgi?id=3D375824

            Bug ID: 375824
           Summary: Wallpaper image size choice is not optimal
           Product: plasmashell
           Version: 5.9.0
          Platform: Other
                OS: Linux
            Status: UNCONFIRMED
          Severity: minor
          Priority: NOR
         Component: Image Wallpaper
          Assignee: notmart@gmail.com
          Reporter: kde@happyjack.org
                CC: plasma-bugs@kde.org
  Target Milestone: 1.0

Wallpapers can include many different image resolutions to better match mon=
itor
reoslutions.  When choosing which image size to use, an algorithm is used w=
hich
takes into account both the aspect ratio as well as how much larger or smal=
ler
the image is than the monitor resolution.  Smaller images are penalized dou=
bly:
an image which is 200 pixels narrower than the desired image is counted the
same as an image which is 400 pixels wider.

I believe this algorithm is suboptimal: given a monitor size of 1920x1200, =
and
a wallpaper with two image sizes, 1680x1050 and 2560x1600, the smaller imag=
e is
chosen even though both have identical aspect ratios to the original.

The problem with this is that upscaling an image looks worse than downscali=
ng
an image, as sharpness is inevitably lost.  In
plasma-workspace/wallpapers/image/image.cpp, in the function =E2=80=9Cdista=
nce=E2=80=9D, the
following is the responsible code:

delta =3D (delta >=3D 0.0 ? delta : -delta*2 ); // Penalize for scaling up

I think that it's probably better to *always* prefer the larger image if the
aspect ratio is the same, but in any case, I think the multiplier here for
penalizing upscaling should at least be greater so that images which are
somewhat larger, but the same aspect ratio, are prioritized over smaller im=
ages
of the same aspect ratio.  In my case, a multiplier of 4 is sufficient, so =
that
2560x1600 is used on my 1920x1200 monitors, but again, I think it's probably
better to err a lot on the side of significantly larger images.

--=20
You are receiving this mail because:
You are watching all bug changes.=