I confess to not being 100% sure which one you thing is in some way
"unexpected".
I *think* you are complaining only about the large size rendering of
cambria in 3.png
BEFORE the fix and that it is equally blurry with or without the fix.
I don't have my usual tools handy to check but that can be as simple as
Cambria
at 48 pixels is unhinted, so turning off hinting makes no difference.
But then no difference is not exactly what people usually mean when they
saying there is a regression ... so I may be misunderstanding you.
BTW only 1.png shows JDK version so its not clear what was used to
render each case.
-phil.
On 4/7/20, 8:31 AM, Mario Torre wrote:
Hi Phil and all,
I was working on backporting this to 11:
https://bugs.openjdk.java.net/browse/JDK-8214481
When I realised this is actually a regression in that it makes
rendering on OpenJDK worse in some cases.
I've been testing with 8, 11, 13 and 15 and 11 patched and comparing
those the fonts rendering becomes blurry when fractional rendering is
used:
http://cr.openjdk.java.net/~neugens/JDK-8214481/1.png
http://cr.openjdk.java.net/~neugens/JDK-8214481/2.png
A you can see the rendering in all cases is cleaner, the font is both
blurry and thicker with the patched version.
This is not the case, however, with the font provided by original bug
report:
http://cr.openjdk.java.net/~neugens/JDK-8214481/3.png
In this case indeed the rendering is better at lower sizes (but not at
higher, where the same blurring is apparent).
I filed a new bug to track this with a reproducer that is derived from
all the various test cases floating around 8214481:
https://bugs.openjdk.java.net/browse/JDK-8242285
Cheers,
Mario
