[Poppler-bugs] [Bug 68986] pdftoppm doesn't render eci_altona-test-suite-v2_technical2_x4.pdf correctly

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=68986

--- Comment #2 from Thomas Freitag <Thomas.Freitag at alfa.de> ---
Created attachment 85949
  --> https://bugs.freedesktop.org/attachment.cgi?id=85949&action=edit
knockout transparency group with subgroups which use blend modes

The patch consists of two parts:
1. The changes in SplashOutputDev::beginTransparencyGroup and Splash::pipeRun
accomplishes that the correct backdrop color will be used for multiplying.
2. The changes in SplashOutputDev::paintTransparencyGroup and the extension of
Splash::blitTransparent is optimization: because of antialiasing effects when
drawing "knock out" the contour of "knock out" becomes visible where it
overlaps "text to". The opimization removes the antialiasing effects in the
overlapping area.

I've regtested the patch and a few tests (only 10) fails. After looking at the
failed tests I could classify them into three groups:
1. I can't see the differences in the "diff" png, because the differences are
too small.
2. I see some small color changes. I couldn't submit that the new color or the
old color is the right one, but I'm given to the new color.
3. I used the openjpeg integration for my regtests and in some of my tests I
got damaged rendering outputs when creating the reference png. This always
happens in area of JPEG2000 compressed images, and these kind of failed tests
have differences in exact that area. When I configured with
--disable-libopenjpeg and use my patch, the output seems to be okay, but I
haven't recreated the references without the usage the openjpeg integration and
rerun the regression test again.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/poppler-bugs/attachments/20130917/15a0ab04/attachment.html>