Am 26.08.2014 15:32, schrieb Tilman Hausherr:
We have problems recognizing YCCK / CMYK jpeg files.
I can't find the issue quickly, it was a few weeks ago from a french
person and about an image about a Porsche event.
https://issues.apache.org/jira/browse/PDFBOX-2128
Tilman
Anyway, what worked for me (most of the time) is to use Apache Imaging
to detect the image type.
I wrote "most of the time" because it is not perfect, although better
than java.
https://issues.apache.org/jira/browse/IMAGING-136
<https://issues.apache.org/jira/browse/IMAGING-136?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel>
I'm not sure if Apache Imaging is still an active project.
Tilman
Am 26.08.2014 14:34, schrieb Timo Boehme:
Hi,
checking the rendering capabilities of PDFBOX 1.8 vs. current trunk I
came across a journal which showed severe problems in both - but
different. Problems of 1.8 are gone, new ones showed up.
While the journal (Chemical&Engineering News, C&EN) does not provide
free PDF editions a sample edition can be downloaded via 'View a
sample issue' at http://cen.acs.org/static/about/digital.html (or
directly via http://www.cendigital.org/cendigital/sample/). I'm
referring to volume 92, nr 27 from 2014-07-07 which I downloaded
yesterday but the same problems also showed up in other journal issues.
The problems (all on Linux, Java 1.6):
- PDFBOX 1.8 (svn 1620380)
- first letters of words in headlines are sometimes missing, e.g. on
page 2 "Getting ..." reads " et ing ...", "Overview" -> " verview"
- bad character spacing because of substituted font
- PDFBOX trunk (svn 1620415)
- no missing letters but heavily distorted and displaced
letters in headlines (e.g. page 2)
- compared to 1.8 correct font is used
- picture colors are completely wrong;
logged warning: org.apache.pdfbox.filter.DCTFilter decode
WARNUNG: Inconsistent metadata read from JPEG stream
- transparent background instead of white
- PDFBOX trunk, no-awt svn 1620487
- font rendering ok
- picture/background problems as in trunk
Since these are multiples problems on different versions and the PDF
is not freely distributable I did not create a JIRA issue.
Nevertheless it is a widely distributed journal and a good test case
for the rendering quality. At least the JPEG rendering problem of the
current trunk should be solved.
Best,
Timo
