Jon Wu created PDFBOX-1868:
------------------------------
Summary: Garbled / distorted fonts during PDF to image conversion
on recent versions
Key: PDFBOX-1868
URL: https://issues.apache.org/jira/browse/PDFBOX-1868
Project: PDFBox
Issue Type: Bug
Affects Versions: 1.8.3
Environment: Java 6
Reporter: Jon Wu
Attachments: script-font-bad-1.8.3.png, script-font-good-1.6.0.png,
script-font.pdf
Rasterized fonts have distorted curves and weird breaks in the letter when
converting PDF to image.
I'm running PDFBox in a web server environment to rasterize PDFs with text in
them. However, I've found that in recent versions, I end up getting garbled
text using the same code, same PDF, and same machine.
In script-font-bad-1.8.3.png, you can see that this big cursive F has a weird
break in it in the middle and all the curves aren't smooth which is really
obvious when you compare this bad image (1.8.3) with the PDF or a good one
(1.6.0).
See script-font-good-1.6.0.png or the attached PDF for the correct rendering.
What's weird is that I don't get the bad image when I use pdfbox-app.jar on the
PDF - it's only with my code. The PDF has an embedded TrueType (CID) font with
Identity-H encoding and was written by iText.
This also happens on other fonts, but I chose this example because it's most
obvious.
Good versions:
1.6.0 (I'm on this now)
1.7.0-SNAPSHOT from approx 10/24/11
1.8.0-20130225.201137-98
Bad versions:
1.7.1
1.8.0-SNAPSHOT from approx 3/1/13
1.8.3
and many others in between
I ended up downgrading to 1.6.0 because I had some issues with transparency on
some of the good snapshots that were working, but I think the transparency
issues are resolved on 1.8.3.
Tilman Hausherr mentioned this could be a regression of PDFBOX-1435.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
