Created attachment 3207 [details]
GPG Key for Igor Foox

Created attachment 3271 [details]
firefox image corruption

Problem in second pdf is "On page 12, the text after the figure is missing" 

As the zlib decoder seems to have problems may i suggest to disable it for the 
moment? 

http://bugs.kde.org/show_bug.cgi?id=111156 suffers from the same issue  

It looks like this problem is caused by inline images. An unlimited EmbedStream
is created for the inline image and the FlateStream reads beyond the end of the
inline image into it's input buffer. This causes things to go out of sync
afterwards. This isn't trivial to fix, so I agree with Albert, the zlib decoder
should be disabled by default.

We could also just fall back to just reading one byte at a time.

That might work, but when writing the code originally I remember having problems with zlib and one byte 
buffers. The right way to solve it is probably to have a new EmbeddedImage stream that specifically looks 
for and of the embedded image and only gives FlateStream data up to that point.

Should be fixed now.