Excuse me, but without seeing details (see
http://nagoya.apache.org/bugzilla/bugwritinghelp.html) I would consider the bug
as INVALID. Provide a reproducible example, please.

Created attachment 3921 [details]
example fo which exhibits the problem

Works fine for me, of course except for
http://doradb1/dora/MediaContentHandler?type=image&id=1&mimetype=image/jpeg
Chances are the problem is there. Try to run without this image, if it helps,
check  that MediaContentHandler returns image correctly (mime-type,
content-length etc).

Created attachment 3926 [details]
Correct sample which exhibits the problem.

Created attachment 3927 [details]
Example servlet which retrieves images.

Using the provided example servlet, RetrieveImage, I can successfully retrieve 
an image to the browser.  However, using the example fo provided, I cannot 
successfully retrieve the image to a pdf document.

Your servlet works very inaccurately with those buffers:
byte[] chunk = new byte[4096];
while (fis.read(chunk) > 0)
{
    dos.write(chunk);
}
Think what a garbage you are outputting in a very last iteration. 
When you are copying buffers always make sure you are writing the same amount of
bytes you are reading. Use the following pattern:

int bytesRead;
byte[] chunk = new byte[4096];
while ((bytesRead = fis.read(chunk)) > 0)
{
    dos.write(chunk, 0, bytesRead);                    
}

RetrieveImage was simply meant as an example (cut and paste from the "real" 
code).  I guess that's what I get for not looking at it thoroughly.  The 
corrected stream read/write "fixed" the problem.
Thanks Oleg for your diligence.

batch transition to closed remaining pre-FOP1.0 resolved bugs