Print artifacts being generated with PDF-Booklet 2.1
Brought to you by:
averell7
Menu
▾
▴
#3 Print artifacts being generated with PDF-Booklet 2.1
FYI When producing a small, but mildly complicated booklet, small artifacts appeared on the printed output. The artifacts were small hourglasses about the size of one character. The initial pages of output were okay. The artifacts increased towards the end, maybe 50 or so on the final (lower right) panel of the last page. This was a 40 page booklet being printed as 10 pages using 2 rows and 2 columns, all in landscape. The most distinguishing thing about this project (and likely precipitating this problem) was that the pages (panels) were being printed completely out of order, eg. 40,1,38,3 then 32,39,4,37 then 36,5,34,7 then 6,35,8,33 etc. (I did this so I could generated a small booklet cut to 8-1/2 x 5-1/2).
Great program by the way.
Discussion
Anonymous
The artifacts are in the PDF by the way, so it's not a printer issue. The artifacts do not appear in shaded or blank page areas. They appear as an overprint on the character content. Specifically, the output was generated using "x pages in line" 2 rows, 2 cols. I tried using PDF-Shuffler to reorder the pages into print order so that PDF-Booklet can print 1-40 in simple order. The artifacts are still there, but not near as many. I've attached a zip file containing the test.pdf input file (re-ordered for use as page 1-40) and the test-bklt.pdf output file. Look at the lower right panels in the last few pages of output. I'm using Win XP with a single core AMD XP 2600+ cpu. Hope that helps you isolate the problem.
The artifacts are characters. You can see that each time you have one of these artifacts there is a missing letter. Example on the last page, < and > are replaced by the artifact.
It happens in the same way on my system, so it is not due to your hardware.
The problems appear when you have more than two pages on a sheet.
So the problem is with the handling of fonts by the library PyPdf. This is a difficult matter. I am not the author of PyPdf but I had to modify it seriously for the merging of pages, and a difficult point is merging the ressources from different pages on a single page. When you have four pages, merging is done three times. For an unknown reason something goes wrong after the second merging.
Pdf is sophisticated and I am not sure I can fix the problem.
You're right, the artifacts (hourglasses) are substituted for characters, but overprint on the next character which is shifted left. I find it curious that an hourglass is often used as a "busy" indicator AND that the problem gets worse with more complex documents. What would make more sense to me is that an hourglass is being used as a place holder until the actual character and size is determined in a later pass. My wild guess would thus be that merge/overlay passes are prematurely terminating with more complex documents.
Unfortunately it is very difficult to fix since the bug is not in our code but in the library and debugging pyPdf is no fun... This is a quite complex work because Pdf format is complex
PdfBooklet 3.0 (not yet released) has fixed this issue by adding a slow mode which is more careful in the merging of objects names.
There is no date planned for releasing this version, but the code is on Github (project : Pdfbooklet, not Pdfbooklet2 which is version 2)