On 11/17/06, Lars Aronsson <lars@aronsson.se> wrote:
Just some basics here: When you digitize old books, the greatest
loss is in the paper being torn, yellow of age, stained by coffee,
and ink having faded. Next step, the scanning or photography
always loses part of what the printed page contains. After this,
it doesn't matter quite so much if the compression of the computer
file is "lossy" or not. Don't get religious about "lossless"
compression. A good result can still be achieved by paying
adequate attention to every step of the process. But
perfectionism just doesn't pay. If you get unreadable results,
you need to go back and redo the steps that failed.
Formats like GIF, JPEG and TIFF existed in 1990. DjVu and JBIG2
are more advanced formats that have appeared later than that and
are still not very common in free software applications.
If you failed in using DjVu, it's not the format's fault.
I'm not saying it is. What I'm saying is that while I realize there will be a loss in quality in every step (including converting to different formats), the image quality *up to* converting to DjVu (from say GIF or PNG) is relatively good. So, if the quality of the DjVu image, using only the freely available DjVu software I can find, is not at all appeasing, then it would be better for aesthetic reasons to keep with the non-DjVu images and do our best to crunch down the filesize (PNGs can get some good compression). That way we can still keep high quality scans around that people will actually want to look at.
Maybe the software you can buy is much better (and I wouldn't doubt it), but I'm not going to drop $400 to get a copy; I'd rather just stick with a more "convential" format.
Z