Barak Pearlmutter -
> So you're saying that this is a theoretical bug, in that if the
> document were scanned at an even higher resolution it is possible that
> this problem would occur even at the default losslevel of 100?
What I'm really worried about is that the problem might occur with
losslevel 100 even at 400 or 600 dpi.
> The documentation has already been updated (in CVS) to discuss this.
> But basically it seems like an old story. "Patient to Doctor: It
> hurts when I use losslevel >100! Doctor's response: Well, don't use
> losslevel > 100."
The analogy would be appropriate if the problem were as easy to spot
as current pain.
If I were confident that losslevel of 100 would prevent the problem,
that would be okay. However, I don't consider u and n to be easily
confused. If the program makes that mistake, it might be making more
subtle ones too, just less often. Mistakes that are very hard to spot
by eye.
I'm setting out to scan AMS55 (Abramowitz and Stegun), with its
hundreds of pages of numerical tables. Even a very small chance of
character substitution (say, replacing a 3 with an 8 on every tenth
page) would call the whole project into question.
- Jim Van Zandt
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
