> On Aug 18, 2015, at 4:20 PM, Noel Chiappa <j...@mercury.lcs.mit.edu> wrote:
>
>> From: Sean Caron
>
>> I have found that even fairly fine detail reproduces okay with a 300
>> DPI scan ... there's no need in scanning with extraneous bit depth and
>> then you start to get people complaining about file sizes
>
> I have found that one can generally have one's cake, and eat it too:
> if I scan at 600dpi in black and white, and then use "CCITT Group 4"
> compression, the resulting images (of prints) are ~200KB per page.
> Is that small enough? :-)
If you can, avoid black/white scans. The reason is that scanners are often
noisy in that mode, so you end up with very large images where the “white”
areas have lots of individual black pixels on them. Copiers used in scan mode
are particularly likely to do this. Such documents are also surprisingly hard
to read, look messy when printed, and utterly fail OCR.
Grayscale works much better in all respects.
You can generate clean bitonal scans, but that requires careful
post-processing, in particular careful (and often page-dependent) selection of
the threshold value.
paul
