ðòïåëôù 


  áòèé÷ 


Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

  óôáôøé 


  ðåòóïîáìøîïå 


  ðòïçòáííù 



ðéûéôå
ðéóøíá












     áòèé÷ :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: filmscanners: Scanner resolution (was: BWP seeks scanner)




> McNamara quotes 60 lp/mm.  I arrived at my initial 75 lp/mm by
> doing a braindead calculation: 4000/25.4/2.  My calculation is
> obviously optimistic - I'm assuming that adjacent rows (or
> columns) of pixels each resolve "1 line".  I would have expected
> (bit being bits and all that) that any degredation from that
> theoretical maximum would have involved a division by 2.  That's
> obvously not the case with McNamara's number.
>
> So the question is:  what factors would conspire to lower the
> resolution fractionally like that?  The suspects I come up with are:
>
> - CCD bloom
> - An interaction between the analog detail on the neg and the
> discrete CCD pixels (leading to something like dither if I'm not mistaken)
> - System noise
>
> Anyone have any ideas?

There are two factors that decrease the resolution of the red channel.
Smear and bloom.  Aside from the red channel, what you are referring to is
aliasing (your statement about "interaction".  That is taken care of by
sampling frequency, and why you sample at twice the smallest detail you want
to acquire.  Noise won't cause a decrease in detail, but will cause
inaccurate tonal values...different dimension.

BTW, CCD smear and bloom are only in the horizontal direction, not in the
vertical.

4000SPI (samples/inch) divided by 25.4 mm/inch = 157.48 samples/mm, which
means it can always resolve a detail that is (157.48 samples/mm divided by 2
for sampling frequency divided by 2 for line pairs) = 39 lp/mm is the
minimum resolution that a 4000SPI scanner can resolve.  That is for line
pairs that are perfectly horizontal or vertical.




 




Copyright © Lexa Software, 1996-2009.