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]

[filmscanners] Re: Film resolution - was: Re: 3 year wait



Austin Franklin wrote:

> But I know what I say IS true as it applies to scanners!  I design digital
> imaging systems, and have been for 25 years.  I have designed about a half
> dozen scanners, and aside from designing them, I also have done extensive
> testing and analysis on how these devices work.

Excellent!  An expert I can learn from and freedom to try
and get technical rather than attempting to talk in broad
terms and bad analogies. :-)

> I believe the part that is important here is that scanners to not sample
> points in time, they sample areas in space.

Yes, I agree, however at least in my mind there's a direct
mapping. But if there isn't a mapping how is Nyquist being
applied?


> That is why 4f is sufficient to
> %100 acquire the amplitude and frequency of a square wave of width f (1/2f
> being the width of the actual black line) with a scanner.  Why not try to
> put that down on paper and understand what it is I am saying.

I think what you say is correct, but wasn't my point.  Perhaps we've
got a heated agreement and I'm arguing against some other's point
and didn't notice your statements were different in detail. :-(

What I'm saying is that Nyquist isn't 2f. Also even in your example,
it's not clear that the original pattern will be necessarily reproduced.
With an alignment of 4f one can get gray-black-gray-white patterns
which isn't the input black-black-white-white pattern.  So what you
say is correct in that both full amplitude white/black extremes are recovered,
but the essence of Nyquist (which you probably know better than I) is
that if you're at twice Nyquist the exact pattern should be resolvable,
at least in theory.


> Not meant at all to be a "mine is bigger" thing, but I am an engineer, and
> have been for 25 years.  As I said above, I design digital imaging systems
> (primarily) for a living.

I'm at 30 years, but your experience is much more directly applicable
so you win at that.  I wasn't trying to do huffing and puffing routine, only
tried to say that I'm not a spring chicken burger flipper either. Get
technical on me IOW. :-)  What I dont' understand is why "2f" is the
proper use of "Nyquist".


> But what you are "reporting/analyzing" applies differently to scanners.  We

My point was that the frequency component in the data being scanned
is of higher frequency than "f", and therefore the Nyquist sample rate is
much higher than 2f.

My assertion is that Nyquist isn't applicable, so although
your statements are very correct about what one can do with 4f I don't 
understand
how that justifies calling 2f the Nyquist rate.

> aren't talking about switching thresholds here, or ringing, or overshoot, or
> undershoot, or knees.  The scanner sensors see the AVERAGED luminosity in
> their FOV, no matter what the components are in the FOV.

Are you saying that the averaging is in effect an input-filter that
reduces the input frequency such that "nyquist" applies (when one looks
at things in terms of post-sampling rather than looking at the thing
being sampled -- which has been my point of view)?

Mike

P.S. - I'm new to ICX SI software, previous ICX work had been done by 
contractors,
       I've been more a system/circuit designer, but I get to do my own
       SI work on my current project.  Fun learning new tools.

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body



 




Copyright © Lexa Software, 1996-2009.