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: Polaroid Optical Density



David,

I commend you for revealing the methodology that Polaroid uses in
determining the reported OD of your SS4000 scanner.  I think it goes a
long way toward de-mystifying the process, and also provides other
manufacturers with a possible structure to work from.

I have only one question, based upon something you alluded to.  What
percentage of errors in reading the density wedges on the Velvia film
does Polaroid allow in measuring the noise floor?  In other words, at
what point does Polaroid consider the noise level high enough to
consider the wedge beyond consideration in the OD readings?

Art

"Hemingway, David J" wrote:
> 
> I have done this before on this forum and hopefully you will find it use
> full to understand how Polaroid determines optical density. I invite other
> manufacturers to be as forthcoming!
> The first thing to realize it the OD is NOT just a density reading, noise is
> also factored.
> Polaroid uses a custom designed target that contains a white patch and a
> whole bunch at the higher density range. This target is expose in , I think,
> Velvia which has the highest density of any E6 film available at the time
> the test was designed. The actual density values as measured using a
> densitometer or spectrophotometer. The target is then scanned and
> statistical analysis is made of every pixel in each patch. As the density
> gets higher the percentage of pixels that are noise increases. This is the
> point where cheating can occur. Statistically how much noise you allow and
> still accept the value as acceptable. You can literally take a single data
> point and manipulate it to give substantially different OD specifications.
> This is where multi-scanning comes in. I will now reveal one of my personal
> gripes. Multi-scanning is generally perceived as "good" and a desired
> feature. In my view it is there to compensate for poor design and/or
> inferior components. Do I wish we had it, yes but only to satisfy the
> "checkbox" buyer. Simply, multi-scanning attempts to identify the pixels
> that change  from scan to scan, These pixels are most likely noise and when
> statistically identified as such can be removed and replaced with values for
> adjacent pixels. While multi-scanning can produce registration  problems,
> more likely it will produce a slight softening of the image particularly at
> higher DPI. Some time ago when Ed Hamrick was writing the driver for the
> SS4000 he tried to implement multi-scanning and he said because of the low
> noise level of the SS4000 multi-scanning produced little advantage.
> 
> In a practical sense let me relate my real word experiences with regards to
> this subject. I work/attend trade show all over the world for Polaroid.
> Invariably potential customers come up to me and ask if they can scan
> "their" slide to look at the shadow detail. They then will try to do the
> same at other manufacturers booths. In the three or more years I have been
> doing this I have NEVER lost this type of shootout with a comparable
> scanner.
> 
> To me the bottom line is :
> 1.      Do not make any decision based on manufacturers specifications
> 2.      If possible do the test yourself or rely on a good reliable
> independent source.
> 
> Thanks for you time
> David





 




Copyright © Lexa Software, 1996-2009.