Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

   


   


   















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

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

Re: filmscanners: VueScan Question



I don't mean to question your authority on this, since I don't own a 
2740 and you probably have worked with one, however, I am trying to 
understand the mechanism of this situation.

I understand that dICE works by doing a comparison of the infrared image 
and the "visible" image and does some sort of subtractive process, to 
assume that stuff that is on the one and not the other is likely dust, 
dirt, fingerprints, scratches, etc.

I also understand that the concept of multipass is to "average out" the 
random noise a CCD produces in the visible spectrum, by basically 
assuming any pixel which alters value through a number of scans is 
likely an artifact of noise rather than "real" data.

What I'm not clear on is this... Does the infrared scan tend to have 
similar amounts af shadow noise during its data acquisition?  In other 
words, is it really necessary to do 4 or 8 or 16 infrared scans of an 
image to get a more accurate infrared scan of the image?  If so, 
wouldn't that mean that dICE actually adds a certain amount of random 
noise (or random artifacting) to the scan when doing a since pass scan. 
  After all, if the visible light scan makes an image and during that, 
the shadow area data contains a certain amount of erronious random 
"noise" artifacting, and then the dICE infrared scan does the same 
thing, wouldn't that introduce a second level of errors, since the dICE 
scan would have different random pixel data (noise) than that of the 
visible light scan, and when the subtraction formula occurs, some data 
would either be subtracted that should not have been, or not subtracted 
that should have been?

Again, I'm guessing, but I would be surprised that the infrared light 
scan would have very much noise artifacting in it.  I'm therefore also 
guessing that taking one infrared scan and then simply using it as the 
subtractive model for all the multi-visible light scans would make 
little difference in the final result.

Ed, if you have access to an Acer 2740 and can write you software to do 
this (take one infrared scan, and then average all the visible light 
scans and then do the subtractive process once) you will find almost no 
difference in the final result, and probably no meaningful difference.

Coming from this with no way of documenting it, but using "gut science" 
;-) only, I very interested if such a test could be carried out.

This would tremendously help any 2740 owners when doing multi-scans with 
dICE.

Art


EdHamrick@aol.com wrote:

> In a message dated 6/3/2001 10:55:46 AM EST, michael@shaffer.net writes:
> 
> 
>> Unlike Nikons, doesn't this scanner insist the IR channel scan
>>  separately from the RGB scan ... ie, a 2nd pass.  I thought the
>>  original post was stating, if he wanted 16x RGB passes, it also
>>  scanned the IR 16x.  There would indeed be no need for this.
> 
> 
> There's no point to increasing the bit depth of the RGB data
> if the IR data doesn't have the bit depth similarly increased.
> The two are combined when doing dust removal.
> 
> This is why there are the same number of passes for the
> RGB data as the infrared data.
> 
> Regards,
> Ed Hamrick





 




Copyright © Lexa Software, 1996-2009.