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: Noise correction algorithms



Jerry wrote:

Jerry wrote:

>I noticed that the HP S20 software was able to paint e.g. in red all pixels
that were being clipped by current histogram mapping settings. To me this
seemed a handy feature, but no other software took over that idea. It seems
that if you can show the user which data is being clipped or is being
considered pixels-to-be-cleaned cq, IR-opaque-pixels, the user would be able
to precisely control if the correct pixels are cleaned.

>This would be a good feature for any owner of a filmscanner without IR.
Small problem is that you have to do a full-resolution pre-view, and the big
problem is... well,
unknown to me, but known to Ed and other software manufacturers. Perhaps it
has to do with patents, but what I hear is 'click-toot-toot...' and since I
hear that often I wonder: does anybody understand what I am trying to get
at?

My point, exactly. This seems to be a software solution, but it's being tied
to hardware/firmware. In (Amiga) programs with shallower bit-depth (which is
why I mentioned 256 colors, rather than the full 16-million+ in 24-bit
color), it is a relatively simple matter to pick a pixel color and *change*
it, globally or in a limited selection. I have found only one Windows
program that will let me do that at 24-bits (PP8), and "simple" it is not!

To me, it seems it should revolve around an "Ask" command that gets the
pixel color and compares it to its neighbors, then moves on--probably
time-consuming since it would involve line-by-line examination of the image
or selection--but doable. Why there should then be any general softening of
the image is irrelevant, if the pixel color or threshold colors are the only
ones affected. I can actually *do* a form of this in Basic (which is,
unfortunately, not recursive so it can't "back up" once it's reached a given
point), but of course Basic would probably take several days to examine a
4-megabyte image. ;-)

But to echo Jerry's thought, what am I missing, here? It seems a selectable
software algorithm would be 10-times better than one a machine decides for
itself.

Best regards--LRA


------Original Message------
From: "Oostrom, Jerry" <Jerry.Oostrom@Alcatel.nl>
To: "'filmscanners@halftone.co.uk'" <filmscanners@halftone.co.uk>
Sent: May 1, 2001 7:16:12 AM GMT
Subject: RE: filmscanners: Noise correction algorithms


I already asked this question to Ed and later to this list, all some time
ago. Ed replied that his algorithms were "of course" already doing such a
thing. Then I asked, where can you set the threshold on black (slides) or
white (negs) for what is considered to be dust and waited... (no answer to
that question).

This reminded me that I should put only one question in a mail (ironically
that was even Ed's suggestion to me, even longer ago), since then it clearly
shows both parties that the other person is not answering / missing that
question. But that is getting OT.

I noticed that the HP S20 software was able to paint e.g. in red all pixels
that were being clipped by current histogram mapping settings. To me this
seemed a handy feature, but no other software took over that idea. It seems
that if you can show the user which data is being clipped or is being
considered pixels-to-be-cleaned cq, IR-opaque-pixels, the user would be able
to precisely control if the correct pixels are cleaned. This would be a good
feature for any owner of a filmscanner without IR. Small problem is that you
have to do a full-resolution pre-view, and the big problem is... well,
unknown to me, but known to Ed and other software manufacturers. Perhaps it
has to do with patents, but what I hear is 'click-toot-toot...' and since I
hear that often I wonder: does anybody understand what I am trying to get
at?

"Bar Bar" applying to the civilized greeks

;-)



> -----Original Message-----
> From: Lynn Allen [SMTP:lalle@email.com]
> Sent: Saturday, April 28, 2001 5:00 PM
> To:   Filmscanners@halftone.co.uk
> Subject:      filmscanners: Noise correction algorithms
>
> This question is for Ed, and any other program-savy people who want to
> answer.
>
> Since dust is always "white" on negs and always "black" on slides, while
> "noise" is usually lighter and "grain" is usually darker than the
> surrounding field of pixels, is this or can it be considered in the
> cleaning
> algorithms?
>
> This suddenly seems so obvious as I experience the problems more, and I
> wonder what I'm missing that it isn't more easy to deal with. (?) Example:
> red pixels in sky colors, when it isn't sunset, green pixels in skin-tones
> and shadow tones at mid-day. It's very perplexing, because I'm pretty sure
> my scanner or its software is actually "seeing" or at least "interpreting"
> those pixels. I could, of course, be wrong, but that's how it looks to me.
>
> Best regards--LRA


-----------------------------------------------
FREE! The World's Best Email Address @email.com
Reserve your name now at http://www.email.com





 




Copyright © Lexa Software, 1996-2009.