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: VueScan clipping & flat images



> -----Original Message-----
> From: owner-filmscanners@halftone.co.uk
> [mailto:owner-filmscanners@halftone.co.uk]On Behalf Of Bob Armstrong
> Sent: 22 November 2001 15:13
> To: filmscanners@halftone.co.uk
> Subject: Re: filmscanners: VueScan clipping & flat images
>
>
> The following is taken from VueScan's help file.  It seems to me
> that B&W points of 0% should leave all of the pixels in the scan.
>  (Having said that, when looking at a histogram, it does seem
> that some pixels either go beyond the black point or there are a
> mumber of pixels _at_ the black point.)

I would agree with this.  The intention is quite clearly to make the data
fill the range of possible values.  For reasons analogous to the use of
16-bit scans (really 10, 12 or 14 bits, generally): to maximise tonal
smoothness and provide resilience under further editing.

>
> It seems to me that setting B/W to 0% results in what Jawed is
> calling a flat image.

Yeah.

> Jawed (if you're reading this!)

Yes... I'm burning the candle at both ends, so to speak.

> does a
> histogram of one of your flat VueScan images extend all, or most
> of the way, from 0 to 255?

Yes, generally.

> I find with my flat VueScan mages
> that do not range across the histogram but leave, particularly at
> the white point end, a lot of unused values.

Sometimes I get these too.  Depends on what experimentation I do.  Depends
on gamma apart from anything else.

I think what you're saying is that you define "flat" scans as those with no
black and white point settings (i.e., a scan in which the entire dynamic
range of the sensor output has been captured, with no attempt at "auto
exposure".  "Colour balance: None" in Vuescan, with a suitable gamma choice,
e.g. 2.2.).

So far, I've intended "flat" to mean any scan with no clipping at both black
and white points of the image.  The amount of space left between the data
and the black and white end stops isn't particularly relevant to "my"
definition.

Does the list want to define a "flat scan"?  Is there a better term?  What
are the parameters that define an "all data" scan?

(I can think of one major point which is caused by using a black or white
point setting:  By manipulating black and white points on a per-channel
basis you are immediately altering the colour relationships of the image.
This, in my experience, throws a major spanner in the works when it comes
time to perform colour cast removal, e.g. skylight correction.)

> If that is the case
> with your images I wonder if you would try to increase the white
> point to 0.05 or 0.1 and see what that does to the image.  I
> would be interested in your results compared to Nikonscan
> especially if you set the white point to _just_ fill the histogram.

I have used these techniques extensively in my experiments.  It is worth
bearing in mind that one can set the options on the dialog in Photoshop so
that pressing Auto gives you a 0.1% white point (or 0.05% or whatever you
choose).  So you don't have to keep re-scanning or re-previewing in Vuescan
(slow!!) to see the effect of this concept. Suffice to say that "auto" is a
technique I never use seriously, because I can build better curves.

The vast majority of my images (all negative scans) from Nikon scan don't
actually have any serious clipping of either black or white in them (e.g.,
100-5000 pixels of clipping in an image with 30 million pixels).  So with
these images, there seems, to me, no point in using Vuescan to try to obtain
a flat scan.  (Sure, there might be other reasons why the user chooses to
use Vuescan, but "all the data" isn't relevant in this case).

When an image from Nikon Scan does clip either black or white points (or
both) seriously, what I find is the "choice" that Nikon Scan has made is,
effectively, the choice I would have made using Vuescan.  Except that Nikon
Scan does a better job than I can with Vuescan.

When you set the white point (say, to 0.1%) you throw the colour balance of
the image off.  Sometimes this corrects for the lighting of the scene, e.g.
tungsten or skylight.  But more often it partially corrects and introduces
"confusion" in the colour.  To finish off the colour correction is then a
complete pain, because the auto level has basically gone behind your back
and screwed the colour up.

Jawed




 




Copyright © Lexa Software, 1996-2009.