Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 




Did you know how to mine xrp - just check on Beaxy.com

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

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

Re: filmscanners: =?iso-8859-1?Q?Re=3A=20autolevels=20was=20re=3A=20filmscann...

Tony, I have no quarrel with your statements (I almost never do) except with
your two-phased processing approach to restore detail in a specular
highlight.  It won't work with a true specular highlight which is what the
original question was about, but only with a diffuse highlight.  A specular
highlight, by definition, is the reflection of a light source by a
"mirror-like” surface.  Specular highlights include such things as sunlight
reflected off a chrome bumper on a car, sunlight reflected off a ripple in a
lake, or a studio light reflected off a model’s wire rim glasses.  Assuming
the light source doesn’t have any inherent detail of its own, then its
mirrored reflection won’t either.  So there’s no detail to recover and it’s
OK to let the specular highlight go solid white on the final print, which
means that it’s not a problem if the white point is set so as to clip
specular highlights.  In addition, the specular highlight is bound to be so
much brighter than the film can record for the exposure being used that it
will have been “burned into” the film and would have lost any detail if it
had any to start with.  For transparency film, the specular highlights will
be the transparent film base. A diffuse highlight, on the other hand, will be
almost as transparent as the film base, but not quite, and will be capable of
holding detail.  So the diffuse highlight is the thing to use when setting
the white point. The white point should be set so that a bright diffuse
highlight, which will have detail, is almost, but not quite, clipped.  That
will allow it to be as bright as possible in the final print, but not so
bright as to lose its detail.  (If you don’t have a bright diffuse highlight,
but only a darker grayer one, then setting the white point becomes a bit of a
trial and error process as another post has indicated.)  I hope this isn't
beating a dead horse (my apologies to the horse) and that someone on the list
will find it useful.  The bottom line is that specular highlights ought to be
ignored when setting the white point since its their nature to "blow out" and
go completely white.  Instead, concentrate on setting the white point for the
defuse highlight.  Similarly, the black point should be set so as to capture
detail in a shadow area that actually has detail, but black areas without
detail should be clipped to oblivion.  It's the stuff in between that gives
me the most difficulty.

Boy I'm long winded, aren't I.  And out of sorts, too.  I just got my new
SprintScan 120 today and it has a "high density" SCSI connector that isn't
compatible with the connectors on my SCSI card or the SS4000 to which I want
to daisy chain it.  I need to buy a new cable or an adapter, but can't do it
until next week due to deadlines I have to meet.  It's like to the scanners
were designed by two different Polaroid groups.  Argggg.

In a message dated 7/26/2001 10:01:04 AM Pacific Daylight Time,
TonySleep@halftone.co.uk writes:

> If so, why is a point of sun reflection in a photograph not a good
> point to use for the white point?  Because it's not representative of the
> majority of the image?

Yes, it is massively out of step with the brightness range of the rest of
the image. If you set it as a white point, the rest of the histogram will
likely be scrunched up to the left - better to clip the specular highlight
and adjust for the rest. Or selectively mask and separately adjust two
layers then recombine, if you want to keep some detail in the specular


Tony Sleep


Copyright © Lexa Software, 1996-2009.