Archived posting to the Leica Users Group, 1998/08/07

[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]

Subject: Re: [Leica] Slide scanners [OFF TOPIC]
From: Alexey Merz <alexey@webcom.com>
Date: Fri, 07 Aug 1998 14:47:20 +0100

Pascal wrote:
> 
> On 05-08-1998 13:01 Alexey Merz wrote:
> 
> >While high resolution will improve dynamic range for very small areas of
> >highlight or shadow, dynamic range is NOT solely a function of resolution,
> >and the Dmax for a given detection system is not resolution dependant.
 
> On 05-08-1998 15:29 John Chapman wrote:
> >Alex Merz wrote that dynamic range was more important than resolution
> >per se. Dynamic range is completely dependent on the resolution you scan
> >at that is a given.

John still says this, and it is still incorrect.

> Now, what is the exact relation between optical resolution and dynamic
> range?
> Take both Nikon scanners, they have the same optical resolution of 2700
> dpi, but there is clearly a difference in dynamic range (30 bit scans for
> Coolscan LS-20 and 36 bit scans for the LS-2000). This probably means
> there is less detail in darker, shadowed areas on slides and negatives
> for the Coolscan LS-20. It is obvious that such detail would be noticed
> at optimum resolution, but what about lower resolutions?

For slides the shadow areas have greatest density and the darkest zones
will be revealed as a series of dark tones with a higher quality scanner,
but as uniformly black (or only a small number of dark tones) with the
lower quality unit. Resolution AND dynamic range are important for 
rendering of small details, while dynamic range will be more important 
in rendering of larger structures. 

Now, it might seem that the opposite would be true for negative film, 
namely that you'd lose highlight detail with the cheaper scanner.
But actually, it's not so much of an issue because (color) negative
films tend not to reach the high densities that slide films do. 
So you could stick to color neg film and do just fine with the cheaper
unit.

> Suppose you are scanning for web publishing purposes, and you limit your
> resolution at 72 or 150 dpi. Two questions:
> - Would this limitation in resolution have an effect on dynamic range, or
> is dynamic range independent on the resolution setting?

Depends on the area of the extreme values. A *point* highlight will give 
a higher output value at 150 dpi because (in theory) all of that light
will be recorded by a single pixel. A 72 dpi pixel, with >4x the area
of the 150 dpi pixel, will record more dark area surrounding that 
same highlight, and take the _average_ of that whole area, resulting in
a LOWER recorded value. In this case, dynamic range is directly dependant
on resolution. 

Now let's take another example, a scene with a lot of snow.
this picture will have large areas of near-white, but probably little
fine detail in the near-white range. Here resolution is not so important
but the ability to make fine tonal distinctions (great dynamic range)
is very important. Now, this is not a hypothetical example. We have a 
rather old flatbed scanner with more than adequate resolution, but 
lousy dynamic range. See:

http://www.webcom.com/alexey/images/warwick600.350.jpg

The ground cover on this frame is completely blown out and this was the
best scan that I could obtain; the resolution is more than adequate,
but the dynamic range was wholly inadequate to show the tonal gradations
in the snow, which are clearly present in the original.

This scan was not really good enough for presentation, but I wanted
to show the *content*, so I posted it anyway.

> - Would the difference in dynamic range between the two scanners be
> noticeable at that kind of low resolution, and if no, from what level of
> resolution would you see the difference?

For large structures such differences can be noticable at very low rez.,
as in the example above.
 
> In other words, I don't doubt that the Nikon SuperScan LS-2000 is clearly
> the better of the two, but if you limit yourself to web publishing (I am
> not into pre-press), would you obtain lesser quality with the cheaper
> entry-level Coolscan LS-20?

Again, it comes down to your image source. If you scan a lot of slide 
film the pricier unit might be worth it. If you scan mainly negs, 
the cheaper unit will probably be adequate. 

One last thing: even with slide films, there will be differences. 
I routinely use Kodachrome in its toe (shadow) regions, while
I use Provia in its shoulder (highlight) regions. As a result, 
I expect that most of my Provia slides are more amenable to 
scanning than most of my Kodachromes - although for pictoral
work I generally have a strong preference for Kodachrome. 
..........................................................................
Alexey Merz | URL: http://www.webcom.com/alexey | email: alexey@webcom.com
            | PGP public key: http://pgp5.ai.mit.edu/ | voice:503/494-6840
            | ...I guess that's why I am pursuing a career in science - 
            | I just spend a lot of time in happy slackjawed wonderment,
            | the saliva dripping from my chin, pitterpattering onto the 
            | linoleum checkerboard of the lab's floor. Mom and dad always
            | told me I was special.