Re: filmscanners: Nikon MF LED light source...

[tflash <tflash@earthlink.net>]

> There's this mantra that capturing the scan
> data in 16 bits obviates all other
> responsibilities at the scanner-driver stage,
> and I've never bought into that.  Seems I
> get by nicely with 24-bit (8 bit/color) scans,
> in spite of all I read here and elsewhere
> about the advantages of 48-bit scans.
> 
> There's no question that 48-bit files will
> allow *certain* scan shortcomings to be
> fixed up around later.  OTOH, it's easy enough
> to obtain 48-bit scans that are fatally
> flawed from the get-go.
> 
> 
> rafe b.

I believe with the leaf, calibration and focus aside, when you get an HDR
file you are getting all that it can get from the scan. You will never need
to scan separately for the highlights, and for the shadows, cause you've
already gotten as deep into them as the scanner is able to deliver. There is
nothing you can do to the scan in the driver that you can't do in PS, so at
that point you can go by which interface you prefer.

In my tests, doing the manipulations in PS did NOT give better results than
doing it in the driver. IOW, if the files were made to look the same, you'd
find the same amount of noise, and the same histogram either way. (why not,
both you and the driver have been working on 16-bit files) But to that point
all your moves have been global. When you start with the raw scan, when both
images are the *same*, you are still in 16-bit, so you can do your selective
moves in 16-bit too. Thus, getting a globally nice 8bit file will be the
same either way, but a heavily manipulated file with selections, will be
better by staying in 16-bit as long as possible.

Whether the difference shows up in print is another matter. I think it
depends on the quality/tonality/character of the original, and the degree of
manipulation you do to the file.

Todd