Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

   


   


   















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

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

Re: filmscanners: pushing dynamic range on the nikon 4000ed



At 10:53 AM -0700 18-11-01, Wayne Williams wrote:
>my specific questions relate to extending the dynamic range of the nikon
>using a work flow that involves bypassing nikon color management and
>building profiles...


Hi Wayne--

I've had my Nikon LS4000ED for about two months now and have spent 
most of my time with Kodachrome 64 slides trying to (a) pull the most 
detail out of the shadows, (b) get the most accurate color and tone 
off the slide, (c) get accurate softproofing on the monitor, and (d) 
get prints that most accurately reflect the original state of the 
slide.  I figure that once I can accurately reproduce the slide at 
each stage, then I can shift my focus to deciding what I WANT the 
image to look like and start manipulating tone, color, etc. confident 
that what I see on the screen is what I'll get on a print.

I'm using NikonScan 3.1 on a Mac, Photoshop 6, an Epson 1200, Epson 
inks and Photo paper, ColorVision software and hardware for monitor 
calibration/profiling and printer profiling, and Profile City's ICC 
Scan for making scanner profiles.

Here are a few thoughts:

o I get the most detail out of the shadows, and best color accuracy, 
by (a) setting the film type to Kodachrome, (b) scanning with color 
management off, (c) applying a custom scanner profile I made from a 
Kodachrome target.

o In NikonScan, choosing "Kodachrome" as the film type makes a big 
difference.  It makes NikonScan display reasonably accurate preview 
images.  It has become obvious to me that Nikon is modifying the 
color data coming off the CCD, before delivering it to you, based 
upon the film type.

o Do a preview scan then look at the histogram in the Curves panel. 
If you don't have histogram data extending all the way to the right 
end of the scale then bump up the Analog Gain and update the preview 
until the histogram data extends to just short of the right edge of 
the scale.  When done properly this can make a vast improvement in 
the shadow detail and overall tonality of the scan.

o I've made dozens of experiments with many variations on using 
Scanner RGB, playing with the gamma in the Curves panel, setting 
black and white points in the curves panel, etc.  And the three steps 
above seem to produce the best results so far.

o Some preliminary experiments seem to indicate that my custom 
profiles do well even if I make gamma and color balance changes in 
NikonScan before doing the final scan.  But don't hold me to that.

o Making custom printer profiles has greatly improved the quality of my prints.
o One of the things that has helped the most is that a professional 
friend mailed me some source files and reference prints made from 
those files.  This gives me an excellent way to judge the performance 
of my profiles and to see where I needed to change the tone or color 
balance and regenerate the profiles.

o Another very important benefit to having reference prints is that I 
can view them side by side against my prints IN ANY REASONABLE LIGHT. 
That is, I don't need special color temperature and intensity 
calibrated lighting to to very accurate and critical color 
comparisons.  Household halogen lights, cloudy or bright sunlight, 
etc. will all work (but avoid flourescent lights for this).

o Despite my prints of his source files being very close matches to 
his reference prints, I find that dark purples in one of my own 
images (scanned as above on the Nikon) are far darker than what I see 
on the screen.  This may be because my viewing lights don't have 
enough purple spectrum, or something else.  I'm still working on 
this.  Other than that I'm getting excellent tone and color in my 
prints.

o Your VIEWING conditions are extremely important to this process.

o To TRULY see what the monitor is displaying: there must be no glare 
on the monitor (get some black foam core board from Office Depot and 
make a hood!), the ambient light hitting the screen must be several 
times less bright than the brightness of your monitor, room lights in 
general should be very dim so that your eyes will adapt their 
sensitivity to the screen's brightness, and the visual field 
surrounding the monitor should be color-neutral.  There's more but I 
think these are the most important points.  This is really important.

o To TRULY see if a print you made matches what's on the screen: (a) 
the light you view the print with has to have the same color 
temperature as the white point of the screen (or vice versa), (b) and 
the intensity of the viewing light reflecting off a white sheet (say 
of your printer paper) must match the intensity of pure white 
displayed on your monitor, (c) the color spectrum of your light 
source and the dyes/pigments in the print have to be "compatible" 
(see next item).

o By the way, a cool trick for comparing a scan displayed on the 
screen to the original slide is to open an empty window with a white 
background and hold the slide up to it, using the monitor as a light 
table that is perfectly matched in color temperature, spectral 
distribution and intensity to...your monitor.

o The dyes/pigments in your prints produce within your brain a 
sensation of color based upon the frequencies of light they reflect. 
So if your light source is missing some of the frequencies that your 
prints reflect those areas of your prints will shift in color and 
tone due to the apparent darkening of the hues in those areas. 
Conversely, if your light source has bright spikes (such as all 
flourescent lights produce) at certain frequencies that your print 
reflects well, then those hues in the print will be brightened and 
result in color and tone shifts.  This is what leads to metamerism, 
where the colors/tones of a print can be different under various 
lights (sunlight, flourescent, halogen, etc.).

o So to compare prints to what you see on the monitor you need a 
light source with a full spectrum, no spikes in the wrong places and 
the same (or similar) color temperature as your monitor.  I don't 
have this problem solved yet, but the Solux bulbs from Tailored 
Lighting are looking promising.

o On the other hand, even if you get a perfect match between your 
monitor and a print viewed under your special viewing light, you 
might find that the dyes/pigments in your printer's inks produce 
distinct color shifts when viewed under other lights.  Some of the 
Epson models have used inks where this effect is particularly 
noticeable.  The bottom line here is that if you're making prints for 
a particular, known viewing environment you should view them under 
the lights used at that environment and make any necessary changes; 
even if the results look wierd on your monitor.


o In summary, I've gotten the best scans OF KODACHROME SLIDES by (a) 
turning off color management, (b) setting the appropriate film type, 
(c) Setting the analog gain if necessary, (d) Applying a custom 
profile to the final scan.  I've gotten the best prints by tweaking 
my profiles to match professionally produced reference prints.  Room 
lighting and print-viewing lighting makes a huge difference in 
whether what you see (on the monitor or on a print) is really what 
you've got.

Good luck,

--Bill



-- 

======================================================================
Bill Fernandez  *  User Interface Architect  *  Bill Fernandez Design

(505) 346-3080  *  bill_sub@billfernandez.com  *  http://billfernandez.com
======================================================================




 




Copyright © Lexa Software, 1996-2009.