Subject: Re: Digital Darkroom under Linux
From: Christopher E. Brown (cbrown@woods.net)
Date: Sat Oct 19 2002 - 23:30:37 AKDT
On Thu, 17 Oct 2002, Peter Q. Olsson wrote:
> I do a fair amount of image editing using linux tools, gimp mostly. I have a
> nikon coolpix 990 3.2 Mpixel camera, and a new Sony model???? 5.1 Mpixel camera,
> a nikon LS2000(?) slide scanner, and an older HP scanner. Also have an epson
> 1270 photo printer.
>
> I use Xsane from gimp to do the scan captures. It has been my experience that
> once you get a good transfer function (color balance bias) for the scanner in
> question, the scanning is fairly straightforward. I also find that scans take a
> lot more post-capture manipulation that do the digital cameras.
>
> With either of my cameras, I have not found much of a problem at all with color
> balance, except in low-light situations, and the images have other problems as
> well. I am sure that this varies from camera to camera. I use a pcmcia card
> adaptor to download images from the camera memory modules. Insert the card and
> mount as a vfat file system.
The cameras raw can be a bit closer. I have found VueScan to be much
better than Sane for filmscanners. It handles things a bit cleaner,
and does a very good job of compensation for the films base color.
Once I select the film type, or better yet use a clear area to
calibrate to the film stock the colors are *right* on.
> The gamut of the printer is a fair amount different than that of the
> monitor/graphics card, and to get a print that looks similar to the screen often
> takes quite a bit of image editing. Not sure if this is an artifact of the
> printer or gimp-print. I suspect the latter.
Unfort GIMP support for ICC profiles and emulated display of
CMYK is non-existant. One must properly calibrate ones display, and
X just to get proper display. Then colorspace conversions come in...
Evil!
> As to 24 vs 48 bit color, I question, perhaps naively, that 48 bit color is all
> that useful, except in a few critical applications. The varying response
> functions of the display media (CRT, lcd, print) from one instance to another
> probably overwhelms the subtle differences enabled by 48 bit color.
It depends. The dynamic range of color negative film is
greater than 24bits. (Slide film is not so much of an issue). Often
if you can scan at 30bits or greater you can putt out alot more detail
in the shadow and highlights. For example, the sky is all washed
out, some is nicely visible, and the shadows are black. If you do a
36bit scan, then control the shift to 24bit you can pull the details
out of the shadows and the highlights, rather than losing one or the
other.
A good example would be an 8 shot sunset sequence I shot last year
overlooking the great salt lake in Utah.
In prints or a 24bit scan you have a nice looking sun, a few hints of
the clouds, and a very dark bottom of the frame.
With a 36bit multipass scan you get the very bright sun (it is fully
in the frame in the first few), *and* you can pull out all the clouds
(even the colors), and the ground in front of the camera and its
feature are clearly visible. *MUCH* closer to what the scene looked
like to the eye.
I really don't bother with home printers. They are ok for scrapbook
stuff, but for something I would hang on the wall I send out to a shop
with a Canon photo printer costing *at least* what I make in a year.
-- I route, therefore you are.--------- To unsubscribe, send email to <aklug-request@aklug.org> with 'unsubscribe' in the message body.
This archive was generated by hypermail 2a23 : Sat Oct 19 2002 - 23:30:22 AKDT