Hi Fernando and all,
> From: Fernando Gonzalez Gentile
> In my case, an horrid noise was 'added' :-)
Seems strange for a _software_ issue... but then, video digitizing is such
an esoteric task for a computer...
well, I do care and mess about on these - my later explorations were in Lab
> mode, but lost my initial enthusiasm- and no, despite Mr. Ling having
> advised me on the contrary, I still get aRGB files from my 4000ED which
> don't look pale at all -
May comparision came from chiping the pics taken in sRGB versus aRGB, on the
300D. Obviously, due to the broader gamut of aRGB, colours look less
saturated so it reduces the risk of 'clipping' but increases processing
artifacts, esp. posterizing. I much prefer the first one, which may happen
in analog systems too, while the second is a digital-only issue.
It's like if your record on analog equipment at a very low level... you'll
avoid any chance of overload, but will notice a lot of noise.
As someone (can't remember now) mentioned later, 16 bit/ch is a MUST for
wide-gamut spaces, and also when heavy corrections are expected. However,
for the final image or straight scans with little processing, 8 bits/ch is
OK. For a sample of the need for extra bits on heavy corrections, see the
pics starting with 'a' at <http://cjss.sytes.net/post/tabernas/> -- C*ntax
RTS2, Tessar 45/2.8, expired Fuji Astia 100F :-(
lol - this is Physics, Carlos: we should have a reasonable reason, not a
> most probable reasonable reason.
But in Quantum Physics, we could only get the "most probable reasonable
reason", am I wrong? ROTFLMAOL!!!
This is just what Ken advised me shortly before you re-joined. In fact, I
> could finally get the Fuji Frontier icc profiles which are currently used in
> Montevideo.
ICC profiles are great and could save you a lot of work and tweaking. But
what if tou miss a single profile of the chain? Or, more likely, if due to
age or whatever the device no longer matches closely its profile? Empirism
is the only way ;-)
> But _ what if I am not in the need of printing ? Recently, I've been asked
> for some photographs to be displayed only, hopefully on a Viewsonic or
> Samsung lcd ... The way both you and Ken suggest, looks too empirical for me
> - which does not imply that I discard it.
As soon as you have calibrated the screen _and_ previous steps on the chain,
should be reasonably OK. For instance, you may photograph a well know
picture (an IT8 card should be great) in controlled lighting with your
favourite film(s). Then try to scan it several times, playing with the
scanner and/or PS and/or monitor controls (taking notes of ALL settings, of
course) until you see someting that resembles the original...
A word of caution on visual comparitions... Metamerism is already mentioned
in another thread. As I understand it, things that look the same colour
under a certain light, may differ under a different one. Sometimes this
could be very tricky. I remember matching Cibachrome prints with the
original slide, under those blue "daylight" bulbs... but the match was no
longer under sunlight :-(
> Understood: correcting an error with another error ... all too common. I
> should move to Utopia :-)
But that IS the way to go! ;-)
Remember you are checking an additive against a subtracting method, beware
> of gamut limits - I know you are aware of this.
There's another surprise here... LCD monitors are, in fact, _substractive_
RGB devices... and their light source is usually a fluorescent lamp, which
has a poor colour quality -- very far from a "black body" source. LED based
illuminants may have an even worse spectrum.
And the word 'native' is misleading: when you read ColorSync, (IIRC) there
> it says apple native gamma is 1.8 while pc is 2.2 etc. Looks more like a
> value to fix anything on, as if it was a 'constant' in Physics. Well, right:
> a constant is a value one must use to get to the real measurement figure
> .... (!?).
The 2.2 gamma comes from this...
A CRT gun's response is far from linear. If we call L the "brightness" of
the screen at a certain point, and V the signal voltage at the input, both
magnitudes will be related (ignoring constants) by the formula (sorry, no
superscript available here):
L = V ^ 2.2
that means: V to the power of 2.2 -- something difficult to understand
without logarithms.
Should the input signal be on a linear scale, shadows would be very dark and
lacking in contrast, which will increase a lot in the highlights. Because of
that, TV cameras and broadasting in general are already compensated for that
unlinearity, applying the original signal (S) a function like:
V = S ^ (1/2.2)
to compensate. This signal is said to be "gamma corrected" for a 2.2 value.
Output devices other than CRTs may behave differently. LCD screens are
_very_ far from linear, with a rather complex response curve. But bulit-in
circuitry compensates for this, usually targeting to a 2.5 gamma -- that is,
a bit tougher than a CRT. However, some LCDs like the Acer X163W are much
more corrected.
Now, a bit of speculation. At late 80s / early 90s most PC video cards
(_plain_ VGA was top of the line back then) seemed to send _uncorrected_
signals to the monitor, in a linear scale. Since the monitor circuitry
didn't compensate at all for this (they were little more than tunerless TV
sets) pictures were rendered with the aforementioned unlinearity (L =
V^2.2). For a MS-DOS screen, Win 3.x icons or a typical game this was no
issue, but for photograph processing it was.
There were the times of PhotoShop 1.0 (!) and Macs were the most suitable
for that task... it _seems_ that their circuitry (in the video cards and/or
the monitors) did some gamma correction, although not in ful. If the applied
formula was something like:
V = S^(1 / 1.22)
then this signal thru a typical CRT of L = V ^ 2.2 would render a somewhat
brighter image, which would need a further
V = S ^(1 / 1.8)
to be fully corrected (linear). Please note that 1.8 x 1.22 = 2.2. Until
very recently, it was normal that a jpeg (or whatever format) image looked
somewhat darker on a PC than on a Mac screen.
Nowadays, things have changed. Most PC video cards are as (colorimetrically
speaking) capable than those on Apple computers. And internet calls for
uniformity. It's said that both Win and OS X aim now to an intermediate
gamma correction, like 2.0, so recent pictures won't look too bad in older
computers of both platforms...
Many people live happy provided they don't ask or get to know anything
> different.
Of course! Consciousness can be the real enemy of happyness... I believe
that's the reason of drugs etc.
All that said, I'd like to end quoting a paragraph from Wikipedia -- this
may not be the most reliable source, but after having confronted it against
my own experience, makes a lot of sense to me:
"As the recommended color space for the Internet, sRGB should be used for
editing <http://en.wikipedia.org/wiki/Digital_image_editing> and saving all
images intended for publication to the WWW."
(from <http://en.wikipedia.org/wiki/Srgb>)
So, unless you're going to four-inks press, stick to sRGB and stop sending
duplicates (of both colour spaces) to Flickr! :-) :-)
Cheers,
--
Carlos J. Santisteban Salinas
IES Turaniana (Roquetas de Mar, Almeria)
<http://cjss.sytes.net/>
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|