Re: 16 bits = 15 bits in Photoshop?
Re: 16 bits = 15 bits in Photoshop?
- Subject: Re: 16 bits = 15 bits in Photoshop?
- From: bruce fraser <email@hidden>
- Date: Thu, 14 Apr 2005 12:04:37 -0700
At 11:13 AM -0700 4/14/05, Ray Maxwell wrote:
I have been following this discussion with great interest.
However, lets get real.
Those who have read books by Dan Margulis or his Colortheory forum
know that he believes that 8 bits is all you need to reproduce
smooth, non-posterized images. While I don't agree with him and use
16 bit per color all of the time, I do think that 16 bits is severe
over kill for a 2.2 gamma space that is approximately linear with
respect to perception. This statement assumes that the final
reproduction is going to be on reflective media.
What Dan's tedious and fundamentally specious arguments deliberately
miss is that the need for greater bit depth has absolutely nothing to
do with reproduction and everything to do with editability.
We can capture much more than we can reproduce. That's why
photography is an art. Higher bit depths let us decide which parts of
the tonal range we've captured we wish to emphasize and which parts
we wish to compress.
Bob Caspe used to posit the example of a woman standing in full
sunlight, wearing a white lace dress, and holding a black
leather-bound book in shadow. That's maybe a 20-stop dynamic range.
You can't reproduce that dynamic range, but the compromises you make
in reproduction depend on the story you're trying to tell.
If the story is Paris Hilton getting married, it's about the dress,
and you expose and develop for the highlights.
If the story is Mrs Smith finding a Gutenberg Bible in her attic, the
story is the book and you expose and develop for the shadows.
But if you can capture a wide enough dynamic range, you can have it
all, and make those decisions post-capture rather than pre-capture.
We need 32-bit floating point channels for those situations where we
want to capture or model real-world scene luminances with contrast
ratios of 100000:1 or more, and manipulate them. With most cameras
avaliable today, that means combining multiple exposures. That won't
always be the case.
At least one camera available today has a 22-stop dynamic range. Check out
www.spheron.de
I've seen this camera in action, it's real, and 16 bits per channel
isn't anywhere near enough to do it justice. (It uses three trilinear
arrays to essentially combine three exposures on the fly.)
Another interesting technology that blows the old assumptions about
how many bits we really need is from
http://www.sunnybrooktech.com/
I've seen this display in action, and while it's a technology demo,
it's also real. I fully expect that sometime in my lifetime, such
displays will be mainstream means of image reproduction that I'll
hang on my wall.
100 L* isn't a dynamic range. It's simply maximum white relative to
whatever your white reference is, and is hence a relative rather than
an absolute value. There ARE papers that can produce a white that is
brighter than L* 100 unless we take steps to define L* 100 in a way
that avoid it-fluorescence takes incoming UV energy and downshifts it
to visible light, allowing more than 100% reflectance. I've seen a
yellow ink with 175% reflectance. (It doesn't profile well!)
256 steps is probably sufficient for printing to reflective copy from
a tone-mapped 11-stop capture. It's nowhere near enough to work with
linear-gamma scene data that will be reproduced in a movie theater
(even an analog one), or to get optimal tone-mapping from a capture
that really holds the scene dynamic range and is displayed on one of
the above displays.
--
email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden