Re: colorsync-users digest, Vol 2 #76 - 13 msgs
Re: colorsync-users digest, Vol 2 #76 - 13 msgs
- Subject: Re: colorsync-users digest, Vol 2 #76 - 13 msgs
- From: Don Hutcheson <email@hidden>
- Date: Thu, 07 Dec 2000 10:01:21 -0500
Roger and Andrew,
In theory Roger is right, but in practice Andrew is.
Let's first agree that scanner dynamic range can be practically defined as
the range of density levels discernable in a test target or live image.
Roger's belief that bit depth defines dynamic range assumes that no analog
processing occurs before quantization. If the scanner digitizes a direct
linear output from the CCD, then Roger wins the cigar. This is the
assumption in most scientific imaging texts.
However, if the scanner first applies any kind of processing - not
necessarily a gamma LUT but perhaps a simple gain and offset - then Andrew
wins, because at that stage the analog signal of even an 8 bit scanner can
be compressed to boost very dark detail into the fixed range of the ADC
(Analog-to-Digital Converter). Most scanners apply a separate gain and
offset to each pixel before digitizing to calibrate white and dark levels
and eliminate streaking. This gain/offset stage can also extend the dynamic
range to any density range available in the raw analog signal.
Many practical tests have shown that even a cheap 8 bit scanner can
comfortably exceed it's theoretical dynamic range limit, proving that
dynamic range is more dependent on analog processing than the number of ADC
bits.
To test high dynamic range, see my RGB Scanning white paper at ProfileCity
.com.
Regards,
Don
*************************************
Don Hutcheson
Hutcheson Consulting
(Color Management Solutions)
11 Turnburry Rd
Washington, NJ 07882
Phone: (908) 689 7403
Fax: (908) 689 5305
email@hidden
*************************************
>
From: email@hidden
>
Reply-To: email@hidden
>
Date: Wed, 6 Dec 2000 07:55:27 -0800 (PST)
>
To: email@hidden
>
Subject: colorsync-users digest, Vol 2 #76 - 13 msgs
>
>
> Andrew Rodney wrote
>
> ADynamic range and bit depth have absolutely no correspondence. You can have
>
> a 10 bit scanner with a dynamic range of 3.8 and a 12 bit scanner with a
>
> range of 3.3. Completely different spec's. Also be aware there is no
>
> universally accepted way of measuring dynamic range.
>
>
I have to disagree with you, Andrew: dynamic range and bits are completely
>
interrelated.
>
>
A 10 bit scanner can represent 1024 discrete levels (2 to the 24 power)
>
which corresponds to a dynamic range of 3. 12 bits of data, on the other
>
hand, yields 4096 discrete levels which is somewhere between 3 and 4, in
>
dynamic range. 14 bits of data (2 to the 14 power) is all you need to
>
accurately represent the 10,000 discrete levels of a dynamic range of 4.
>
>
I agree this is aside from any implementation issues and marketing hype on
>
the part of hardware manyfacturers.
>
>
I would suggest you turn to Phil Green's excellent "Digital Photography"
>
book, recently published by Pira (ISBN 1-85802-207-X), for a short, to-the-
>
fact discussion on this concept (page 29, last paragraph).
>
>
Regards.
>
>
Roger Breton
>
Quebec Institute of Graphic Communications