Mike Engles wrote:

ftp://ftp.alvyray.com/Acrobat/9_Gamma.pdf

Yes, very good information, from a rather heavy weight professional,
his Bio btw is at: http://alvyray.com/Bio/default.htm Even if it was
written in 1995 it is perfectly valid today (only that in case the
image has an embedded ICC profile we need not to *guess* the transfer
function, ICC color-management was not very popular at that time).

They must be transmitting/recording in at least 18 bit. That is
the bit level that Chris Cox et al say is the minimum necessary
for linear images, without gamma encoding.

No more bits are necessary for digital imaging than what the sensor of
the acquire device is able to provide (according to it's S/N ratio) in
other words there is no need to store pure noise. Say you buy 12 eggs,
then you do not need a trailer truck to bring them home. And these
days the so called pro scanners and pro digital cameras can not reach
even 10-bit. Ther real pro devices (like the EverSmart Supreme ll
scanner that has cooled CCD) can do nearly 12-bit.

And the so called banding issue in 8-bit/c is linear is enormously
exaggerated, it is in fact quite an academic case:Here is an example
from a thread on my forums:

16-bit/ edit:
http://www.aim-dtp.net/aim/temp/sad_...edit-16bit.jpg
8-bit/c edit
http://www.aim-dtp.net/aim/temp/sad_...-edit-8bit.jpg

The original was a linear converted RAW from D60. You can read the
details from the thread:'Linear workflow and 8bit/channel' if you like
to go there. Rather demanding picture in regards to the horrible
banding issue, can you see *any* problems in the 8-bit/c edit?

It does seem that what we have today is two types of digital imaging.
One is the truly scientific one that uses ALL linear data.

Yes, the scientific imaging is done in linear.

The other is a convenient engineering one that delivers the goods
simply, by pre compensating the linear data to display on non linear
displays.

This the _easy solution_ for the ordinary consumers. It is *not* an
engineering issue but a marketing issue.The industry simply needs a
way to sell the digital imaging gadgets to the mass market consumers
without stressing the consumers with the workflow issues.

And the third type of digital imaging is the high end professional
imaging that you see in the better magazines etc. It is still done in
linear domain like it has been done for the past 30 years. For the
very reasons Dr. Alvy Ray Smith lists on his above mentioned memo like
"all computer graphics computations assume linear images", this
includes Photoshop CS also. When the computations are applied over
gamma compensated image data there will be the Gamma Induced Errors.

Timo Autiokari