Thread: "16-bit" mode.
View Single Post
  #224  
Old November 28th 04, 12:11 AM
Mike Engles
external usenet poster
 
Posts: n/a
Default

Mike Russell wrote:

Mike Engles wrote:
[re alvy gamma article]

Hello

Yes there does seem to some confusion about PC Gamma, but he [Alvy] is
absolutly clear about the need for linear processing. As for not
understanding Gamma encoding, that is not clear from the article. He
has
been around a long time, and does know a thing or two. If gamma
encoding
were that important he would have mentioned it. He is totally clear
about not applying any non linearity to a image, just to the display
device. I assume from this he means the same for storage, but I don't
know.

Suffice to say that he and his collaborator are MAJOR digital graphics
imaging authorities who on the face of it supports Timo Autiokari's
lonely stance. His last words in the gamma article are telling.

Mike Engles


Right on. Alvy has another article, written in 1995, that goes into further
detail re gamma issues:
http://www.cs.princeton.edu/courses/...s/smith95d.pdf
. In this 1995 article, Alvy states:
"Nonlinearity should never be stored in an image. Or, if it is, then this
nonlinearity must be noted in the storage format in such a way that it is
known how to remove it to retrieve linear data."

This comment, as fundamental as it is to graphics algorithms, plus others
relating to the concept of working versus display space, came years before
Photoshop 5 commercially introduced the concept of working spaces, as part
of color management. Not bad for someone who "doesn't understand gamma".

As for Timo's lonely stance - he appears to be in good company now, having
been debunked together with Alvy Ray Smith and Dan Margulis, all in the
space of a few days. :-)
--
Mike Russell
www.curvemeister.com
www.geigy.2y.net



Hello

This curiuosly was the article I was referring to.
I was not sure of the legality of quoting from it. Since you have I will
also.

These are the last words.

"Gamma can be confusing, as the above probably illustrates. Here are the
simple rules Altamira Composer uses and what I am advocating that
imaging applications do as a matter of course: Images are always assumed
to be linear. Gamma is applied only to the display of images and not to
the data of the images.The display is assumed to be nonlinear (because
it is). Applications separate computation from display cleanly, and
gamma correct for the local display only in the display process.

To get compatible results between imaging applications written under the
(I trust you believe sensible) “new” guidelines offered here and those
written the “old” way: Set the monitor gamma assumption in all the “old”
imaging applications to the same (greater than 1) value—presumably to
that matching one’s usual display monitor. Most applications provide a
way to do this. This transfers the nonlinearity correction in those apps
from the computation process to the display process, as it should be,
leaving linear data in the images themselves.
A desirable consequence of all this is that it would be very convenient
for imaging software if display devices provided gamma correction tables
settable by software. That way, each imaging app could work completely
in linear space,knowing that the display step would be correctly
compensated by the local monitor for its local nonlinearities.

Believe it or not, this was the way it was done
20 years ago, but the idea got lost along the way, leading to the mess
described in this memo. Unfortunately, it is probably too late to
change. The technique offered here is the best that can be done short of
changing all the hardware."

It seems that there was a differnt way and the only people who use it
now are the scientists.

I honestly do not know which is right and do not know enough to be able
to know, but feel in my bones that the old way was right.

Mike Engles