[opendtv] Re: RGB mania
- From: "Alan Roberts" <roberts.mugswell@xxxxxxxxxxxxxx>
- To: <opendtv@xxxxxxxxxxxxx>
- Date: Tue, 18 Jan 2005 19:25:37 -0000
Craig, this is nonsense.
Video transfer characteristic has a "toe" defined by the noise floor of the
amplifiers and adcs, and a "shoulder defined by the clipping level. If that
were the only thing to consider, you'd be mostly right, but most video
cameras use non-linear pre-processing that simulates the "shoulder" response
by compressing highlights, so you effectively get a curve not unlike a film
curve. Notionally, that is straightened out in digits before actual gamma is
applied, but the gamma curve is no more a pure power law than is the overall
curve of film. Power laws are involved, but with offsets and limits.
By the way, have you guys worked out how many angels can dance on the head
of a pin yet? This thread is rapidly getting as interesting and significant
as such a discussion.
----- Original Message -----
From: "Craig Birkmaier" <craig@xxxxxxxxx>
To: <opendtv@xxxxxxxxxxxxx>
Sent: Tuesday, January 18, 2005 2:51 PM
Subject: [opendtv] Re: RGB mania
> At 9:55 AM -0600 1/17/05, Doug McDonald wrote:
> >Craig, you lose! Negative film gamma is EXACTLY "the way we take the
> >range of samples from any
> >source and map them to the limited range of the encoding system".
>
> Let's try to agree that there is a relationship between Film Gamma
> and Video/Display gamma.
>
> I agree that the notion of film gamma is similar to that of video
> gamma, as least in so far as film gamma attempts to describe the
> process of mapping brightness into a narrower range of densities.
>
> But we are also talking about acquisition systems with VERY different
> response characteristics. Unfortunatley, there is no toe or shoulder
> in the response of video sensors, which have an essentially linear
> response. And CRT displays have a nonlinear response curve that is
> different than both film and video acquisition devices. So in real
> world terms, Gamma correction is the way that we map the
> characteristics of one kind of device into another.
>
> Rather than continuing to argue around the edges of this topic, let's
> just agree that it is a complex subject with many variables, and that
> it is easy to mislead, or be misled, depending on the particular
> aspect of the subject we are discussing.
>
> This discussion began when I suggested that we would like electronic
> acquisition systems to better emulate the characteristics of films
> non-linear response curve. That is, that we need more sensitivity at
> the ends of the range than in the linear area in the center, and that
> we need the ability to encode the information that is acquired in
> such a way as to preserve as much of it as possible for post
> production processing.
>
> The application of a video gamma response in the camera encoder does
> just the opposite. We throw away much of the potentially useful
> information in the linear signal before the signal ever gets out of
> the camera. In essence this is ANOTHER legacy practice that was a
> pragmatic compromise to deal with the cost of CRT displays. Turns out
> that it is easier to do the gamma correction at the source, than it
> is to build it into a cheap CRT based TV.
>
> While researching this subject - again - I stumbled upon an excellent
> set of slides (with many illustrations and images) from the
> University of Frankfurt in Germany. Mark Schubin may be particularly
> interested in the excellent information of the Modulation Transfer
> Function. The slides begin with the response of the Human visual
> system. They also track the history of both imaging systems and
> displays, and include some useful info on the relative response of
> LCD displays versus CRT displays.
>
>
http://www.cvg.physik.uni-frankfurt.de/teaching/04ws_dvap/DVAPVideo251004fold4.pdf
>
> >The idea is exactly the same. It is true that the "display system"
> >for B&W film has a huge variability, in that there are different
> >contrast grade (print gamma) papers. Color film and print paper
> >NOMINALLY has no such range .... that's what the books and data sheets
> >say, but they are wrong ... you can play with negative color film
> >just like negative B&W if you want to be "arty". People
> >stopped doing that with the advent of Photoshop, however.
>
> All correct. I read a number of papers about gamma as it relates to
> still photography and printing.
> They all point out the DIFFERENCES between still imaging systems and
> the use of film for motion imaging systems. As I said, this is a
> complex subject with many variables; what is defined as "gamma" for
> one application may not be the same as the definition for another
> application, although the underlying issues are very similar.
>
> All of this leads us into another important discussion.
>
> The notion that a display system can, or should, be optimized for a
> single kind of source imagery is beginning to be challenged by
> modern realities. Although it has ALWAYS been true, there are
> significant differences in the luminance and color characteristics of
> various image acquisition systems for both still and motion imagery.
>
> At one time it was cost prohibitive to build display systems for
> which the response could be modified to deal with these differences,
> especially for mass produced consumer appliances. TV as we know it
> had ALL of these parameters locked down in order to keep the cost of
> the receiver as low as possible. NTSC assumes an end-to-end closed
> system.
>
> Then the PC happened - actually to me more correct, the Mac happened,
> along with the revolutions in Desktop Publising and Desktop Video.
> Apple had to deal with these differences in a very critical way.
> Their solution is ColorSync, which is built into the OS and the
> Digital media architecture, QuickTime.
>
> Every "device can be characterized with a profile that defines the
> luminance characteristics (gamma) and the colorimetry. This includes
> all types of displays, printers, and source imagery for both still
> and video formats. It is also possible to create and store custom
> profiles.
>
> If Bert had ColorSync on his new LCD panel he could create presets
> for each of the setups he has tried, then switch between them to see
> how they behave with different kinds of source imagery. He would
> probably find that there is NO single setup that is ideal for all
> applications, and that he might even change the setup to deal with
> differences in ambient lighting.
>
> IF Apple gets into the media center business, as the NYTimes article
> that Monty posted yesterday suggests, ColorSync is going to play an
> important role.
>
> Thanks for the spirited discussion Doug. I think we have both learned
> something, and in so doing, perhaps we have exposed some people on
> this list to the realities of modern imaging and display systems.
>
> Regards
> Craig
>
>
> ----------------------------------------------------------------------
> You can UNSUBSCRIBE from the OpenDTV list in two ways:
>
> - Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
>
> - By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
>
>
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts:
