I have been trying to understand gamma and gamma correction. Some previous threads here (especially in CRT Projector forum) are interesting, but don't seem relevant to the basic notion.

As I understand it, tube televisions initially proved incapable of producing light intensity that was proportional to the broadcast signal. As a result, broadcasters built in compensation (I gather through the gamma encoding process) for this non-linear relationship by selectively boosting the signal. Current digital monitors, however, are capable of a linear response so must account for this historical practice. They have built-in gamma compensation that attempts to replicate the performance of earlier tube technology. Ideally, the gamma curve defining the monitor's response exactly matches the inverse gamma compensation curve in the source material. The result is the desired linear response. The problem with some monitors is that they do not have an accurate gamma so the image is biased in one or more ways (usually obscuring shadow detail?).

Am I even remotely close to understanding this?

Thanks.

Greetings

More or less it.

Regards

Thanks, again, for your helpful response.

Jim

Just to add a correction, LCDs do not have a linear response— they have an inherent s-curve gamma which has to be corrected for with internal processing.

http://i32.tinypic.com/107whnt.png
source (http://www.sony.co.uk/res/attachment/file/78/1133797573278.pdf)

Thank you for the clarification. This would also be the case with first cousin LCoS? I assume, then, that plasmas do have a linear response. Is that correct? Is this also true of DLP?

In any event, the main point was to be clear that gamma describes the compensation for difference between original CRT technology and current digital technology (essentially LCD (or LCoS), DLP, and plasma).

I have been trying to understand gamma and gamma correction. Some previous threads here (especially in CRT Projector forum) are interesting, but don't seem relevant to the basic notion.

As I understand it, tube televisions initially proved incapable of producing light intensity that was proportional to the broadcast signal. As a result, broadcasters built in compensation (I gather through the gamma encoding process) for this non-linear relationship by selectively boosting the signal.

Partly, although human vision also plays apart, because we also see nonlinearly. So gamma is still very much beneficial in the digital domain and with linear displays, regardless of legacy CRT reasons.

Current digital monitors, however, are capable of a linear response so must account for this historical practice. They have built-in gamma compensation that attempts to replicate the performance of earlier tube technology.


Yes, though again, it's not just the nonlinear nature of CRT. Otherwise we'd have linear digital content and reverse gamma LUT just for CRTs, but we don't, because there is still huge benefit from nonlinear coding. Linear coding is extremely wasteful, because we do not see linearly.

Ideally, the gamma curve defining the monitor's response exactly matches the inverse gamma compensation curve in the source material. The result is the desired linear response.

Almost. It's actually not totally linear through the system for subjective viewing reasons because we view content on dim displays in dim surround situations that don't match the original luminances of the original scene. The natural CRT gamma of about 2.5 is actually undercompensated, which leaves and end-to-end flow of about 1.25, rather than 1.

The problem with some monitors is that they do not have an accurate gamma so the image is biased in one or more ways (usually obscuring shadow detail?).

Am I even remotely close to understanding this?

Thanks.

The basic essence yes.

See also:
http://www.poynton.com/GammaFAQ.html

In any event, the main point was to be clear that gamma describes the compensation for difference between original CRT technology and current digital technology (essentially LCD (or LCoS), DLP, and plasma).

No, this is farther away than your original post. It is irrespective of the difference, because gamma existed way before digital displays came into play, and still exists today for a variety of reasons that don't have to do with this difference. The reason that de-gamma processing *has* to occur in a linear digital display (unlike a CRT which you can think of as doing it inherently) is because there is a difference. However, nonlinear coding is extremely advantageous both for signal-to-noise ratio in analog circuits, and for efficient digital coding because of the fact that humans see nonlinearly. That this nonlinear vision also basically matches the nonlinear nature of a CRT made this an easy and fantastic coincidence at the advent of video. Now with linear displays, we still maintain gamma because of the benefits from a coding efficiency standpoint, despite the added difficulty of needing to add additional de-gamma processing to linear digital displays.

Thanks. This is much more complicated than I thought, but you have helped clarify things well enough for my feeble mind to handle.

Edited addition: OK, I looked over the Sony document and the Poynton FAQ. Very helpful. My head still hurts, though!

I have been trying to understand gamma and gamma correction. Some previous threads here (especially in CRT Projector forum) are interesting, but don't seem relevant to the basic notion.

As I understand it, tube televisions initially proved incapable of producing light intensity that was proportional to the broadcast signal. As a result, broadcasters built in compensation (I gather through the gamma encoding process) for this non-linear relationship by selectively boosting the signal. Current digital monitors, however, are capable of a linear response so must account for this historical practice. They have built-in gamma compensation that attempts to replicate the performance of earlier tube technology. Ideally, the gamma curve defining the monitor's response exactly matches the inverse gamma compensation curve in the source material. The result is the desired linear response. The problem with some monitors is that they do not have an accurate gamma so the image is biased in one or more ways (usually obscuring shadow detail?).

Am I even remotely close to understanding this?

Thanks.


That's basically right. Some software permits viewing the gamma curve (sometimes called luminance histogram or luminance graph) for R, G, & B separately so you can gauge how accurately each color follows an appropriate gamma curve. Errors in the gamma curve show up as errors elsewhere also... RGB tracking will reflect the Gamma errors, so will color temperature errors expressed in dE - because when you see errors in Gamma (especially when you are looking at separate curves for R, G, & B), they "must" appear elsewhere in the measurements.

Gamma is not some standalone factor... everything is interrelated.

The natural CRT gamma of about 2.5 is actually undercompensated, which leaves and end-to-end flow of about 1.25, rather than 1.Sony, Microsoft and Hewlett-Packard proved that is not true in the real world.

http://download.microsoft.com/download/1/6/1/161ba512-40e2-4cc9-843a-923143f3456c/MSColorCaseStudy.doc
Sony provided studies that proved optimally calibrated CRTs in their native state are 2.2 gamma and have HDTV primaries. [Katoh and Deguchi, "Reconsideration of CRT Monitor Characteristics]

http://www.imaging.org/store/epub.cfm?abstrid=2202
Society for Imaging Science and Technology
Naoya Katoh and Tatsuya Deguchi (Research Center, Sony Corporation)
Historically, CRT characteristics were investigated in the 1910’s by Child and Langmuir and later in the 1950’s by Oliver, but not much research has been done until recently, as of colorimetric characterization. In this paper, four basic characteristics of the CRT monitor are reconsidered, i.e.,
1) Tone Curve Characteristics
2) Phosphor and Additive Color Mixture
3) Gamut
4) Viewing Flare

FWIW jimmacintosh "gamma correction" is really an analog non linear voltage convention and most of the time is implemented by sampling a pre distorted analog waveform (emphasis) and regenerated via a given LUT (de-emphasis). The tonal shift towards the dark end of the intensity scale is almost always done in the analog domain. For example the link below reveals the actual distribution of digital intensity.

http://www.displaymate.com/ShootOut_Part_3.htm
Digital Granularity
"This is the most efficient method for specifying the intensities accurately, but it’s not the method that’s used because a linear spacing is more convenient and makes signal processing a lot easier. In reality, the digital intensity steps are all separated by equal differences rather than equal ratios.

See Figure 1 in the link above, plotted there are the percentage change in brightness per digital step from an 8-bit intensity scale”

This has been the topic of heated debate in the past. Please leave it in the past. Move forward in a civil matter. Lets keep the moderators from having to get involved. Thank you!!