Basic Guide to Color Calibration using a CMS (updated and enhanced) - Page 60

Samsung and LG, the Samsung is kind of old (2009 series). The LG (LK450) only has Color(Sat?)/Hue control.

1 more thing, when doing gamut calibration which factor is the most importance ? Sat/Hue/Luminance or DeltaE ? I always try to keep Hue and Sat as close to the target as possible then recheck DeltaE.

Luminance for Primaries and Hue for Secondaries is typically order of importance when doing gamut.

Jason

yes and in general, I'd say from most important to least important:

1. Hue (this is what separates one color from another)
2. Lightness (more obvious than saturation errors)
3. Saturation (least obvious and contrast ratio/black levels will affect perception of saturation)

Is ChromaPure much better than HCFR in terms of real-time measuring? I mean, Gamma readings/DeltaE (for each IRE) and CMS target values display...

Thanks.

Obviously it's more complicated but generally speaking this has been the 'logical' approach for many years (at least for the last 10 years that I have been watching ISF'd displays). I believe this is what is taught at the ISF classes as well, is it not?

Jason

When I took it in January that was still the basic rule for ISF. I'll look at the book to verify that.

I don't doubt your math here Tom but what I question is if the math coincides with the visual errors we would actually see first (most objectionable).

Jason

Sounds like there would still be unknown "unknowns" (how certain combinations of variables chain together and combine at certain levels affect other variables creating unknown results)
It would seem the complication of this (and calibration in general) is that some errors combine together to form different variances to the human eye in non-linear combo's so to speak.

If we are dealing with extremely visible errors, then I'm sure we can apply these more general rules since the error would be overwhelming the other smaller errors, but in other cases I cannot imagine they always hold true.

HCFR (original recipe) *can* be an effective tool when used in combination with other (excel based) tools floating around, but it has some major "faults"/limitations and in some cases might even lead a newly (or not so newly) minted calibrator to adjust things in the opposite direction of where they should go.

If all you want to do is dial in the greyscale then HCFR is sufficient, beyond that things get "interesting."

PS: I'm assuming that the new/revised fork is an improvement, but I haven't looked at it.

I believe it is MUCH more complex to describe what our eyes see given any specific combined amount of color error than what has been said so far... Not that it can be described any better, just that our current methods are far from perfect.

I'm not a Physics or Math guy, but I'm pretty sure the generalities of the Navier Stokes problem applies here (even though we are dealing with color instead of fluid and the equations themselves are not directly applicable)...

A direct reference to some of the problems in logic are contained in this PDF about physics (http://www.math.jussieu.fr/~boutet/P..._equations.pdf)

From the above source:
-------------------------------------------
The present problems are: how can one describe the phenomena with adequate equations, how can one compute them, and use and visualize
the results in spite of their complexity.

The question is the prediction, computation and representation of phenomena which contain no “mystery” in their generation but which are too complex
in their realization to be accessible to simple computations.

The equations involve some physical parameters and turn out to be relevant when these parameters have certain values. Therefore as an introduction it is natural to consider a “chain” of equations (and coderguy would add --- also to try to break everything down linearly is our human nature, but it doesn't always work out when testing the phenomena directly in the real world), so in hoping, as is often the case, that the next equation will become relevant when the structure of the phenomena becomes too complicated to be computed by the previous one.

It is an averaging process and the “turbulent model” starts to be efficient when the original
Navier Stokes are out of reach by direct numerical simulations.
In this averaging appears a hierarchy of moments which has been studied “per se” (cf. section
(3) and [24]). However this is not sufficient for the following reasons: (and goes on and on)....

-------------------------------------------

This problem is often cited in mathematics to show there can be limitations in the accuracy of measuring interpreted phenomena (like our eyesight or even the weather for instance), these limitations are in chained theories and equations as it relates to direct physical results (even a few variables could produce millions of varying results due to varying values within each of those variables), and the main limitation is that we cannot achieve certainty even if we attempt to solve all the resulting hierarchies of equations without actually being able to test each pattern against physical phenomena (in this case what our eyes interpret is the untested physical phenomena). That is why weather models get more accurate over time (they are using actual tested phenomena from historical data). In the color situation, many of this is untested because there are too many potential subjective results in each different chain in the end equation. So even if you think an equation or theory is perfectly sound mathematically in a linear fashion, the point is basically that with physical phenomena you potentially left it untested in thousands of other scenarios.

So to truly solve ALL of the color questions on what the EYE is seeing in any given COLOR ERROR situation, we would have to produce equations of nearly infinite magnitudes at varying degrees of percentage estimations in varying situations. However, the good news is we can still achieve some satisfying result without having to be this perfect, but we are really doing nothing more than using a system of predictions. You can be completely wrong some of the the time, or be ONLY partially wrong at varying levels most of the time.

Sorry for the RANT, and I have no opinion as to the exact amount this principle pollutes the calibration issue, but I am just saying the problem is there. I am not sure why I wrote all this, I guess I like throwing wrenches into things...

So the keywords in the above quotes are "hoping" and "averaging", and I guess the best we can really do in calibration is to average and hope...

coderguy, you miss the fundamental concept of color science, it is already based on an approximation. All of the coordinates, equations, etc. are derived from an average response (the "standard observer"). Any individual person may or may not be close to that average but it is the best that can be achieved given the complexities you describe. There is no attempt in color science to derive the actual physical mechanism of an individual's eye response. As far as describing the sensitivities to various errors in hue, saturation, and lightness it is amazingly accurate nontheless.

I am not arguing the entirety of color science or calibration. We were discussing how small color errors are perceived and describing that as such in relation to the equations. I am not questioning the ability to try to approximate the calibration using only mathematics to as good as we can get it, I am questioning the ability to describe the error as perceived by eye.
I agree the point of color science is not to perfect this or even really solve this problem, but this was the topic of one of the posters.

Sensitivities in what situation, you're quoting a problem that would need multivariate analysis but only using one to three variables. In dealing with extremes, I would agree with you, but in dealing with combinations of smaller errors, it is nearly impossible to test the theory. There are probably somewhere around 200 variables.

What viewing content, is the content they are usually watching shot under D65 lighting (if they watch soaps it might be), what is the average content the viewer watches (blurays), on what person's screen, in what lighting conditions, what age is the viewer, what is the person's perception of color in the first place, is their source video causing any intermittent results that are immeasurable by equipment but at times visible by eye, under what image noise levels, with what compression, with what artifacts, I will stop there...

All of what you are describing is irrelevant to calibration which does not attempt to tell you an individual's sensitivity to any particular color/lightness shift in real content. Calibration tells you how close a test pattern measures to the perceptual benchmark of a standard observer, period. The formulas used to quantify "close" have worked very well for 81 years now and have greatly improved the accuracy of film and video reproduction.

Well, we were discussing the theory of what the eye can see rather than just calibrating it purely based on a mathematical assumption of one result (it is a 3-dimensional problem).

I trust math and science more than I do my eyes.

i had a 10th grade color-blind chemistry teacher who always used to say, "That blackboard isn't green, it's brown, dammit!". He used to always wear green suits because he preferred that shade of "brown".

This is an extreme example, and the teacher was making fun of his colorblindness, obviously, but even if your eyes are reasonably close to accurate, if you see something wrong or out of whack with a calibrated image, the math or measurement (or measuring device) is wrong.

Yes, I agree to a point, but we are usually left with an error of some sorts and must make decisions and concede something. Also, you just can't measure certain things with calibration equipment (like posterization for instance, at least not that I know of).

BTW, the entire set of equations are still not completely solved, and the Clay Institute will pay someone a million if they solve it (I got no chance)...

Somewhat surprisingly, given their wide range of practical uses, mathematicians have not yet proven that in three dimensions solutions always exist (existence), or that if they do exist, then they do not contain any singularity (smoothness).

Kurt Gödel strikes again....

Sorry I usually don't rant on theory that much, to me most of those mathematical concepts are too far out and I would just lose my sanity if I ever tried to solve any of them (not that I could). Thank god I didn't go into mathematics theory.

...

Up late and not much else to do but POST, and even though you didn't address me --- I doubt anyone would argue about the purity of the math (or against trying to hit Target 0), but maybe just the non-linear effect to the eyes at varying levels of H-S-L as the values stray (or in combined with other factors). We are dealing with a three-dimensional problem that is based on the perception of a living mechanism (being our eyes). I didn't mean to scare anyone off, I really just posted that because the "quotes of logic" from that math problem were really interesting as related to this issue (at least I thought so), the math itself is too intense, but the problem is interesting to view in a sense of logic.

I mean the logic issues in the Navier Stokes problem applies pretty much universally to measuring something like this as in measuring an active phenomenon perfectly with formulas, and even the best mathematicians in the world haven't fully solved the issue. They are still trying to work out many of the equations. If you can finish the problem they put forth which is different but similar logic issues, they really will pay you 1 million dollars (The Clayton Institute).

If some color scientists came up with that pre-determination even when dealing with SMALL ERRORS, then I'm sure it's fairly applicable, but it appears no-one really exactly agrees when there are small errors the true effect to the human eye in most viewing (or do they), and that is a pretty complex thing to determine even for a panel of scientists. I mean it's obvious if you change any value too extreme that it will stick out to our eyes, but beyond that it would take an art-form or just incredibly complex math to determine some average of what we are seeing across varying content.

That is not to say anything you put forth was incorrect, but it's just to say that it probably has limitations (much like limitations in predicting the weather).

I could be wrong, but that is my opinion anyways.

Hehe, no beef with anyone, just curious why it is that it has evidently been taught wrong for so many years by industry experts if "The numbers are fairly conclusive on this subject".

The deeper into calibration the more realization that there are way too many varied opinions for this to be a 'known' science, the math is done but how do we know the formulas are correct. As I said in another thread, we need standards for our standards. Even ISF and THX appear to teach different procedures, who's correct or does anyone even know for sure. It appears there is an abundance of bickering/nit-picking based upon varied opinions on aspects that are not completely understood.

The bothersome aspect here, if true, is the probability of Pro's out there being paid to calibrate 'wrong' based upon what they have been told/taught...?

Jason

Maybe I've been watching a bit too much SCI FI lately. At least this is based in real logic and not FRINGE science like that show Fringe (btw some ok acting in that show, but the science behind it is NOT even Fringe science, more like optimistic witchcraft).

Anyways as a Disclaimer, I am not even close to a mathematician, but as a programmer I like to study problems of logic and certainty (or in this case uncertainty).


People are paid to predict the weather incorrectly every day, the main thing is just that they get close most of the time, that's IMHO anyways.

Hehe, oh I give you major credit for jumping in head first at the shallow end, brave man, very brave (and much deeper than I have any desire to explore)!

I have been doing this stuff for years, even remember when some of the current experts were on here asking questions that I knew the answer, but as someone that is simply trying to do things as accurately as possible it's frustrating that so many experts have so many varied opinions (this makes it rough on those of us that are mere enthusiasts/hobbyists and don't have a dog in the fight).

Jason

I gotcha...

I have no reservations about being certain about the unfinished math of representing the measuring of phenomena (especially in 3D space) with equations in a perfectly precise and accurate manner (especially measuring a living response), and anyone can go read a thousand books on that one, as it is one of the unsolved mysteries of the world. So no worries there.

The degree of effect it has on this I have no idea, but it obviously affects it.