AVS Forum banner
1 - 20 of 23 Posts

·
Guest
Joined
·
867 Posts
Discussion Starter · #1 ·
Why do calibrators promote such accuracy to d65 greyscale color temperature?


The greyscale color temperature for commercial cinema projectors seems to be alot more lax. For moive theaters Smpte 196M white point is 5400 +600/-200 for a range of 5200 to 6000. Measurements taken from movie theaters gave a even wider range below 4700 to above 6800. Even Hollywood screening rooms and labs had a range of below 4800 to above 6700. I would expect audiences were not walking out of the theaters in disgust, and the Hollywood screening rooms and labs where being used for critical viewing.


Apply the same lax standard +600/-200 to D65 would give 6300 to 7100 as being acceptable. Would there be significant loss of picture quality using 6300 or 7100 in comparison to the accurate D65? Since the eyes are apparantly better at adapting toward daylight color temperatures maybe a region of D5400 average summer sunlight at noon to D7100 light summer shade, would be more tolerable.


At what point would a display become noticeably wrong to you just by looking at it? At what point would it be intolerably inaccurate to you?


How much does viewing environment impact on dislay color perception? Vision attempts to maintain color consitancy so if the viewing environment is not D65 neutral the viewer is going to perceive the colors on the display differently. Vision is also chromatic adaptive so if the source of light is not D65, viewers are going to perceive colors differently. Would a batcave home cinema make correct greyscale color calibration more critical or the viewer more adaptive - tolerant?
 

·
Registered
Joined
·
6,723 Posts
Greetings


That's why we have dE stuff.


10 = Tolerable

5 = Very good

3-4 or less = beyond perception


And it isn't D this and D that .. D D D ...


There are A points .. B points .. C points ... D points ... E points ...


Like how the E point can be used for B/W films from the old days. 0.333, 0.333 ...


regards
 

·
Guest
Joined
·
867 Posts
Discussion Starter · #3 ·

Quote:
Originally Posted by Michael TLV /forum/post/17044487


Greetings


That's why we have dE stuff.


10 = Tolerable

5 = Very good

3-4 or less = beyond perception


And it isn't D this and D that .. D D D ...


There are A points .. B points .. C points ... D points ... E points ...


Like how the E point can be used for B/W films from the old days. 0.333, 0.333 ...


regards


How does dE translate to color temperature. 10, 5, 3-4, would translate to what temperature ranges. Most reviewers, give a chart showing color temperature tracked across the greyscale.


I appreciate that D65 refers to green in addition to red and blue, while color temperature just covers red-blue. So I will expand my original question to cover how much must green be out for you to be able to notice it by eye or find it intolerably inaccurate to watch. Since the eyes red-green hue discrimination is better than red-blue I would expect it become more noticeable quicker, especially since it would effect skin tones.
 

·
Registered
Joined
·
6,723 Posts
Greetings


If you dig up the formula used to calculate dE ... you get your answer. The formula looks at x,y coordinates versus 313,329 ... and also accounts for the way we see color.


It should be noted that color temp is irrelevant. I can give you 6500 that is purple ... or green. both are 6500 ... 6500k is a line on a graph. ... A LINE ... not one point on the line. There are then infinite points on this line that stretches infinitely. Because of this, I can give you something that is 10000K and it can be easily more accurate than some 6500 point. (Basic calibration theory ... x,y coordinates ... not some color temp number.)


regards
 

·
Registered
Joined
·
7,998 Posts
SMPTE has established standards for DCI installations. It is +- 4.0 CIELAB dE.


So you need to know both the dE value and the color space in which it is calculated. CIELUV is a little different, and CIE94 is different for color than it is for white. CIE94 white is the same as CIELAB, but for color the tolerances are much smaller.


Because of what you point out about color temp ignoring green, it is a metric of limited value. It is a legacy holdover from before the time when dE was widely used that I wish would just go away. (The video industry is incredibly conservative. dE was endorsed by CIE in 1976, but it has only been widely used in the video industry in the last ten years.)
 

·
Registered
Joined
·
7,998 Posts

Quote:
Originally Posted by Michael TLV /forum/post/17044744


If you dig up the formula used to calculate dE ... you get your answer. The formula looks at x,y coordinates versus 313,329 ... and also accounts for the way we see color.

Actually, the dE formulas do not use xyY at all because they are not sufficiently perceptually uniform. The 1976 dE formulas use either Luv or Lab. All subsequent formulas use Lab only. One may use xyY in calculations, but it has to be converted to Lab or Luv first before the actual dE formula is applied.
 

·
Registered
Joined
·
6,723 Posts
Greetings


And u is derived from the graph and v is as well. I've seen the formulas.


regards
 

·
Registered
Joined
·
7,998 Posts
Michael, not sure exactly what you mean. u and v are part of the Luv color space. u' and v' (not the same) are intermediate steps in a XYZ-Luv conversion. They are used as chromaticity coordinates in the 1976 CIE chart in the same way xy are used in the 1931 chart.
 

·
Registered
Joined
·
6,723 Posts
Got it ... my bad ... they were u' stuff


regards
 

·
Registered
Joined
·
3,871 Posts

Quote:
Originally Posted by dovercat /forum/post/17044454


Why do calibrators promote such accuracy to d65 greyscale color temperature?


The greyscale color temperature for commercial cinema projectors seems to be alot more lax. For moive theaters Smpte 196M white point is 5400 +600/-200 for a range of 5200 to 6000. Measurements taken from movie theaters gave a even wider range below 4700 to above 6800. Even Hollywood screening rooms and labs had a range of below 4800 to above 6700. I would expect audiences were not walking out of the theaters in disgust, and the Hollywood screening rooms and labs where being used for critical viewing.


Apply the same lax standard +600/-200 to D65 would give 6300 to 7100 as being acceptable. Would there be significant loss of picture quality using 6300 or 7100 in comparison to the accurate D65? Since the eyes are apparantly better at adapting toward daylight color temperatures maybe a region of D5400 average summer sunlight at noon to D7100 light summer shade, would be more tolerable.


At what point would a display become noticeably wrong to you just by looking at it? At what point would it be intolerably inaccurate to you?


How much does viewing environment impact on dislay color perception? Vision attempts to maintain color consitancy so if the viewing environment is not D65 neutral the viewer is going to perceive the colors on the display differently. Vision is also chromatic adaptive so if the source of light is not D65, viewers are going to perceive colors differently. Would a batcave home cinema make correct greyscale color calibration more critical or the viewer more adaptive - tolerant?
"SMPTE 431-2, “Reference Projector and Environment

for Display of DCDM in Review Rooms and Theaters,” is

a recommended practice that defines the characteristics

of the reference projector and its controlled environment.

It also defines the acceptable tolerances around critical

image parameters for both review room and theater

applications. The goal of 431-2 is to provide a means for

achieving consistent and repeatable color image quality

for all digitally projected images."



"SMPTE 431-2 states: “The Reference Projector is a

practical device.” There is no physical reference projector

in an institution in the sense that there is a primary

standard against which all other devices or materials

are compared. It is instead a device defined by its

characteristics and capabilities, not by its technology. It

is doubtful that there is any physical device that exactly

matches the defined characteristics of the reference

projector. However, there are many physical digital

projectors that match the defined characteristics of the

reference projector within acceptable tolerances of the

reference characteristics. Therefore, the term “reference

projector” refers to two different devices: (1) the theoretical

device defined exactly by the characteristics given in the

various standards, and (2) any physical device that meets

the characteristics, within the specified tolerances, defined

for the theoretical reference projector."


SMPTE recommended tolerances for white in digital cinema projection:

For Reviewing Room- +0.002 x/y

Commercial Theater- +0.006 x/y


see: SMPTE Engineering Guideline #432-1, and 'Color Processing for Digital Cinema 4: Measurements and Tolerances' by Thomas O. Maier,SMPTE Journal Nov/Dec 2007.


One must always keep in mind that film-is-not-digital cinema-is-not-rec709 HD video-is-not-film, etc. Each imaging format is unique and must be understood on its own terms and in its own context. What is tolerated in one format should not be construed to serve as a guide for another.


Display calibrators work towards a standard reference target value and strive to remain within recommended tolerances, as much as a given device functionality and practical time allow. We work from a basis of what is our achievable best, in pursuit of a reference target, NOT how much compromise any individual viewer won't notice. Best practices preserve image fidelity and picture quality. Probing for the outer limits of tolerable compromise seems a fool's errand to me. What may or may not be noticed as "wrong" by any individual viewer will vary widely.


Best regards and beautiful pictures,

G. Alan Brown, President

CinemaQuest, Inc.

A Lion AV Consultants Affiliate


"Advancing the art and science of electronic imaging"
 

·
Guest
Joined
·
867 Posts
Discussion Starter · #11 ·

Quote:
Originally Posted by Michael TLV /forum/post/17044744


Greetings


If you dig up the formula used to calculate dE ... you get your answer. The formula looks at x,y coordinates versus 313,329 ... and also accounts for the way we see color.


It should be noted that color temp is irrelevant. I can give you 6500 that is purple ... or green. both are 6500 ... 6500k is a line on a graph. ... A LINE ... not one point on the line. There are then infinite points on this line that stretches infinitely. Because of this, I can give you something that is 10000K and it can be easily more accurate than some 6500 point. (Basic calibration theory ... x,y coordinates ... not some color temp number.)


regards

To lazy to do the math, so downloaded ColorHCFR. It gives the following figures.

X .3001 Y 0.329 dE10 Tolerable 6953

X .3064 Y 0.329 dE5 Very Good 6732

X .3077 Y 0.329 dE4 Beyond Perception 6686

X .313 Y 0.329 dE0.2 6500

X .3178 Y 0.329 dE4 Beyond Perception 6333

X .3191 Y 0.329 dE5 Very Good 6288

X .3255 Y 0.329 dE10 Tolerable 6067


So thats 6067 to 6953 tolerable, a range of about 900, so actually bigger than the smpte -200/+600 range. 6288 to 6732 very good, a range of about 500. 6333 to 6686 beyon perception, a range of about 350.


How far out does it need to be before you can see it by eye. How far out does it need to be before you find it unbearable to watch. These figues would give the answer 6953 is unbearable and 6732 is very noticeable.

These seem to me to be centered around 6500 when I would of expected more leeway to red than to blue, if it was based on human perception. The noticeable and tolerable ranges also seem very small given human color consistance and chromatic adaption. In practice are my assumptions incorrect, are these correct.
 

·
Guest
Joined
·
867 Posts
Discussion Starter · #12 ·

Quote:
Originally Posted by GeorgeAB /forum/post/17045128


One must always keep in mind that film-is-not-digital cinema-is-not-rec709 HD video-is-not-film, etc. Each imaging format is unique and must be understood on its own terms and in its own context. What is tolerated in one format should not be construed to serve as a guide for another.


Display calibrators work towards a standard reference target value and strive to remain within recommended tolerances, as much as a given device functionality and practical time allow. We work from a basis of what is our achievable best, in pursuit of a reference target, NOT how much compromise any individual viewer won't notice. Best practices preserve image fidelity and picture quality. Probing for the outer limits of tolerable compromise seems a fool's errand to me. What may or may not be noticed as "wrong" by any individual viewer will vary widely.


Best regards and beautiful pictures,

G. Alan Brown, President

CinemaQuest, Inc.

A Lion AV Consultants Affiliate


"Advancing the art and science of electronic imaging"

Striving for perfection beyond the point we can perceive the difference seems a bit pointless. Consumer displays are not usually reference quality so you are unlikely to find perfection at a price most are willing to pay.

Having a better idea of how bad they need to be before they start to detract from the viewing experience, seems like something that would be nice to know. It also gives a more realistic goal for buying a display rather than just looking for D65 perfection that you are unlikely to get. How inaccurate do you need color temperature to be, before you go that display is rubbish despite its superior contrast, etc... Choosing a display is an excercise in compromise for those who do not have infinite money.
 

·
Registered
Joined
·
7,998 Posts

Quote:
Originally Posted by GeorgeAB /forum/post/17045128

"SMPTE 431-2, Reference Projector and Environment

for Display of DCDM in Review Rooms and Theaters, is

a recommended practice that defines the characteristics

of the reference projector and its controlled environment.

It also defines the acceptable tolerances around critical

image parameters for both review room and theater

applications. The goal of 431-2 is to provide a means for

achieving consistent and repeatable color image quality

for all digitally projected images."



"SMPTE 431-2 states: The Reference Projector is a

practical device. There is no physical reference projector

in an institution in the sense that there is a primary

standard against which all other devices or materials

are compared. It is instead a device defined by its

characteristics and capabilities, not by its technology. It

is doubtful that there is any physical device that exactly

matches the defined characteristics of the reference

projector. However, there are many physical digital

projectors that match the defined characteristics of the

reference projector within acceptable tolerances of the

reference characteristics. Therefore, the term reference

projector refers to two different devices: (1) the theoretical

device defined exactly by the characteristics given in the

various standards, and (2) any physical device that meets

the characteristics, within the specified tolerances, defined

for the theoretical reference projector."


SMPTE recommended tolerances for white in digital cinema projection:

For Reviewing Room- +0.002 x/y

Commercial Theater- +0.006 x/y


see: SMPTE Engineering Guideline #432-1, and 'Color Processing for Digital Cinema 4: Measurements and Tolerances' by Thomas O. Maier,SMPTE Journal Nov/Dec 2007.


One must always keep in mind that film-is-not-digital cinema-is-not-rec709 HD video-is-not-film, etc. Each imaging format is unique and must be understood on its own terms and in its own context. What is tolerated in one format should not be construed to serve as a guide for another.


Display calibrators work towards a standard reference target value and strive to remain within recommended tolerances, as much as a given device functionality and practical time allow. We work from a basis of what is our achievable best, in pursuit of a reference target, NOT how much compromise any individual viewer won't notice. Best practices preserve image fidelity and picture quality. Probing for the outer limits of tolerable compromise seems a fool's errand to me. What may or may not be noticed as "wrong" by any individual viewer will vary widely.

I was referring to the following in the same Maier document you cite:


"Specification of the Color Accuracy of Displayed Colors


Within the minimum color gamut specified for the reference projector, all colors need to be accurately reproduced within a tolerance of 4 delta E*ab."


However, on second reading it appears that this standard applies to color, rather than to white. If one takes the theater +-0.006xy as the recommended tolerance for white, then that provides an average dE of 5.5, or a little higher tolerance than what is recommended for color. This would be quite odd since errors in the white point are generally more perceptually offensive than similar errors in the gamut.


It's also interesting that whereas SMPTE lists separate tolerances for white chromaticity for review rooms and reference rooms, they do not make this distinction for color. Thus, I think that the tolerance for a reference environment +-0.002xy is probably a more reasonable choice if we are looking for recommended guidelines for white point error. That translates to an average of 4.1 in CIELAB--almost exactly the same as for color.


Finally, I have read your other comments several times, and I can't quite make out what you are trying to say. I don't think that calibrators generally allow larger than necessary errors than they have to simply because they are within some established tolerance. We get it as good as we can within the limits of the equipment, environment, and time available. The importance of tolerances is to provide some guidelines in reporting that indicate whether the calibration has resulted in display performance that is within industry standards.


As for the differences between video, film, and various gamuts, since neither CIE nor ISF has ever published ANY official tolerances that I know of, then the DCI standard--which at least has the advantage of being recent--is as good a place to hang one's hat as any.
 

·
Registered
Joined
·
7,056 Posts

Quote:
Originally Posted by dovercat /forum/post/17045330


Striving for perfection beyond the point we can perceive the difference seems a bit pointless. Consumer displays are not usually reference quality so you are unlikely to find perfection at a price most are willing to pay.

Having a better idea of how bad they need to be before they start to detract from the viewing experience, seems like something that would be nice to know. It also gives a more realistic goal for buying a display rather than just looking for D65 perfection that you are unlikely to get. How inaccurate do you need color temperature to be, before you go that display is rubbish despite its superior contrast, etc... Choosing a display is an excercise in compromise for those who do not have infinite money.

yeah but I've never touched a display that could hit D65 pretty spot on from cuts to 100%.


I guess that's the reason people get anal about it, is because it's achieveable on almost everyt set if you go to the service menu.
 

·
Registered
Joined
·
3,871 Posts

Quote:
Originally Posted by dovercat /forum/post/17045330


Striving for perfection beyond the point we can perceive the difference seems a bit pointless. Consumer displays are not usually reference quality so you are unlikely to find perfection at a price most are willing to pay.

Having a better idea of how bad they need to be before they start to detract from the viewing experience, seems like something that would be nice to know. It also gives a more realistic goal for buying a display rather than just looking for D65 perfection that you are unlikely to get. How inaccurate do you need color temperature to be, before you go that display is rubbish despite its superior contrast, etc... Choosing a display is an excercise in compromise for those who do not have infinite money.

Context explains most points of view, especially when the context is revealed and understood. It's been my experience that most video consumers will tolerate much more that's wrong in an image than any of the standards and recommendations mentioned thus far. Display manufacturers seem to count on that.


Since this is the display calibration section of the forum, I was thinking more in that context. I've also observed that no one has "infinite money" (except governments that can print their own) and the wealthy may or may not regard the higher cost and performance of a superior video display to be justified for them. A disciplined shopper will determine what they can afford and find a solution that gives them the most quality for that amount. You are right about compromises. It certainly can help to understand recommended tolerances.
 

·
Banned
Joined
·
889 Posts

Quote:
Originally Posted by GeorgeAB /forum/post/17045487


Context explains most points of view, especially when the context is revealed and understood. It's been my experience that most video consumers will tolerate much more that's wrong in an image than any of the standards and recommendations mentioned thus far. Display manufacturers seem to count on that.


Since this is the display calibration section of the forum, I was thinking more in that context. I've also observed that no one has "infinite money" (except governments that can print their own) and the wealthy may or may not regard the higher cost and performance of a superior video display to be justified for them. A disciplined shopper will determine what they can afford and find a solution that gives them the most quality for that amount. You are right about compromises. It certainly can help to understand recommended tolerances.

Heck, I'm as poor as dirt and I think I would appreciate a calibrated display(CRT, LCD, DLP - 500 to 1080 lines, whatever it's capable of) more than most wealthy people who just throw a display up on a wall(or hang it from a ceiling if it's a pj), and the only settings they touch are ON, OFF, Volume Up/Down, and Channel Up/Down. LOL!


What don't people get when I try to explain to them that video calibration is akin to making sure a foot = twelve - not 11, not 14 - but TWELVE inches from construction site to construction site. They just can't seem to wrap their heads around that in video(or audio for that matter) terms. To them, if the Sub or Surrounds aren't turned all the way up, they're not getting what they paid for.
 

·
Guest
Joined
·
867 Posts
Discussion Starter · #17 ·

Quote:
Originally Posted by sotti /forum/post/17045414


yeah but I've never touched a display that could hit D65 pretty spot on from cuts to 100%.


I guess that's the reason people get anal about it, is because it's achieveable on almost everyt set if you go to the service menu.

Your first statement seems to be contradicted by your second statement?

Assuming you mean it is acheiveable on almost all displays.


Looking at a recent issue of the UK Home Cinema Choice Magazine, this seems to be untrue, or maybe they are not very good at calibration. Even if it is achievable on the display it gives a better idea of if you are going to have to sacrifice contrast for color temperature to get an exceptable picture quality.

From a recent issue of Home Cinema Choice Magazine.

The Mitsubishi HC5500 LCD projector measured 6573 out of the box, and 6500 calibrated so pretty much perfect.

The Samsung LE40B650 LCD flat panel measured 9101 out of the box, and 6522 calibrated. So it would definitely need calibration, but that better greyscale color temperature comes at quite a cost to contrast.

The Toshiba 42ZV555D LCD flat panel measured 9730 out of the box, and 7326 calibrated. So it would qualify as intolerably bad. But gets described as a real cracker, sublime, gorgeous. So having an idea what the hopefully objective measurements mean maybe useful.
 

·
Registered
Joined
·
7,056 Posts

Quote:
Originally Posted by dovercat /forum/post/17045660


Your first statement seems to be contradicted by your second statement?

Assuming you mean it is acheiveable on almost all displays.


Looking at a recent issue of the UK Home Cinema Choice Magazine, this seems to be untrue, or maybe they are not very good at calibration. Even if it is achievable on the display it gives a better idea of if you are going to have to sacrifice contrast for color temperature to get an exceptable picture quality.

From a recent issue of Home Cinema Choice Magazine.

The Mitsubishi HC5500 LCD projector measured 6573 out of the box, and 6500 calibrated so pretty much perfect.

The Samsung LE40B650 LCD flat panel measured 9101 out of the box, and 6522 calibrated. So it would definitely need calibration, but that better greyscale color temperature comes at quite a cost to contrast.

The Toshiba 42ZV555D LCD flat panel measured 9730 out of the box, and 7326 calibrated. So it would qualify as intolerably bad. But gets described as a real cracker, sublime, gorgeous. So having an idea what the hopefully objective measurements mean maybe useful.

I have a toshiba 42rv530, I"m guessing they didn't go into the service menu or did something else very wrong.


My Tosh is right at 6500 +- 100 or so from 30+
 

·
Premium Member
Joined
·
1,093 Posts

Quote:
Originally Posted by dovercat /forum/post/17045279


To lazy to do the math, so downloaded ColorHCFR. It gives the following figures.

X .3001 Y 0.329 dE10 Tolerable 6953

X .3064 Y 0.329 dE5 Very Good 6732

X .3077 Y 0.329 dE4 Beyond Perception 6686

X .313 Y 0.329 dE0.2 6500

X .3178 Y 0.329 dE4 Beyond Perception 6333

X .3191 Y 0.329 dE5 Very Good 6288

X .3255 Y 0.329 dE10 Tolerable 6067


So thats 6067 to 6953 tolerable, a range of about 900, so actually bigger than the smpte -200/+600 range. 6288 to 6732 very good, a range of about 500. 6333 to 6686 beyon perception, a range of about 350.


How far out does it need to be before you can see it by eye. How far out does it need to be before you find it unbearable to watch. These figues would give the answer 6953 is unbearable and 6732 is very noticeable.

These seem to me to be centered around 6500 when I would of expected more leeway to red than to blue, if it was based on human perception. The noticeable and tolerable ranges also seem very small given human color consistance and chromatic adaption. In practice are my assumptions incorrect, are these correct.

Can you point me to the specific place in ColorHCFR where these figures are given?


Thank you,

Garry
 

·
Registered
Joined
·
7,998 Posts
At the risk of sounding like a broken record and just repeating myself, 4.0 dE in Lab or 5.0 dE in Luv are reasonable tolerances for white point error.


Color temperature is such a flawed metric, quoting it just doesn't mean very much. A given color temp can be a fair low dE or a fairly high one depending on the RGB mix.


Reviewers continue to cite color temp only because--as I mentioned--it is an incredibly conservative industry.
 
1 - 20 of 23 Posts
Top