When dividing the screen luminance measured for full-on by the screen luminance measured for full-off are people aware that the two levels, one generally within a decade of 10 fL and one generally within a decade of 0.001 fL, are in two different regions of human vision requiring two different kinds of metering? The regions are called the photopic and the scotopic. See the attached curves (from The Science of Color).

ATTACHMENT DISAPPEARED, SEE NEXT POST.

Red light counts very much less in the scotopic. A projector which doesn't do a good job of dousing its red band in the "off" state can still produce very low "off" luminance. It shouldn't be penalized by measuring the "off" signal with a photopic lightmeter. Likewise a projector which doesn't do a good job of dousing its blue band doesn't produce low "off" luminance, and shouldn't be credited with low "off" luminance by measuring the "off" signal with a photopic lightmeter. Considering that the projector's three bands might be centered around for, example, 470 nm, 540 nm, and 630 nm, compare the scotopic and photopic luminosity curves to see what huge differences occur.

People using an illuminance (lux) meter held much nearer to the projector than the screen is, can measure an "off" signal seemingly in the photopic range. These people are deluding themselves if they think the luminance in a plane near the projector is proportional to the luminance at the faraway screen, when the first luminance is in photopic range and the latter in scotopic range. Luminances on the screen are what should count in FOFO contrast measurment. Any measurement procedure which uses just a photopically corrected lightmeter yields hugely misleading FOFO contrast measurements in cases where projectors douse their R,G,B beams significantly unequally.

Is there an established ("official") test method for FOFO contrast measurement which the manufacturers follow? Does it specify photopic correction for both the "on" and "off" measurements? Then it is a boner. It will encourage projector design which pays too much attention to dousing the red beam and too little to dousing the blue beam.

Off the point, the AEMC CA813 lightmeter mentioned in other strands is a joke in respect to spectral correction. The AEMC webpage states its spectral response as the "CIE photopic curve", but without specifying the degree of fit to the curve the statement is vacuous. The webpage illustrates the CIE photopic curve and marks some points on it, as if those are data points for the CA813's sensitivity. The definitely aren't. Luxmeters costing several thousand dollars can't approximate the CIE photopic curve nearly this well.

attachment intended for previous post

People using an illuminance (lux) meter held much nearer to the projector than the screen is, can measure an "off" signal seemingly in the photopic range. These people are deluding themselves if they think the luminance in a plane near the projector is proportional to the luminance at the faraway screen...

I know of at least one AVS forum member that preaches exactly what you have outlined and he is completely "delusional"

dcouzin,

I don’t want to “pollute” your thread considering you probably know more about this subject matter than anyone here on this forum btw..I linked you to a comment that pretty much sums up my perspective. It will be my only post in your thread..and thank you for sharing your expertise! I've learned alot from your postings.



http://www.avsforum.com/avs-vb/showthread.php?t=1144400&highlight=

..But contrast ratio is one of the total horseshit measurements. It tells you NOTHING useful because something with poor black levels can have a high contrast ratio if the peak white level is very very bright. And some future technology with "perfect" blacks could have an infinite contrast ratio but only be able to achieve 1 fL (movie theater screens are typically 8 to 16 fL, home panel-type displays are typically set to 30-50 fL for dark-room viewing). Manufacturers cheat with CR measurements constantly (projectors have far more realistic specs than panels though). Most manufacturers blank the panel (turn the panel off) to measure black level - a condition that will never happen during real programming. I've see a projection demo at CEDIA where the contrast ratio was 100:1 and it was probably the best projection demo I've seen. You don't need a high contrast ratios to get great images.

dcouzin,

On/off CR definitely isn't perfect and I think it is a good point that vision is different as far as which colors we see in bright stuff compared to dark stuff. I've pointed this out before. But, I don't know of anybody who uses 2 different meters for the test or 2 different settings as far as this goes with one meter. As I've mentioned before, my modifications to an Optoma H79 took advantage of the fact that our vision works this way and although it measured about 9k:1, the effect was likely a little more because the on/off CR for red was lower than for blue and green with my design. And a projector with a higher on/off CR as measured with the same meter could look like a little lower on/off CR because of this factor, like you said.

But I don't know of anybody who skews their numbers based on that factor. Yes it is a possible error as far as what people will see, but I also don't expect people to use 2 different meters. And people can look at the color balance coming out of the projector for video black to see if it has a big skew toward blue by putting a piece of paper near the lens.

And why do you think it needs to be measured on the screen? Are you saying that because of color shift of the screen, because a meter could still measure with a photopic curve and scotopic curve near the projector. With black levels way down into an area where we can still see differences I think it is going to be tough to find a meter that can measure accurately that low off a screen. The Minolta LS100 reads down to 0.001 LUX with some error and I don't think that is good enough for some of these.

As far as tbrunet, he has been dissing on/off CR for years in really ignorant ways. Like claiming that more than 219:1 CR wasn't possible with 8 bit video, telling somebody that they must have measured wrong because they measured higher than that, claiming that people can only see 100:1 and so CR didn't matter, and for years has been playing this game about our video being linear even though gamma was pointed out to him long ago. Even more than 2 years after the discussion started he backed up that gamma didn't apply to digital video and that a 50%stim (or 50 IRE) level would result in 5 ft-lamberts if reference white was 10 ft-lamberts, along with answering my question about how bright 10%stim would be by saying 1 ft-lambert in that case. It isn't just that lack of comprehension of the subject matter, but it is the intellectual dishonesty of continuing to claim he understands this while people like Greg Rogers doesn't, even as he doesn't even comprehend the simple stuff like that. Many people have him on their ignore list for some of his underhanded tactics.

As I said elsewhere, one difference between you and tbrunet is that when I pointed out that you had missed gamma you were a man and admitted that you had left it out. This is much different than tbrunet who instead spent a couple of years trying to mislead people about what levels would be all while never once providing a single measurement he has ever taken of light levels. I don't know if he'll ever stop his BS about things being linear and so he was right.

As far as Doug Blackburn's claims, I wonder if he would say that color measurements are complete horse**** because the color measurements can be perfect and other things weak. No one spec tells everything. They are each indicators of performance. ANSI CR is one. On/off CR is another. Color gamut is another. RGB tracking is another. White level is another. It is rather ridiculous IMO to claim that one is horse**** because another can be poor. If your white level is way too low that is a problem no matter the CRs. If it is way too high, then with projection it can be dimmed down (including with darker screens or ND filters). Manufacturers can cheat or lie with basically any spec. As I have mentioned before, all of these things matter after the display is setup well.

--Darin

And why do you think it needs to be measured on the screen?

I don't. Sorry if my writing didn't make that clear. Whatever is read for the "on" and the "off" must be in the same proportion to the luminances on the screen, that's all.

As I've mentioned before, my modifications to an Optoma H79 took advantage of the fact that our vision works this way and although it measured about 9k:1, the effect was likely a little more because the on/off CR for red was lower than for blue and green with my design.

I remember that. And why believe it was likely a little more? It could have been much more, say by a factor of 2. If everyone's using photopic meters for both measurements they're probably stating false values, and worse, they're dissuading manufacturers from going your route of sacrificing the red dousing for the green and blue dousings. This is an example where having a better test procedure leads to better products.

Incidentally, you might not be able to buy a scotopic corrected light meter. It would be simpler to buy a cheap spectroradiometer and use it for both measurements.

I remember that. And why believe it was likely a little more?You're right. Just a matter of perspective. Some (including me) might have called it much more if I had measured that accurately.

With the H79 it worked out pretty well because the UHP bulb was weak in red, so the thing I did to balance the colors to get close to D65 for white optically also helped the on/off CR for blue and green more than red, going well with our vision.
If everyone's using photopic meters for both measurements they're probably stating false values, and worse, they're dissuading manufacturers from going your route of sacrificing the red dousing for the green and blue dousings. This is an example where having a better test procedure leads to better products.I agree with you in a way. I wonder if it would make things harder enough to do accurately for many people though. I already see people not understanding margin of error and giving 4 digits of accuracy with a black level reading that was only one digit on the meter. I'll sometimes give a little more than strict rules might call for, but some of these go way too far.
Incidentally, you might not be able to buy a scotopic corrected light meter. It would be simpler to buy a cheap spectroradiometer and use it for both measurements.Is the spectroradiometer and software with it likely to automatically adjust to the scotopic range, or do you think people need to do some special calculations for the difference?

--Darin

Is the spectroradiometer and software with it likely to automatically adjust to the scotopic range, or do you think people need to do some special calculations for the difference?

There's no reason to have a meter with 100,000:1 linearity. Mount the spectroradiometer at distance1 for the "on" measurement and distance2 for the "off" measurement and also measure distance3 to the screen. Choose the d1 and d2 so the instrument's within its range. Inverse square law is more accurate than any ND filter.

Ocean Optics makes good cheap instruments. http://www.oceanoptics.com/Products/spectrometers.asp

There's no reason to have a meter with 100,000:1 linearity.I agree as long as not taking things at screen level.
Mount the spectroradiometer at distance1 for the "on" measurement and distance2 for the "off" measurement and also measure distance3 to the screen. Choose the d1 and d2 so the instrument's within its range. Inverse square law is more accurate than any ND filter.That would work as long as people take into account the image size and where it effectively starts. Another method I've mentioned is to measure 100 IRE and 20 IRE from the screen and 20 IRE and 0 IRE from close to the projector, then multiply the ratios.

What I'm confused about is how you get the scotopic values vs photopic values with the meter. Like whether it is automatic when doing it this way.

--Darin

I'd like to see some actual data.

What I'm confused about is how you get the scotopic values vs photopic values with the meter. Like whether it is automatic when doing it this way.

The spectroradiometer produces a list in a text file which we manually pasted into an Excel worksheet, but any programmer can make this happen unaided. Sample attached. The Excel worksheet convolutes the spectral data with the two different visual sensitivity functions, essentially making the two lightmeters from it. A cheap spectroradiometer offers better fit to the functions than any filtered meter can. It is less stable than the filtered meter and has a smaller range of linearity.

I'd like to see some actual data.

So would I. Right now the "actual data" is darinp2's report that his H79 projector (after modification) shines a black which is reddish when viewed close to the projector lens but doesn't look reddish on the screen. This is the same phenomenon which makes the choice of red for fire engines a mistake when they drive at night.

If darinp2 can estimate how reddish the black beam close to the lens looks, e.g., by visual comparison with CC filters or something we can trust on his computer screen, then we can calculate how erroneous the photopic measured FOFO contrast ratio was.

If darinp2 can estimate how reddish the black beam close to the lens looks, e.g., by visual comparison with CC filters or something we can trust on his computer screen, then we can calculate how erroneous the photopic measured FOFO contrast ratio was.Unfortunately that was years ago, I dismantled it, and pretty sure I couldn't do it from memory.

One of the best calibrators, Ken Whitcomb, setup a dual stack G90 CRT system for Cliff on the CRT forum and did some measurements shooting into the individual tubes to try to figure out the on/off CR. My memory is that one of the red tubes had a pretty high black level reading compared to the rest and this lowered the calculated value. But I pointed out to Cliff that that red amount might be hard to see down that low and asked if the image looked like it had a skew toward red off the screen with blackouts. He reported that it didn't. This was with a calculated on/off CR of something like 500k:1, but with your system would have likely measured much higher (probably more than 2x from what I recall of the readings).

If you want to see one of the best CRT setups before it goes away and are willing to travel that far, I'm sure you would be welcome at Cliff's near Chicago in the US. He now has a triple stack of G90s setup by Ken. I was going to see it a couple of weeks ago, but with some flight problems didn't get enough time in Chicago to go there. I'm not sure how long he will keep it a triple stack, but I'm thinking about trying to make it back there before he dismantles it back to a dual stack.

BTW: I wouldn't mind that much if on/off CR was done in the perception range totally. Not just photopic vs scotopic, but more along the lines of our vision being largely logarithmic. Then people might not see a big number and assume that people really couldn't see beyond that and they might see that it takes a lot of improvement to on/off CR to get much improvement in perception space. As I've mentioned before, to get a medium sized number in perception space takes a much larger number in light power (what on/off CR is done in) space. I haven't done the math, but getting 100:1 in perception space takes a lot in on/off CR space.

EDIT: As far as the H79, instead of trying to do it visually I could say that based on the sizes of the irises it would probably be reasonable to assume that the on/off CR for just red without consider scotopic may have been in the 5k:1-6k:1 range.

--Darin

So would I.The reason I mention it is that comparing the CA 813 against a tristimulus device--whose luminance accuracy I independently verified with a spectro--indicated that the CA 813 was accurate at least down to 0.1 lux.

This is starting to sound like there might be a need for a standard to be established for measuring on\off CR or a revision to one if there is one already.

Obviously there is a standard for measuring ANSI CR. What does that standard for the instrumentation yada for measuring the black squares?

This is starting to sound like there might be a need for a standard to be established for measuring on\off CR or a revision to one if there is one already. Obviously there is a standard for measuring ANSI CR. What does that standard for the instrumentation yada for measuring the black squares?

It is appropriate to use photopic sensitivity when measuring the black squares in ANSI CR because someone seeing a screen half covered with white will be photopically adapted. In my view the scotopic problem arises with FOFO CR. Every projector manufacturer specifies FOFO CR without citing the measurement standard used. Perhaps they have an informally agreed upon method. Perhaps not, in which case they should all be fearful.

If the "off" output is spectrally neutral, then it doesn't matter whether it's measured photopically or scotopically. From such basic considerations as the small mirrors diffraction contribution to DLP's FOFO CR we know that the different color bands douse differently. (Actually this factor will cause redder black for which photopic measurement exaggerates the contrast loss.)

Who will be the Ralph Nader for video projectors?

Call Ralph whoops Joe Kane. Maybe he will do it. Not me. The manufacturers agreeing in secret or informally? I doubt they have much of a clue as to what is being talked about here.

The reason I mention it is that comparing the CA 813 against a tristimulus device--whose luminance accuracy I independently verified with a spectro--indicated that the CA 813 was accurate at least down to 0.1 lux.

I wrote: "Off the point, the AEMC CA813 lightmeter mentioned in other strands is a joke in respect to spectral correction." The matter of the CA813's photopic correction is separate from the main point about the need for scotopic correction for the FOFO CR "off" measurement.

Luxmeters are made for measuring more or less white illumination. The test of a well-corrected luxmeter is its ability to measure the full variety of white illuminations including blackbody, various fluorescent lamps, triband LEDs, fluorescing LEDs, etc. etc. CIE suggests a measure f1' of photopic spectral fit which does decently for evaluating luxmeters for different white lights. If all white illumination were, for example, 2856K blackbody, then a meter could have horrible photopic correction, could even be sensing at just one wavelength, and be a perfect luxmeter. So a fair test of the AEMC CA813 is to compare it with a trusted reference luxmeter on a wide variety of white illuminations.

A tristimulus meter is a funny choice for trusted reference meter, since the tristimulus meter incorporates three separate sensors to measure CIE xbar, ybar, zbar, and none need to be extremely accurately fitted to produce the meter's chromaticity measurements. When measuring lux, the tristimulus meter reports just its ybar sensor. So the $3000 Minolta CL-200 tristimulus meter has the exactly same CIE f1' = 8% fit to photopic sensitivity on its ybar sensor as the $1000 Minolta T-10 luxmeter has for its sole sensor. Still, CIE f1' = 8% isn't horrible, and it would be interesting to know how the AEMC CA813 compares with the Minolta T-10 on a wide variety of white illuminations.

As for your having "independently verified" the "luminance accuracy" of the tristimulus instrument "with a spectro", I don't know what you mean. Did you verify the CIE f1' spectral fit of the tristimulus instrument's ybar sensor? Or did you use one white illumination and find agreement in the lux readings over a wide range? From your statement "accurate at least down to 0.1 lux" I'd guess that linearity, not spectral accuracy, was your prime concern in all your comparisons of the instruments. Linearity certainly does matter when one meter will be measuring FOFO CR from one position. But linearity has nothing to do with spectral correction, which is the topic of this strand.

Perhaps I shouldn't have called the AEMC CA813 "a joke in respect to spectral correction" since I've seen no test data for it's spectral correction. AEMC's disclosure of its instrument's spectral correction is definitely a joke. And having seen how hard the first rate luxmeter manufacturers work to achieve their spectral correction, which they report as an f1' value or as a curve or a tabulation, AEMC's disclosure being vacuous is not encouraging for what's inside the case.

The manufacturers agreeing in secret or informally? I doubt they have much of a clue as to what is being talked about here.

I don't know what kind of industry it is. In friendly industries companies cooperate ahead of standardization. In combative industries standards bodies serve as referee.

One simple sentence, "For full on/full off contrst meaurement scotopic luminosity measurement is required for the full off state", would be immediately understood by engineers at all the companies. There might be an objection, which I expected to hear in this strand, that since scotopic adaptation takes time and the length of a dark scene is unknown, photopic measurement is as good as scotopic measurement. I defer to darinp2, who made and enjoyed a projector with high red in the black, that the temporal objection won't fly.

I wrote: "Off the point, the AEMC CA813 lightmeter mentioned in other strands is a joke in respect to spectral correction." The matter of the CA813's photopic correction is separate from the main point about the need for scotopic correction for the FOFO CR "off" measurement.My only point is that I have seen zero empirical data to suggest that the CA 813 is unsuitable for on/off contrast measurements. In fact, the data I have suggests that it is quite accurate, at least down to 0.1 lux.

The tristimulus meter is not my "trusted reference meter". A 5nm spectroradiometer is. What I said was that I used the spectro to verify the tristimulus meter's luminance accuracy. I then used the tristmulus meter for the remaining tests because is is much faster and has a wider dynamic range than the spectro.

Unless you are suggesting that there is an amazing coincidence in which both the tristimulus meter AND the luxmeter are both inaccurate and in exactly the same way, I see no reason to doubt these numbers.

BTW, the upper limit of the scotopic range is 0.01 cd/m2, so if the contrast measurement in question involved a black level higher than that, then your concerns about spectral correction in the scotopic region is not relevant. This would apply for any front projector whose on/off CR is approximately 5,000:1 or less or any flat panel whose on/off was 12,000:1 or less. BTW, the measurements I took were on a flat panel whose on/off CR was well below below that limit.

But linearity has nothing to do with spectral correction, which is the topic of this strand.No, the topic of this thread, at least judging by the title, is what instruments are suitable for on/off contrast readings. You are making an argument about that by way of a more technical discussion of the need for scotopic spectral correction.

What I said was that I used the spectro to verify the tristimulus meter's luminance accuracy.

This statement simply repeats your previous vague statement, which caused me to I ask what you meant by this verification: "Did you verify the CIE f1' spectral fit of the tristimulus instrument's ybar sensor?" That would be slightly more informative than reading it in the tristimulus instrment's specification.

We're talking about the spectral fit of the CA813 to the CIE photopic luminosity function, right? So naturally it's the spectral fit of the tristimulus meter to the photopic luminosity function that you need to verify in order use the latter to prove the former.

Unless you are suggesting that there is an amazing coincidence in which both the tristimulus meter AND the luxmeter are both inaccurate and in exactly the same way, I see no reason to doubt these numbers.

What numbers? I don't know what you measured! You only say that the tristimulus meter and the CA813 agreed down to 0.1 lux. What is the relevance of 0.1 lux when we're talking about spectral correction? You can determine whether the tristimulus meter and the CA813 agree in spectral correction at a single midrange lux level. I don't think you determined this. If they both said 100 lux in some simple illumination, especially under a 75 watt incandescent lamp, then this only says they are both calibrated correctly for lux. If you didn't compare them under a wide variety of illumination types, not illumination levels, then you didn't compare their spectral sensitivities.

BTW, the upper limit of the scotopic range is 0.01 cd/m2, so if the contrast measurement in question involved a black level higher than that, then your concerns about spectral correction in the scotopic region is not relevant.

I agree. If the projector is very bright and its FOFO CR is middling, then its "off" luminance could be over 0.01 cd/m². This is why I suggested in a post above (concerning software) that distances d1,d2,d3 should be measured. d3 (projector to screen) would be used to check if the "off" luminance on the screen is in scotopic range. (Actually we need the screen gain entered too.) For the mesopic range (between the photopic and scotopic) the two sensitivity functions could be blended. There's a further issue here. Why should a manufacturer need to think about d3 or screen gain when specifying the FOFO CR of a projector? If someone wants to project tiny images at 1000 cd/m² he should get a believable FOFO CR too. So the manufacturer can throw up his hands, and then measure FOFO CR any way he pleases.

We can limit our attention to the participants in this forum, who are projecting at about 50 cd/m² and expecting at least 5,000:1 CR, often 10 times that. Their "off" luminances are in scotopic range.

Originally Posted by dcouzin: But linearity has nothing to do with spectral correction, which is the topic of this strand.

No, the topic of this thread, at least judging by the title, is what instruments are suitable for on/off contrast readings.

Meter linearity is not strictly necessary for measuring projector FOFO CR. It's only necessary if you're trying to measure the "on" and "off" in the same planes. But d1=d2 is not necessary.

Good spectral correction is somewhat necessary for measuring projector FOFO CR even if you figure that your "off" should be measured photopically. This is because the "on" and the "off" lights have different spectral power distributions.

The published f1' for a CA813 is 8%. I find that to be reasonable based on what I have seen.

The effect of vision changing with the relative state of dark and light adaption of the viewer reduces the value of on/off contrast measurements for comparison purposes as you transition to lower light levels. With full dark adaptation not only does our sensitivity to various frequencies shift, but so does the efficiency of our eyes. In a home theater the relative state of dark adaptation of an individual will depend on many factors including the viewers age, cumulative effect of room light level and the duration of a given dark scene.

I know when I look at my screen when it is dark it looks nearly pitch black at first and slowly gets brighter as I sit there. This is for a system which runs 0.00034 fL with about 41,000:1 on/off contrast on the scotopic scale. When I am not fully dark adapted those values are going to be close, but as I dark adapt my perceived black level is rising temendously.

Just because the light level falls into the scoptic or mesopic range does not mean we are dark adapted to that level. If you are watching a movie like Dark City it might be possible to be dark adapted to some degree, but most very dark scenes are short enough that most viewers will not fully dark adapt.

Here is a link to more on this subject http://optics-vision.med.uoc.gr/pdf/Plainis-Dark%20adaptation.pdf.

What numbers? I don't know what you measured! You only say that the tristimulus meter and the CA813 agreed down to 0.1 lux. What is the relevance of 0.1 lux when we're talking about spectral correction?
http://www.avsforum.com/avs-vb/showpost.php?p=16249621&postcount=81

Since these measurements never got into the scotopic region, I suppose it hardly matters. But actually that was my point in the previous post. Even if everything you say is correct and all the implications are sound, it's not relevant for dark level readings above 0.003 fL, so the real-world implications of your post are a good deal less dramatic than the absolutist language you used (e.g., "hugely misleading" "a joke") would lead one to believe.

Again, no one in this thread (or in this forum for that matter) cares the least about "spectral correction" per se except to the extent that it substantially impacts the ability to obtain measurements of displays within a reasonable range of tolerance, in this particular case the ability to measure on/off contrast.

Your original claim was that using an instrument that fails to employ spectral correction results in "hugely misleading FOFO contrast" and that the "AEMC CA813 lightmeter. . . is a joke in respect to spectral correction" from which I suppose we were expected to conclude that the AEMC CA813 lightmeter results in hugely misleading measurements of on/off contrast. You seem to have moderated that view so that now the joke only concerns AEMC's documentation, rather than its actual performance, with which I gather you have no first hand experience.

To be quite honest I guess I was a little irritated by your summary dismissal of an instrument's performance on purely theoretical grounds and without the slightest amount of empirical data to substantiate your skepticism. The last thing this forum needs is more armchair theorizing.

If you really believe your thesis, you might consider investing the time and resources into performing the necessary tests that would reveal such instrument's capabilities. Now THAT would be useful and welcome.

The published f1' for a CA813 is 8%.

Published by whom? It's not in the specifications on the AEMC website http://www.aemc.com/products/pdf/2121.20.pdf

With full dark adaptation not only does our sensitivity to various frequencies shift, but so does the efficiency of our eyes.

I think you are saying that on the way to full dark adaptation the eye's spectral sensitivities shift and also its absolute sensitivity shifts. I agree with the latter. As Hecht in The Quantum Relations of Vision (1942) put it: "It is obvious that the eye must be thoroughly dark adapted. This means at least a 30-minute stay in the dark before measurements can be begun."

But I don't agree that the spectral sensitivity is shifting significantly all the while the absolute sensitivity is. The spectral sensitivity is practically scotopic as soon as the rod signals greatly outweigh the cone signals, and the rods simply continue to improve as they unbleach.

but as I dark adapt my perceived black level is rising temendously.

Yes, the perception is increasing -- the black becoming lighter -- until at full adaptation it's a noisy grey.

You might be questioning why the number of scotopic nits stays constant throughout this time while the perception keeps changing. Why do we pay attention to the eye's (relative) spectral sensitivity and ignore its absolute sensitivity? This is a deep quesion. A short answer is that contrast ratio is an objective quantity which does not measure the difference between two perceptions but rather measures what the projector can produce which then provides many perceptions.

Finally, it might sound a little fishy to be dividing the number of photopic nits by the number of scotopic nits. I don't know if the scotopic lumen or candela has been defined by any respectable body. I am content to normalize the two curves (accoding to their integrals) so 1 watt at 555 nm makes 683 photopic lumens and 1 watt at 507 nm makes 752 scotopic lumens.

Here is the user manual for the CA813. The f1' is listed in this on page 6.

...the real-world implications of your post are a good deal less dramatic than the absolutist language you used (e.g., "hugely misleading" "a joke") would lead one to believe.

You are throwing together two parts of my initial post which were separated by the phrase "Off the point".

I said: "Any measurement procedure which uses just a photopically corrected lightmeter yields hugely misleading FOFO contrast measurements in cases where projectors douse their R,G,B beams significantly unequally." I should have qualified this to cases where the "off" screen luminances are definitely in scotopic range.
I expect cases where using a scotopic meter for the "off" measurement will make a factor of 2 of difference to the CR, so I stand by this "dramatic" statement.

Then "Off the point, the AEMC CA813 lightmeter ... is a joke in respect to spectral correction." I later reduced this charge to the spec sheet being a joke. If you understood specifications of spectral correction you'd agree. Then UMR reported that "the published f1' for a CA813 is 8%." Well, if this is true, the off the point point disappears and the CA813 is a great bargain.

Your original claim was that using an instrument that fails to employ spectral correction results in "hugely misleading FOFO contrast" and that the "AEMC CA813 lightmeter. . . is a joke in respect to spectral correction" from which I suppose we were expected to conclude that the AEMC CA813 lightmeter results in hugely misleading measurements of on/off contrast.

No. No. No. You mistate my original claim. It was that using "just a photopically corrected lightmeter yields hugely misleading FOFO contrast measurements in cases where projectors douse their R,G,B beams significantly unequally." The key word is "just". When two meters, a photopic one and a scotopic one, are called for, using just one is a boner. This statement has nothing to do with the quality of either meter. I'm not implying that a f1'=8% meter isn't plenty good enough as the photopic meter. What you say you "suppose" is your logical confusion.

I wish I had not included the last "off the point" paragraph in my first post, because it really is off the point of the rest of the post and yet a reader might be unable to believe that.

If you really believe your thesis, you might consider investing the time and resources into performing the necessary tests that would reveal such instrument's capabilities.
I don't know what you mean by my thesis. I regard it as the part of my first post before the words "Off the point", never mentioning the AEMC CA813 lightmeter. But when you say "reveal such instrment's capabilities" I fear you're talking about the damned AEMC CA813. My thesis isn't about revealing any instrument's capabilities. It's about what projectors do and how to reveal that by use of well specified instrumentation and calculation.

Here is the user manual for the CA813. The f1' is listed in this on page 6.

Thanks. I wish AEMC had included this in the spec sheet on their webpage and spared me my error. The AEMC CA813 appears to be of high value for its price.

...I think you are saying that on the way to full dark adaptation the eye's spectral sensitivities shift and also its absolute sensitivity shifts. I agree with the latter. As Hecht in The Quantum Relations of Vision (1942) put it: "It is obvious that the eye must be thoroughly dark adapted. This means at least a 30-minute stay in the dark before measurements can be begun."

But I don't agree that the spectral sensitivity is shifting significantly all the while the absolute sensitivity is. The spectral sensitivity is practically scotopic as soon as the rod signals greatly outweigh the cone signals, and the rods simply continue to improve as they unbleach.

....

You need to study the results of CIE TC1-58. The mesopic range is a mix of the spectral response from scotopic and photopic.

http://www.lightinglab.fi/CIETC1-58/files/MOVE_Report.pdf

Thanks. I wish AEMC hadn't omitted this from the spec sheet on their webpage and spared me my error. The AEMC CA813 appears to be of high value for its price.

I have had two break on me in a few years. It has been the least reliable piece of gear in my kit.

I would look at Extech instead. I believe they have a similar meter with an f1' of about 6% and I hope it would be more reliable.

You need to study the results of CIE TC1-58. The mesopic range is a mix of the spectral response from scotopic and photopic.
Actually I have attended one or two meetings of CIE TC1-58. The "off" luminance will generally be in scotopic, not mesopic range.
Let me expand my claim that "The spectral sensitivity is practically scotopic as soon as the rod signals greatly outweigh the cone signals, and the rods simply continue to improve as they unbleach."
Immediately upon dropping the luminance to, say 0.005 nits, the cones produce insignificant signals. So the photopic luminosity function simply doesn't pertain to the sensation. Initially the rods produce insignificant signals too. So we're briefly blind. But as the rods unbleach they increase in absolute sensitivity. However, their spectral sensitivity doesn't change. Thus on sudden passage from 50 nits to 0.005 nits there is sudden change from photopic spectral sensitivity to scotopic spectral sensitivity. There is no transitional mesopic spectral sensitivity, no mixing of rod and cone vision.

I would look at Extech instead. I believe they have a similar meter with an f1' of about 6% and I hope it would be more reliable.

I'm a firm believer in spectroradiometry, not photometry. I'm saving up for an Ocean Optics model.
Actually the CIE f1' measure nauseates me, as I've expressed in CIE and ASTM meetings for years, but it is sooo entrenched.