I recently calibrated my RS-2 with a Lumagen HDP with all DeltaE's at or below 10 and Luminance ranging from -10% or less on the primaries and less than + 10% on the secondaries, except magenta, which is at about +15% luminance. I can't really get much closer with the HDP. Greyscale is perfectly flat and gamma is perfect 2.2. Since I have never knowingly witnessed a calibrated display with correct colors, I can't be sure that what I am seeing is "correct." I notice that the fleshtones that I am seeing on caucasians tends to be a bit pink. Not horrifically so, and the overall picture looks incredibly real, but like I said, flesh tones do look a little pink. However, as I look at my own hands while typing, they too look just a little bit pink.

For those of you with far more experience than I with calibrated displays, did you find that flesh tones for caucasians looked a bit pink at first (due to cooler color settings taking the natural pink out as they lean towards blue), or does this sound more like a problem with my magenta color setting?

P.S. I now can see the mild blotchy patches of imperfection on Hollywood's "beautiful" people.

I recently calibrated my RS-2 with a Lumagen HDP with all DeltaE's at or below 10 and Luminance ranging from -10% or less on the primaries and less than + 10% on the secondaries, except magenta, which is at about +15% luminance.

These are significant errors if these are the true measurements. Consequently, this is what you are seeing with real world video content.

I am confused. Significant or insignificant? I was under the impression that deltaE's of below 10 were very good and 3 or less were imperceptible. All my deltaE's are at 10 or below. Magenta measures from 9.9 to 10 per measurement. The others range from 4% - to 8%. All luminance is right about 10% negative on primaries, and under 8% positive on yellow and cyan, with magenta being the slight outlier at plus 15%. I do not have my saved file with me at the moment, but I can post it later tonight.

I am confused. Significant or insignificant? I was under the impression that deltaE's of below 10 were very good and 3 or less were imperceptible. All my deltaE's are at 10 or below. Magenta measures from 9.9 to 10 per measurement. The others range from 4% - to 8%. All luminance is right about 10% negative on primaries, and under 8% positive on yellow and cyan, with magenta being the slight outlier at plus 15%. I do not have my saved file with me at the moment, but I can post it later tonight.

You will get different delta readings depending on the formula you are using. It has been determined that the human eye can perceive color differences down to a delta error of 3. Even less on side by side comparison. Your luminance error percentages are quite high. -10% on primaries, yet +15% on Magenta. That is a discrepency of 25%.
You are obviously seeing color errors with video content. That should be your confirmation that something isn't quite right.

A few other things to be considered is the accuracy of your meter. I'm not sure what you are using, but anything less than a spectroradiometer will yield inconsistent results. Room reflections can also be skewing your readings.

I certainly have a less than ideal meter, as it is simply an Eye1 LT. Without a Lumagen radiance, it is impossible to get my particular display to have 3% or less DeltaE's without messing up the Luma more than it is because I have no control of the secondaries or individual Luma. My hope was simply to get reasonably close. I can say that my DeltaE's and Luma errors were much higher with any of the default settings (or at least they were on at least 1 or 2 colors). So, it sounds like my set is somewhat calibrated, but still perceptibly not ideal. So, back to my original question, do people tend to find caucasian skin tones to be a bit pinkish when they first see them on a properly calibrated set?

I guess a different way of skinning this cat is trying to find a reference display somewhere in Dallas that I can see for myself.

You are confusing the discussion by referring to dE as % - the calculated dE value has nothing to do with % of anything. It's a unitless number.

The LT is a very limited instrument. It needs to be on for at least 1 hour aimed at the display you are going to calibrate and after that hour, you must perform the dark calibration on a black, opaque surface in complete darkness. Even then, calibrations several days apart will produce surprisingly large differences in the measurements - that's just what you get with the LT.

While I've worked with a number of Radiance processors, I've not used an HDP. But if it doesn't have CMS adjustments to make the colors more accurate there's nothing you can do about them anyway. If you find flesh tones too pink, I would suggest looking for a red or magenta bias in your grayscale in the70%-90% range (and you may need intermediate steps to see it... like 65%, 75%, 85% if you can find such patterns on some disc you have).

The error you are seeing could be "built-in" to the LT or you may not have warmed it up long enough - also make sure the light the LT is measuring is on the same axis as the light you are viewing from the screen. Meaning... when you are in your "best" seat watching the screen, the LT's sensor should perfectly block your view and be aimed along the same axis as your eye-to-screen-center axis. If you put the meter too high or too low or too much off to one side, you'll be measing an axis you don't view the images on and some screens change the light quite a bit off-axis.

If you want to talk about luminance of colors, it's best to simply refer to the target and measured luminance in fL (or nits -- cd/m2, either is fine as long as you stick with just ONE measurement unit). Your target luminance value comes from your white measurement... so if you measure 100% white, you MUST measure 100% colors also. If you measure 75% white, all the colors should be measured at 75% also. Failing to follow that convention will result in wrong results. So it's better to say something like "Magenta measures 4 fL when it should measure 5 fL"

Also... the LT really isn't very good with displays that use projection lamps. It does better with displays using phosphors to produce the image (CRT and plasma). Projection lamps have a very very different light spectrum. And light from an LCoS projector reflected from a screen is very very (tempted to add a 3rd very) different in spectral content compared to a plasma or CRT display. So you may be fighting a losing battle using the LT. It is very likely to mislead you given your setup.

Wow, very informative post Doug. Thank you. I am measuring a Lcos projector with the LT, so it sounds like I am fighting a losing battle. I was under the mistaken impression that I only had to warm up the LT on a plasma screen, not for a projection system. I will try that and perhaps it will at least help a bit. My screen is neutral gain, so I do not really get any hot spotting or viewing angle issues. However, by dumb luck I placed the meter right in the middle at eye level from my normal seat.

The Lumagen HDP now has the ability to adjust red, green and blue on each of the primaries and white. You can also adjust these 3 colors and overall Luma within the grayscale controls. I can go to any % white I want in an 11 point scale (i.e., I can pick custom white levels like 7.5%, but I only get a total of 11 points). What I cannot do is adjust luma on the individual primaries or do anything purposeful on the secondaries. The secondaries are the big problem with the HDP, because by pulling in the oversaturated primaries, the secondaries tend to get moved out of position a bit and tend to get too much luma. The luma is typically the larger magnitude problem.

The way I calibrated was to calibrate the meter with a flat black surface, then calibrate white using the meter and the RS-2's onboard color controls. I ended up having to drop blue a ton and green a reasonable amount to get white to be at the proper x,y coordinates. Once I had white dialed in using a 100% white full screen test pattern/field, I then checked R,G,B,Y,C,M. Then I went through an iterative process of slowly pulling the primaries in by adding blue and red to green and substracting green from green, etc., After each change I would retest all of R,G,B,Y,C and M. Almost always the readings would have the expected result (or change in some magnitude in the expected direction). After getting the DeltaE's to under 10 on all primaries and secondaries with Luma errors of 10 or less (- for primaries and + for secondaries, except +15 for magenta), I then calibrated my gray scale and gamma curve using the 11pt controls. Once these were flat with a flat 2.2 gamma, I rechecked colors. Things moved somwhat of course, but still within the 10 point tolerance.

One odd thing I found is that my FTL reading on white has ranged from 7-9 FTL, which is really low. However, when watching film, the picture does not appear dim to me. Dark scenes are dark, but light scenes are light. I see a ton of shadow detail. I believe that 12 ftl is about what I "should" want, so either I am weird or my meter is really far off on FTL measurements.

I will give it another try tonight after letting the meter warm up and see how different my results are.

Thank you again.

One odd thing I found is that my FTL reading on white has ranged from 7-9 FTL

On my Sanyo Z2000 PJ, I also get around 8 for the FtL, but the picture looks ok to me as well.

However, I'm getting low FtL using window patterns, when using full screen patterns, the FtL is much higher. I'm also using an i1 LT.

I wouldn't worry too much about the low FtL.

I'm no expert, so it's just my 2 cents.

Pitou!

Most people have way dimmer images than they realize. The lumen and gain values people assume are generally optimistic and it only gets worse as the lamp can lose 50 percent of its output at the end of life. I tell you more front projection systems than not that I run across have 6 to 9 fL maximum with low lamp hours. This comes from pushing screen size with machines that just do not put out the lumens required either from too long a throw or just because the machine is not really delivering what people think they will.

An 8 fL maximum system will not look incredibly dim in a dark room. However, it will not have the punch of one that can put out 12 to 15 fL.

I also wonder if distance from the screen is an issue. I have my Eye1 LT mounted on a tripod, but to put in front, it also has a center channel speaker in the way, so the meter ends up being about 18 inches away from the screen.

Don't forget that when I said the LT meter is not very good at measuring luminance that the fL measurements you are talking about ARE luminance... you are probably working in xyY space... x and y are the coordinates that tell you whether your white or shade of gray or primary or secondary color are accurate color-wise. But the Y is the same thing as the fL you are measuring and the LT meter isn't very good at that. There's no way to know whether it reads high or low, or by luck, reasonably accurately for the Y (fL) measurements. So you probably shouldn't be too wrapped around that axle.

As umr said, most people end up with too little projector for the size screen they are using because the screen/projector combo can be rather pricey. When you want 100" to 120" for the screen, you need a LOT of light from a projector to get 12-14 fL on the screen. But the projectors with the light capability can cost double the price of a projector that seems like it might just be good enough (on paper) for a screen that large. You're looking at an $8000 projector plus a screen and all the published numbers tell you the projector should produce enough light for a 100"+ (wide) screen, but the numbers stretch the truth and you end up with considerably less light than you thought (wished) you might. It turns out that the $15,000+ projector the same company offers has the amount of light necessary to get you to 14 fL on a 100"+ screen... but a lot of people either can't or won't spend that much more on the projector... but they don't want a smaller screen either. So it's a real tug-of-war between budget, screen size, and screen luminance.

Is it better to have a dimmer calibrated PJ or a brighter not calibrated PJ?

Any thoughts?

Pitou!

I am not a professional calibrator but in my opinion.
I would use gamma to determine what the contrast (white level) of the display should ideally be.
Given the option with a flat pannel I would sacrifice brightness for greyscale color accuracy. With a projector you might have to sacrifice color accuracy to maintain decent gamma.
Dropping actual luminance by 50% is only perceived as about a 20% drop in luminance. In a dark/dim viewing environments the brightest thing is percieved as white even if in comparison to brighter things it would look grey.
So it comes down to how much contrast ratio and gamma are being effected by luminance. A three fold increase in contrast is only going to look like a doubling, and anything less than a doubling of contrast is going to be difficult to notice except in direct comparison.
So it comes down to gamma. Gamma of 2.2 looks ok on living room flat pannels, according to EBU crt tvs average 2.35 gamma, but I prefer gamma of 2.5 on a projector, commercial cinema dci projectors which have a different gamma curve track about 2.6 gamma.
With dim displays like projectors increased luminance can also look better due to the eyes increasing contrast sensitivity and improving color vision at higher brightness. So brighter equals looking sharper and more colorful.

It also comes down to how uncalibrated?
Greyscale color temperature is most critical to perceived image quality at 50-70 IRE, next most important is 20-50 IRE, less critical are 70-100 IRE, and least critical is 0-20. Red/Green balance is also more critical than Blue. A delta E of 10 or below across the entire greyscale is good, delta E of 3 or below is excellent.

Is it better to drink terrible beer or terrible wine?

Is it better to drink terrible beer or terrible wine?

So you're suggesting that none are good.

What can we do then? Replace the lamp? Buy another projector maybe?

Pitou!

So you're suggesting that none are good.

What can we do then? Replace the lamp? Buy another projector maybe?

Pitou!

If the projector was calibrated while the lamp was so old the color had drifted, replacing the lamp will definitely "de-calibrate" the projector. If the calibration was done in the middle of the life of the lamp and now the lamp is older and dimmer but the color has not shifted yet, replacing the lamp should not change the calibration very much but the projector should be brighter... perhaps 50% brighter depending on how old the lamp is and what mode you select for the lamp (and iris opening if that is adjustable).

If the projector is never bright enough or never calibrated very well, you would have to decide if spending money on a newer projector that is brighter and has more calibration adjustments is worth the cost.