Official JVC RS20 / HD750 Calibration and CMS thread (NEW FIRMWARE V1.1) - Page 19

You've got some fairly big Hue adjustments in there. I spent five or six hours yesterday recalibrating and had to make some large Hue adjustments as well. When I measured the THX settings, my whole triangle was twisted clockwise a bit. Red was more towards magenta, green towards yellow, blue towards cyan. I dropped in lawguys' and lovingdvd's and ended up making changes to lovingdvd's settings as they were closer to being right. What really scares me is that I'm using an I1D2 and I don't have much confidence in it anymore. My grayscale looks good but I'm not so sure about the colors...

I was able to bump up my Green gain on the recalibration of the grayscale though, but I had much different settings for low bulb vs high bulb. I had to cut both green and blue a lot more in low bulb than in high. Jason originally calibrated my RS20 and the green gain was somewhere around -30ish on high bulb. When I recalibrated yesterday, I only needed green gain at -15.

I'm worried that the I1's may have problems on calibrating the color on these JVC's and I may be doing more harm than good. For those with good spectroradiometers, have you had to make big hue adjustments, or just pull in saturation and adjust brightness? If Hue was really incorrect, as my meter seems to suggest, would that show up when using the filters to adjust the color hue/tint from the DVE discs? If those don't show errors, are our meters reading wrong?

Manni, The 'dummies guide' says to put the meter as close to the screen as possible, just a few inches from the pics they show. Have you confirmed with others about what you are suggesting here? Tx, Bill

From what I have seen, the errors in the RS20 gamut are almost exclusively in saturation. Hues are off by only a very little and the lightness values are also very close to correct. However, you must get the grayscale right first. Here's what I am seeing from User 1 after grayscale calibration but before CMS calibration:

	R	G	B	Y	C	M
L	1.5%	-0.4%	-5.8%	0.0%	-0.7%	0.1%
S	12.6%	22.6%	5.9%	17.2%	24.0%	7.1%
H	-1.0%	2.3%	-1.9%	1.4%	-10.0%	4.5%

Negative values indicate clockwise errors in hue and positive values indicate counterclockwise errors.

The only difference in closer vs. farther is the aim point. Close readings measure higher luminance but over a smaller area. Farther away the level is lower but the area of the screen from which you read is larger. These two factors cancel each other out, so from the standpoint of the final luminance reading it shouldn't make a difference. A couple of feet away sounds about right to me. The angle of acceptance of these meters is fairly large so the viewing cone gets bigger fairly quickly as you move away from the screen, which makes getting an accurate aim point more difficult.

BTW, this is one of the advantages of high-end meters. They generally have either a lens or a spot laser that defines the area from which the device reads and a narrow angle of acceptance, which makes getting a good aim point easier and more precise.

OK, thanks Manni and Tom. I will try to do my grayscale with my I1LT 1 to 2 ft from the screen. I want to get the grayscale as best I can on my own before UMR arrives to do this magic--to see how far wrong I am even trying my best!

The strange thing in my case, however, is that although I have to make these big adjustments in hue, the visible result (eyeballing with R,G and B glasses to the different hue and color decoder checks) is 100% perfect!
Same goes for grayscale, by the way. That's excellent as well.

On one side, I am not 100% sure about the I1 LT, but on the other side: if things were off, it really should show with the visual color decoding checks and hue bars. And that is not happening.

As Mrlittlejeans remarked: measuring THX and most settings from others, results in a clockwise tilted triangle.

Well, I hope my I1 pro will arrive soon to put an end to my doubts.

Is this your new lamp that has gone dim now Manni?

I've have given up with using my I1-LT as I found that I can't repeat any readings with a reasonable accuracy even on my TV which I can't imagine will drift that much once warmed up. I can get a greyscale within a delta of 2 using my Lumagen to adjust, but if I check it the next day it has a delta of 5 or more. I don't know which reading to trust, so I've cleared the greyscale adjustments in the HDQ and I'm just getting on enjoying films as it is.

I might consider an ISF calibration as I've reached 100+hours (and I'm not looking for my next PJ either ).

In digital terms, white is R,G,B=235, black is R,G,B=0. If you want a color that resides half way on the line between white and Red, your signal would be R=235, G=125.5, B=125.5. There is no "calibration basis" AFAIK, for needing a color measurement with "75% saturation" as you describe.

If the gamma tracking for each of the three primaries is accurate, then all derived colors should be "OK" provided there are no decoding errors. We all know this doesn't happen in reality. To get to the 50% color you described does involve "Y" because it is derived from a combination of all three colors.

If the REC 709 colorspace tends to collapse or expand as stimulus levels change, those are CMS issues, generally not fixable. We just need to derive a useable compromise.

As you try to correct "Red" to REC 709, you add Blue and Green to affect Hue, but the additional amounts of B/G affect "Y" (saturation/color). As you get close, the filters will look good for decoding, measurements are required to verify properly.

These values give some support to my larger Hue correction for Cyan. In fact, putting aside Y values (which will be affects by Saturation adjustments), all of your measures, with the exception of blue, seem to correspond with my CMS values.

What do you do for blue? I simply could no influence its "y" without extreme adjustments and even then, I could never get it close to correct.

What that cheap thing? It's like a toy meant for children and is to be used stricty for fingerpainting color-correction purposes.

Feel free to PM me if you need any help. I don't promise to know the answer though!

You can always compensate for whatever effect the diffuser has by using an offset, but that assumes you have a valid off-the-screen reading to begin with, which I gather is precisely the problem, at least with blue.

Blue can be a pain because it is so dim. I think that your suggestion of using zoom to get a smaller, brightner area to read from is a good one. Assuming you have verified that the RS20 is linear from 75% to 100% stim (I haven't checked this yet), then use 100% test patterns. The combination of these two strategies should allow you to get a good off-the-screen reading without resorting to a diffuser.

Using fields is probably a good idea for this type of display if for no other reason than it eliminates problems with the aim point. The difference in brightness should be minimal. You will see a small difference because fields throw a lot more light into the room, which then reflects back onto the screen.

I didn't have any problem with blue. The i1Pro's poor low-light sensitivity may cause problems trying to get a valid reading for blue. I can only suggest the same things I did to Manni: use 100% patterns, full fields, and zoom the image down to make it smaller and brighter.

I think this is right. When I took constant readings, blue's numbers jumped around quite a bit. I'll try zooming in. Great idea.

I did too when I calibrated with my i1LT. I no longer do with my i1Pro. My largest is on Cyan, which Tom's measurments confirm has a hue issue.

mrlittlejeans' and Tom's comments on Hue made me sit down and compare my old i1LT calibration to my i1Pro calibration setting. They are actually very close. The biggest difference in colors other than blue is in the Hue settings. I don't know what to make of blue.

i1LT Calibration

r: 9, -26, -2
y: -6, -52, 34
g: -32, -48, 11
c: 49, -60, 29
b: 60, 34, -21
m: 0, -27, 1

i1Pro Calibration

r: 6,-30, 3
y: -6, -50, 33
g: -11, -47, 13
c: 40, -54, 29
b: 11, 8, -16
m: 8, -26, 10

This is what I once assumed as well, but because of gamma actually it is a little different. Using % values it is approximately R100%, G46%, B46%.


There can be and you allude to it later in your post. If the colorspace is not linear in that different levels of saturation do not fall where they should on the CIE chart, then compromises need to be made. Stereomandan has done some good work in this regard. It's not a problem with the RS20, but my Sharp 20K is very non-linear in this regard. When I get a fully saturated red correctly calibrated then all of the less saturated reds are significantly off from where they should be in the color space. Thus, I get better measured performance throughout the color space by allowing a little oversaturation at its edge.


Any color involves Y. It is part of how a color is defined. What I meant was that for the purpose of checking saturations, it doesn't matter how bright the red is. All that matters is where it falls on the CIE chart. The disc-based test patterns out there use constant brightness for the range of saturations, even though, all else being equal, if you simply reduce red's saturation and move it towards the white point, e.g., R100%, G46%, B46% it becomes brighter. Keeping constant brightness is a harmless methodological choice.

It is quite possible that--again because of its poor low-light sensitivity--you are getting a LESS accurate reading of blue with the i1Pro than you were with the D2.

That may very well be.

I adjust the CMS settings with the constant read feature on, but I always run full a set (or sets) of numbers using the gamut measuring function in which ColorHCFR, presumably averages many reads to improve accuracy for low light readings. I can tell it takes a long time when measuring bue.

Thanks Tom for sharing what you know about this. I feel I have come a long way in appreciating accurate color and it is in many ways thanks to you being such a persistent advocate.

I had it checked in my most recent calibration.

If anyone is interested in analyzing data sets, here are all the calibrations that we have to date. Any common denominators?


Marcel W

R 36, -23, 16
Y 8, -46, 36
G 24, -37, 28
C 4, -43, 34
B 36, -17, 10
M 8, -37, 25

Lawguy i1LT Calibration

r: 9, -26, -2
y: -6, -52, 34
g: -32, -48, 11
c: 49, -60, 29
b: 60, 34, -21
m: 0, -27, 1

Lawguy i1Pro Calibration

r: 6,-30, 3
y: -6, -50, 33
g: -11, -47, 13
c: 40, -54, 29
b: 11, 8, -16
m: 8, -26, 10

Manni (1.2 gain screen)

R=-4 -22 6
Y=9 -45 38
G=-4 -44 35
C=-2 -51 43
B=0 -20 4
M=1 15 7

Manni (1.4 Gain Screen)

R= -4 -22 3
Y=6 -46 37
G=-11 -43 31
C=-3 -51 39
B=30 0 -7
M=1 0 -5

Lovingdvd

R 2 -24 10
Y 3 -44 34
G -10 -42 32
C -2 -48 38
B 0 -25 9
M 2 -31 16

JeffY

R -4,-22,11
Y 5,-44,37
G -6,-42,35
C -2,-46, 37
B 6, -4, 3
M -2,-4,0

Karrih

R 16,-27, 12
Y 6, -48, 37
G 5, -45, 39
C 2, -49, 39
B 36, -18, 4
M 1, -30, 15

008

R; 3 -22 9
Y; 6 -44 37
G; 15 -37 27
C; 2 -43 35
B; 26 -5 -3
M; 5 -22 12

Nelson4u

Red -3 -22 6
yellow 2 -48 39
green -19 -45 35
cyan 1 -52 44
blue 11 -25 10
magenta -5 -29 14

Ignace

Red 6 -29 8
Yellow 1 -48 35
Green -24 -46 33
Cyan 8 -56 43
Blue 18 -24 8
Magenta -10 -32 15

I hope you are just 'unlucky' however, I'd be tempted to ask them to look at the circuitry that powers the lamp incase there is an issue with it not allowing the correct current/voltage to the lamp. It's a shame you're some distance from me otherwise you'd be welcome to try swapping lamps to see how bright yours is in mine & visa versa. I've got 110 hours or so on mine and I'm still mainly using iris at 13 (service menu tweak for the HD350 ) and low lamp on a 112" wide 2.35:1 screen 1.5 (claimed) gain.