Official JVC RS20 / HD750 Calibration and CMS thread

[Mark Petersen]

This thread is the official calibration thread for the RS20 and HD750. Please post any CMS and other calibration (greyscale, gamma, etc.) tips, techniques and tweaks into this thread. Please use the owners thread and other RS20/HD750 threads for other discussions not relevant to calibration. As time goes on I'll update this post with links to key posts and other references related to calibration.

Due to lack of time I haven't been able to post a formal writeup that recaps most of what is in this thread. Manni01 has graciously taken the time to do this though and you can read his write up here:
http://www.avsforum.com/avs-vb/showt...6#post15654856

When the final writeup is done I'll be sure to maintain the link above and also add any new links that come along.

[Mark Petersen]

Reserved.

[kutlow]

Mark wil you be posting your settings?

[Lawguy]

I wonder if the service menu has anything we can play with that might help us use the CMS?

[Mark Petersen]

Originally Posted by Lawguy

I wonder if the service menu has anything we can play with that might help us use the CMS?

Good question, JVC has changed the remote keystrokes to get into the service menu. Looks like we'll have to dig for it

edit: SM keystrokes are exactly the same, but it's harder to use than the RS1 because the remote is balky.

[Lawguy]

Originally Posted by Mark Petersen

Good question, JVC has changed the remote keystrokes to get into the service menu. Looks like we'll have to dig for it

I was afraid of that.

[lovingdvd]

Originally Posted by Mark Petersen

Good question, JVC has changed the remote keystrokes to get into the service menu. Looks like we'll have to dig for it

Perhaps there will be some settings in there helpful for taming the CMS...

[TomHuffman]

Until someone posts the raw xyY data, it is hard to tell much of anything. A CIE chart by itself is not very helpful, and even luminance graphs don't help much either unless you know the chromaticity points.

[mark meyers]

Originally Posted by TomHuffman

Until someone posts the raw xyY data, it is hard to tell much of anything. A CIE chart by itself is not very helpful, and even luminance graphs don't help much either unless you know the chromaticity points.

Makes perfect sense to me!

[Kris Deering]

Does this help Tom? This was for the THX mode.

[mrlittlejeans]

So using Kris's numbers and Gregr's color calculator I come up with the following dE's

White 10.7
Red 13.2
Green 2.5
Blue 13.6
Yellow 10.2
Cyan 14.2
Magenta 8.9

I wonder how much a grayscale calibration would help those numbers. These weren't really good.

[lovingdvd]

Originally Posted by mrlittlejeans

So using Kris's numbers and Gregr's color calculator I come up with the following dE's

White 10.7
Red 13.2
Green 2.5
Blue 13.6
Yellow 10.2
Cyan 14.2
Magenta 8.9

I wonder how much a grayscale calibration would help those numbers. These weren't really good.

Color dE aside, it seems very odd that white would have a dE of 10.7. I think this indicates the grayscale is quite off, at least at 75% or 100% or whatever level was used to measure the color.

Speaking of which Kris - for reference did you use 75% or 100% levels, and were these fields or windows? Thanks.

EDIT: Yes indeed the grayscale is off. Tom Hoffman nailed it just by looking at the hue shift in the CIE chart. Nice work Tom. I'm basing this off of what Kris posted here: http://www.avsforum.com/avs-vb/showt...3#post15325743 . Note the dE of 10 in that chart at 100% which matches the dE reported in Kris's white point reading in the .doc he posted above. So that also answers my question - the gamut measurements were taking at 100%. It would be interesting to try at 75% which has a dE of about 7 and see if the readings improve a tad, just as an experiment.

[TomHuffman]

Originally Posted by Kris Deering

Does this help Tom? This was for the THX mode.

Kris: Very much so.

OK, we have a good idea now what the THX mode offers.

dE Performance (CIE94: 1.0 minimum perceptible error, 1.5 maximum acceptable error)



Chromaticities



LSH Analysis



Luminance Performance



As presets go, the chromaticities are pretty good, though red is more orangy than I would like.

The dEs are more of a problem. None of the colors, except green, are within a reasonable set of calibration targets, though it is certainly a big improvement over the RS1/2. The secondary dEs would be much better after a gray scale calibration. Here the white point is shifted away from blue. But the red and blue dEs are about double of what I would like to see and gray scale calibration wouldn't help these.

The biggest problem is the brightness of the the colors along the line of purples. Oddly, the remaining colors are nearly perfect. This is what takes the dEs of red, blue, and magenta considerably above what they should be.

The LSH analysis shows exactly where the errors lie. The Lightness errors are different from the Luminance errors because they are based on a perceptually uniform brightness scale compared to the Rec. 709 standard. The luminance errors are based on linear brightness compared to the expected values from measured primaries, rather than the Rec. 709 ideal.

Again, this is a preset, not a post-calibration result, which I assume would be better.

Note: I also compared these numbers against a SMPTE-C reference. It made little difference. The performance was better in some ways, worse in others, but all in all about the same.

[TomHuffman]

Kris: If you could also post the raw xyY data for the grayscale, we can get a better idea of both gamma and grayscale performance in the THX mode.

[Kris Deering]

Originally Posted by lovingdvd

Color dE aside, it seems very odd that white would have a dE of 10.7. I think this indicates the grayscale is quite off, at least at 75% or 100% or whatever level was used to measure the color.

Speaking of which Kris - for reference did you use 75% or 100% levels, and were these fields or windows? Thanks.

EDIT: Yes indeed the grayscale is off. Tom Hoffman nailed it just by looking at the hue shift in the CIE chart. Nice work Tom. I'm basing this off of what Kris posted here: http://www.avsforum.com/avs-vb/showt...3#post15325743 . Note the dE of 10 in that chart at 100% which matches the dE reported in Kris's white point reading in the .doc he posted above. So that also answers my question - the gamut measurements were taking at 100%. It would be interesting to try at 75% which has a dE of about 7 and see if the readings improve a tad, just as an experiment.

Color was done with 75% windows, that is what the program (Calman) is setup for. Grayscale was also done with windows. I used the AVS Blu-ray disc for source and a PR650 for the measuring.

[Kris Deering]

Originally Posted by TomHuffman

Kris: If you could also post the raw xyY data for the grayscale, we can get a better idea of both gamma and grayscale performance in the THX mode.

I'm such a push over sometimes.

[lovingdvd]

Originally Posted by Kris Deering

I'm such a push over sometimes.

Thanks Kris. Based on this data it yields a dE of about 9.3 - 11.3 across the range. It is about 10% short on blue across the range, about 1% high on green from 20-80 and about 2-3% short on red from 20-60. This assumes no errors with the source patterns or instrumentation.

If you have it handy, can you post the default 6500 xyY measurements for the default non-THX mode (non calibrated so we can see what that mode is OOTB).

I should have my RS20 today with some preliminary info posted tonight so we can compare.

EDIT: Also meant to add that giving it credit for a perfect Y value at 10 IRE (only data for 20-100 was supplied) the gamma for this data works out to 2.22.

[Lawguy]

Originally Posted by mrlittlejeans

So using Kris's numbers and Gregr's color calculator I come up with the following dE's

White 10.7
Red 13.2
Green 2.5
Blue 13.6
Yellow 10.2
Cyan 14.2
Magenta 8.9

I wonder how much a grayscale calibration would help those numbers. These weren't really good.

These numbers are typical of (perhaps a little worse than) the numbers we were getting using the Lumagen HDQ, which in reality did a very good job even when compared to my RS1 calibrated with a Radiance.

Still, I think that we have to be able to do better.

My RS20 has apparently been delivered to my office but I am out of town until tomorrow night!

It will be a busy Saturday.

[DIY Guy]

Are you guys forgetting to take into account that you're taking measurements from a fresh bulb and haven't factored in the lamp decay down to it's linear portion of the lamp's output?

Hopefully JVC is using a predictive model in their calibration process to account for this.

[lovingdvd]

Originally Posted by DIY Guy

Are you guys forgetting to take into account that you're taking measurements from a fresh bulb and haven't factored in the lamp decay down to it's linear portion of the lamp's output?

Hopefully JVC is using a predictive model in their calibration process to account for this.

This is a very good point to consider.

[lovingdvd]

Originally Posted by Lawguy

These numbers are typical of (perhaps a little worse than) the numbers we were getting using the Lumagen HDQ, which in reality did a very good job even when compared to my RS1 calibrated with a Radiance.

Still, I think that we have to be able to do better.

My RS20 has apparently been delivered to my office but I am out of town until tomorrow night!

It will be a busy Saturday.

Here is a reference by Tom I like to go by:

For CIELUV, a reasonable set of tolerances are 2.5 for the minimum perceptable error and 5.0 as the maximum acceptable error. For real-world performance, any display whose pri/sec colors are below 10 is quite good performance, but if you have a CMS, I'd try to get that under 5.0.

The values posted in the thread above by mrlittlejeans are based on CIELUV. It seems likely we will be able to get this closer to (and perhaps even under) 5 with custom calibration. So in THX mode we have red, blue and cyan over the 10 "quite good" threshold. This could be for Kris' particular bulb, or as DIY Guy suggests perhaps this will improve as the bulb wears in.

[mrlittlejeans]

I'm going to pick mine up from AVS tomorrow. Jason calibrated it so I'll measure it this weekend with my I1LT with HCFR and see what I come up with (time allowing). I was also wondering whether JVC was taking into account bulb aging in the auto calibration process. It seems reasonable that they would, but who knows.

Kris - do you have xyY data for your calibration for grayscale and gamut? Thanks for your feedback so far. Very helpful to those of us who haven't received theirs yet.

[pmeyer]

That's raises a good question:

Has anybody gotten a Jason calibrated projector and done measurements yet? I'm guessing not, because anybody that could do the detail color measurements themselves would likely lean away from the Jason calibration. I'm curious how close Jason was able to get.

In some ways I'm glad I don't yet have my eyeOne. I have *no* other projector to compare it to, and no way to measure it's accuracy or black levels other than 'it looks great!', or maybe the t-shirt or sock-puppet test.

I suspect I'll be much happier this way.

Paul

[TomHuffman]

One could argue that the color performance of the THX is reasonably good for a preset. I wouldn't say that about the grayscale performance.

Gamma

90% - 2.30
80% - 2.26
70% - 2.24
60% - 2.23
50% - 2.23
40% - 2.20
30% - 2.19
20% - 2.18
Ave. 2.22


This is spot on to the THX standard of 2.2.

Grayscale

ΔE (CIELUV)
20% - 10.9
30% - 10.6
40% - 10.1
50% - 9.9
60% - 10.4
70% - 9.3
80% - 10.6
90% - 10.4
100% - 11.3
Ave. - 10.4

The dE is much too high, at least double of what one would look for.



Blue is much too weak across the entire range, which would give the image a yellowish/greenish bias. I've seen considerably better grayscale results than this on other presets (the Panasonic plasma Warm mode or Pioneer Kuro plasma Movie mode, for example), but the gamma is good.

The grayscale numbers are a little disappointing. Unless you can do a custom grayscale adjustment, I wouldn't recommend the THX preset on the RS20. It really needs custom calibration.

[Lawguy]

Originally Posted by TomHuffman

Blue is much too weak across the entire range, which would give the image a yellowish/greenish bias. I've seen considerably better grayscale results than this on other presets (the Panasonic plasma THX or Cinema mode for example) and the gamma is not even close.

These numbers are disappointing. If they are typical, I wouldn't recommend the THX preset on the RS20. It really needs custom calibration.

This explains the yellow/green tinge that apparently had been reported by some as a gamut error. The gamut needs work too, unfortunately.

There is no getting around the need for a real calibration.

I just can't explain the thinking that went into not letting users tweak the THX setting.

[Hughman]

Hi Tom,

Your gamma numbers are a little funky, when I put the numbers into HCFR gamma is fairly good around 2.2-2.3. I looks like you may have entered the incorrect values starting at 30% continuing to 70%, appears you entered 20%values at the 30% window etc.

[Kris Deering]

I was going to say the same thing. According to Calman the gamma tracked nearly perfect at 2.2 in the THX mode. Unlike the RS2, this thing does track gamma almost perfectly in its presets. So if you select a 2.4, it actually is 2.4.

[TomHuffman]

Originally Posted by Kris Deering

I was going to say the same thing. According to Calman the gamma tracked nearly perfect at 2.2 in the THX mode. Unlike the RS2, this thing does track gamma almost perfectly in its presets. So if you select a 2.4, it actually is 2.4.

Correction. I forgot that I have an extra data point in my spreadsheet, so the numbers were off by one level. I revised the numbers. This is much closer to the 2.2 standard. My mistake.

[Hughman]

Originally Posted by Kris Deering

Unlike the RS2, this thing does track gamma almost perfectly in its presets. So if you select a 2.4, it actually is 2.4.

I consider the gamma tracking issue of considerable significance and is something I've been fairly vocal about. Unlike your RS2 both my RS1 and RS2 initially tracked perfectly but within several hundreds hours they both developed the declining gamma issue. Thankfully I can calibrate the RS2 to reasonably flat but at about 400 hours I can't get anything much over 2.2 with it and am concerned about what the next several hundred hours will bring. Using only the 2.6 preset without "the considerable tweaking" now provides an average gamma of a little under 2, 2.3-2.4 at 10-20% and declines signif from there.

Hopefully Darin and others will measure the gamma performance over time and report any anomalies.

[Bulldogger]

Originally Posted by lovingdvd

This could be for Kris' particular bulb, or as DIY Guy suggests perhaps this will improve as the bulb wears in.

Wouldn't it get worse and not better? I was under the impression that the bulb loses red as it ages? If the red luminance numbers are already low wouldn't that get worse?