Official JVC RS20 / HD750 Calibration and CMS thread - Page 2

Maybe Tom can sneak and give them to us!

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

Because then it would no longer be the THX setting; if you developed a Lawguy preset would you want people to be able to change it?

Not being able to copy the settings into another memory as a starting point is another matter.

My RS20 is on order and Jason will be calibrating it. At some point I may want to have an in-home calibration performed. From what I've read it seems that such a calibration is best done after the bulb has had a chance to age a bit. Is that true, and if so, is there any consensus as to how many hours of use should occur before having it professionally calibrated in-home?
Thanks.
Mike W.

75-100 hrs. of use is a good idea.

Ohlson suggested over in the 'users thread' that it might be possible for JVC to provide FW updates for the CMS, as both they and we learn how to use this new feature. This would be sweet! Do you guys think this is feasible? likely?

Feasible? Yes, the thing has a serial port, I'm sure they have a way to update.

Likely? I wouldn't bet on it. Unless the CMS is so bad that it affects the reputation of the projector enough to affect sales, the cost of managing the software updating process (along with returns for bricked projector when the update failed, etc.) would be enough to prevent them from doing it.

I would love to be proven wrong, however.

Hi everyone,
Not sure if this will be useful, but I've added some more data regarding measurements of the out of the box presets for the THX and 6500K user presets, before any calibration, in my calibration attempt thread here: http://www.avforums.com/forums/dlp-l...n-attempt.html
My HD750 is about 30 hours old now.
Still unable to tame green and cyan though...

Can someone confirm when trying to adjust for more accurate colors with the CMS, does the overall brightness suffer? If so by how much? And if you were to use an external scaler, would you see a similar brightness loss? (ie. maybe the HD350/RS10 with an external scaler is a better bet after all).

I took it to mean that if you lower the saturation of a colour you also need to up it's brightness.

Have you tried to measure the THX settings with lamp mode HIGH?!?

I get much better results in HIGH mode which leads me to the assumption that the THX calibration has been done in that mode!

Obviously you can't get it right for both lamp modes as the spectrum changes quite a bit!

This makes sense, I was told by someone within JVC who should/would know that all their stated specs, lumens, contrast ratio, gamma, temps., etc are derived with the lamp in High mode. As you've stated high mode does change the lamps native temp somewhat typically with more available red.

I wish that someone would post the approach(es) that they have employed in using these controls.

We need to memorialize these somewhere so that we can all know what works and what doesn't.

I am hopeful that there is some way that the existing controls can be used with good results.

Assuming the weather in New York cooperates and my flight isn't cancelled, I will get to work on this tonight.

I had a chance last night to begin calibrating the RS20. I have experience with custom calibration, but this is the first time I've had the luxury of an 11 point grayscale system to work with. As a result I have a few questions I'm hoping someone can clear up quickly so I can get back to dialing this in.

When you change a level at a particular % with the custom gamma control, is it normal for the system to then automatically make changes to that parameter at the other nearby levels, overriding what I already had set at those other levels?

This drove me nuts and made custom grayscale calibration using the gamma controls a very slow and challenging process, so I'm wondering if there is an easier way or if I was doing something wrong.

Here's an example. At 5% I changed the default level of 48 for R so that it was 50. This gave me a dE of 0.9 at 5%. Nice! But not for long... So then 10% needed some adjustment. I increased the R level at 10% from 104 to 109 and got a dE of 0.5 at 10%. But then I went back and remeasured 5% and the dE at 5% had jumped to 6. I then noticed that my R value of 50 at 5% had been changed from 50 to 53!

The challenge is that, when it makes these changes, it wreaks the nice dE I achieved at those prior levels I set. Maybe this isn't such a big deal in the brighter ranges like 40+%, but especially down low like 5% and 10% a slight change to one parameter is the difference between a nice dE and a way-wrong dE.

Overall the process I followed was to first use the global gain/offset controls to get the grayscale as flat as possible. This gave me a dE of less than 3 for 20-100, but with a dE of 12 and 6 at 5% and 10% respectively.

So to correct that I then turned to the custom gamma controls. I set 5%, then 10%, then correct 5%, then correct 10% and repeat this until they were both at acceptable dEs. Then I did 15%, which throw off 5%, so I fixed 5%, 10%, then had to fix 15%, then 10%, then 5% etc etc until all three where in balance. Then the same craziness when I did 20%.

Am I doing something wrong here? Is there a better way?

Although once it was complete I had a dE of 0.5 to 2.3 from 5-100 IRE, I was really expecting to be able to get closer to a dE of 1 or less across the entire range (and to be able to do this much more quickly). Seems like I could have easily and quickly if it wouldn't keep changing these parameters on me!

I would greatly appreciate any tips and advice. Thanks!

OK so after hours of tweaking the grayscale it was very late and I didn't get much time to work on the CMS. But after all the chatter hear about it I just had to at least take a quick look at it.

The first thing I did was to dial in Red. This seemed to go incredibly quick and easy, although I did not check what impact if any it has on the other colors. Anyway, I lowered Red's saturation way down and was then able to bump up its brightness (to compensate for what the saturation control did) right on target.

Then I turned my attention to green which did not go as smoothly, or should I say, it just didn't go period. I turned the Saturation for green all the way to -30 (lowest setting) and it barely moved down toward the white point at all.

I'm still confused at how that could be. I'm actually not striving for perfection here with Rec 709. Rather, I just want something fairly close and way better than OOTB settings. And I cannot use THX mode (even if that gets us close) because it offers no custom gamma control.

From this quick experiment things didn't look too promising with regards to green. However I am optimistic that based on what Darin and Kris and cine4home and others had posted here it seems the custom CMS can offer an improvement, even if not perfect.

Can someone that's had some level of decent progress with the CMS comment on the procedure they've used to do this so far? After seeing green not really budget I'm at a lost for where to go from here. I have a lot of experience with the CMS on Sharp projectors and the RS20 certainly isn't operating like that CMS (would sure be nice if it was!). Thanks!!

Welcome to the world of calibration. It is typical that one adjustment will affect surrounding readings if not all readings. The multistep adjustments, including just cuts and drives are processed and interpolated into the the LUTs. Unlike analog, every step of luminance from each color is discrete, driven by a LUT. This is also why you might make 2, 3, 4... click adjustments with very little change, then a big change with the next click. Think of a 256-step grayscale (0-255) and having 10 or 11 points to adjust its uniformity, there has to be some mathematical formula to uniformly adjust/average the points between.

I have spent entire evenings, tweaking the internal curves on my RS2 trying to get grayscale and gamma correct, using a combination of the R, G, B & W curves, finally settling on flattest grayscale. Trying to correct gamma only, after grayscale was flat, resulted in a horrendous grayscale. Fortunately, I have the Radiance to further correct grayscale and gamma.

I predict the RS20 will be sort of a "poor-man's" (for lack of better words) solution to have a good system with grayscale, gamma and CMS. The RS20s performance can probably be greatly improved with the addition of a Radiance and still be cheaper then any other option with similar performance. I am assuming the RS10, just as the RS1, with a Radiance, to be a less than equal performing system, due to poor black level performance. However, the Iris in the RS10 could make a big difference and go a long way.

I would be interested in hearing what a few of you think a "package" (RS10 and RS20 + RadianceXE (HDMI 1.3 version) + pre-calibration) should/could sell for. Please PM me with your thoughts...... Obviously I can't put this package together below dealer cost and nothing for calibration. The RadianceXD will still be available ($1K less) and the package also without calibration.

Thanks Glen. I didn't get to the point of adjusting White in the custom gamma yet (since I spent all night on the grayscale!).

Do I assume correctly that one uses the White point to increase or decrease Y at specific levels (i.e. 5%, 10% etc) to put Y on target reference for the gamma?

For instance after I was done tweaking the gamma RGB my gamma curve dropped from 2.2 to 2.17. In looking at the curve I can see that 5-30% is raised a bit causing the 2.17 gamma, and I'm sure its not a coincidence those are the levels that I had made my RGB gamma tweaks in.

But from what I gather from your post it sounds like the necessary bump I'd make to the white point in the gamma to adjust for this will likely cause the grayscale to then be off again? Sigh. I think I read about that as being an issue with the JVC gamma control implementation. Perhaps it is something they improved with the RS20...

Apart from the above issues just to add a little more gloom to the party on my RS2, for some unexplained reason, when adjusting gamma white the red, green and blue don't adjust at the same luminance level. For instance if I set it at 50%, only the red and green actually are adjusted at 50%, blue gets adjusted at 60%. Blue also suffers from this when adjusting by itself, whatever % luminance level I intend to adjust has to performed at the 10% lower luminance level. I pity the fool who attempts to use these adjustments after a glass or two of wine, it can get messy very quickly.

I guess that's the best tip I can provide anyone using this gamma control.. calibrate sober.

That is about all the White points can do, but they change RGB levels to do that.

Well, nothing works like we would want. When adjusting grayscale, you should also consider what your needed gamma corrections need to be. If Y is too high at a adjustment point, you dont want to correct grayscale by adding/increasing any of the RGB points.

All these adjustments are like a complicated tax return..... change one number and you nearly cause a change every line on every page of the tax return. Then you find another change...... this was murder when doing manual tax returns .... before computers

Doesn't the wine add a sense of tolerance, allowing you to adjust, and adjust, and adjust ........ "reset", then adjust, and adjust, and adjust ....... "reset" again...... then repeat...... I've done it without the wine..... and really don't want to have to do it again...., but we never get what we want....

Hey Glen

I know how you feel there. I've spent probably 12-18 hours of calibration time alone on my RS2 in the last week. Had to take advantage of the PR650 while I had it up here. That is one frustrating projector to work with and in the end it is still nowhere near what I would like it terms of overall accuracy but boy does it throw a spectacular image now. I guess if I wanted pure accuracy I would own a Samsung, but that would mean giving up a lot in actual perceived image quality.

What would you give up, except better black levels? And yes I know this can make a big difference.

The Samsung I think would be just as good/superior in most every other area of an image quality standpoint... no?

And before anyone posts, SOWK is bashing JVC again, I'm not bashing, just honestly asking.

It's not a matter of giving up, it's a matter of getting the best picture desired.

I wouldn't give up high native contrast ratio
I wouldn't want to give up a 3-chip design
Black levels can be achieved with gray screens, larger screens, ND filters, but nothing can replace high native contrast ratios (by nothing, I don't mean not throwing serious dollars for a solution)

Again, A RS2/RS20 with a Radiance is a sweet configuration with absolute REC709 colorspace, flat grayscale and uniform gamma. What more?

I'm not sure if the wine adds tolerance as I don't have a baseline to compare. I don't think I want to admit to myself how much time I've spent obtaining a suitable gamma and greyscale especially in the sub 20% range but finally last weekend I pulled a half-nighter and got it to the best it's been. Now I'm scared to turn the on the PJ. Maybe I can squeek out that last 1%, oh crap.......reset.

Why would you prefer a 3 chip design?

That only calls for more "issues", at least in my thinking.

Kris,

Darin has commented a couple times since the RS20 comparison that you've tweaked your RS2's color gamut and it looks very good, are you using an outboard processor or have you discovered something worth sharing?

You are now the second user to report this.

http://www.avforums.com/forums/8382069-post30.html

TomHuffman, he was able to do better. Check post 48, http://www.avforums.com/forums/dlp-l...attempt-4.html

Correct. I think the trick is, as he later went on to post, that you have to take the color control down a lot (-30 I think he said), and then use the CMS controls to compensate. With hindsight this is probably what cine4home's early remarks were about when he talked about first calibrating overall saturation, you think?

Hi guys, I guess it's easier to follow the discussion here. Yes, I did manage to get a better result (I think) but it would be nice if some could confirm I'm heading in the right direction.

LovingDVD, I can confirm that it's really annoying when playing with the gamma steps, it does affect the neighboring ones. I found out two things:
1) a higher gamma makes it MUH MORE difficult to flatten the greyscae (at least for me), so I think I'm going to stick to 2.3 maximum (as a reference) which is probably going to end up as an average of around2.2 anyway.
2) When you look at the uncorrected RGB curve, try to minimise the points you're trying to correct. Initially, I tried to go step by step, and it just drove me mad. Then I started again, and tried to corect the big bumps first, to lower the average dE, then work down on the smaller bumps (if that makes sense). It may be a well known approach, but as I said, I'm learning...

Regarding going down to -30, it was a bold approach to test the theory (go much further than recommended to set the red at 21%, and then try to correct each color individually with the brightness control.

I have no idea if there is an inherent flaw in this theory, and someone more experienced may have a look at the uncorrected gamut and know instinctively how much to push the color and tint to get the right starting point for the primaries/secondaries, and then improve from there.

But it looks like the right approach at the moment. Before adjusting the brightness, I had all the primaries (including green) spot on, and only cyan outside. Of course Y was all over the place.

I am going to start again from scratch when I have a bit more time, following Tom's excellent guide and using the table for xyY in REC709 to try to get closer values (in my last attempt I didn't have the right percentages for Y and I just did alot of guesswork to bring dE down). I know, I should have started from there but I found the "calibration for dummies" at my level for greyscale, it was just not detailed enough for a proper CMS system...

From what I understand from Ekkehart's very vague descrition, he seemed to think that the approach to bring color way down to bring the primaries and secondaries inside the gamut, and then pushing them back towards the right point was the best approach (that's from the translation, my German is not good enough).

Hop this helps....

Here are the CIE94 numbers for the THX preset.





Here are the CIE94 numbers on Mann01's latest "improved" technique using the CMS.



This is no better than what was available with the RS1/2.

Not an encouraging result. Other than red, none of the colors are within acceptable limits, and some of them (green, yellow, and cyan) are grossly inaccurate.

This is the result, despite the fact that the 1931 CIE chart looks reasonably good.



This just shows the inadequacy of relying exclusively on the CIE chart (especially the 1931 version) to grade color performance.

I haven't worked on one personally yet, but from the data of those who have it is beginning to look like the functionality built into the RS20 isn't a true CMS at all, but a pseudo-CMS with limited (if any) value.

You can select the THX preset, which gives reasonably good color but a quite yellowish gray scale; or, get a good gray scale with a custom calibration and live with the poor OOTB color. The included CMS controls don't seem to be able to improve it.

There may be very good reasons to like the RS20 (better optics in particular), but unless something new comes up we haven't seen yet or JVC offers a fix, better color performance isn't one of them.

I'm sure that this statement will annoy some of you, but that's what the data shows. If we get better data, then I'll be happy to revise this judgment.