Official JVC RS20 / HD750 Calibration and CMS thread - Page 29

Us DIY calibrators have this instinct to maximize the lumens at the optimum grayscale by maxing out the controls. I suspect that this projector requires a 'though' compromise to get it exact.

Trying to get things exact in consumer products is not going to work for any product when you really examine what is going on. It is even more unlikely if you do not have tools that are capable of measuring to the tolerance that you are targeting. What you need to do is focus on the parameters that define the best picture quality and make your compromises where necessary.

Last night I accomplished two things specifically with regards to the CMS: 1) I fixed an issue with cyan being too bright, and 2) I created an entirely new CMS calibration the opposite of how I did the last one, to see how the two would compare.

The Goal

My CMS Setting 1 was created by keeping color at 0 and using 100% patterns to calibrate, then fixing the Y at 75% so that it was accurate rather than overblown, at the expense of Y at 100%. This also caused 75% to be a bit more saturated than my target gamma. With this approach I attempted to balance the dE fairly evenly (split the difference as much as possible between errors at 75% and 100%. This looked great but cyan was running too bright. I resolved that last night.

When I asked Tom whether I should try and split the difference between errors at 75% vs. 100%, or whether I should try and favor one level, he understandably recommended I give priority to 75%. So with this in mind I wanted to try a new approach designed to do so. This lead me to CMS Setting 2.

The Approach
My CMS Setting 2 was created last night completely differently than CMS 1 in an attempt to give priority to 75%. With this approach I set color to -10, and calibrated things to be nearly spot on at 75%, with low values for Y at 100%. But unlike earlier calibration attempts, I was able to keep the xy mostly near the same points at 100% as they are at 75%, with the exception to green. With CMS Setting 2 I had to increase the green gamut point because otherwise 100% would be too under saturated. This was a bit undesirable but alas this is also about compromise and balance. Also I had to reduce Y for cyan below spec even at 75% to stop it from being too bright in general.

Essentially when I was done with CMS Setting 2 the measured gamut at 75% was nearly spot on for Rec Ldvd and when measured at 100% was close to the xy of Rec 709, with some points being slightly undersaturated mainly green, and Y values being a low pretty much across the board at 100%.

Initial Testing and tweaking
The first thing I noticed when testing CMS Setting 2 was that it was not clipping at all, unlike CMS Setting 1. So that was a favorable start. I did find it difficult to find a proper balance for green brightness. Initially I had it a bit down from its target at 75% in order to help it not get too under saturated at 75%, but the slight dimness was apparent as a result.

So to compensate I extended the green gamut point up, allowing me to make it brighter at 75% without under saturating it too much at 100% relative to its Rec 709 point. This improved things, but it was still a bit dark. So ultimately I pushed the brightness for green up a few more clicks and subsequently have undersaturated green a bit more than I would like at 100%.

How It Looked and Compared
I was pleasantly surprised to find that CMS Setting looked surprisingly close to THX mode +0 Color in general, except for a few things.

CMS 2 is deeper in green and red, but this of course is intentional. So far so good.

CMS 2 was a bit washed out in cyan. No doubt this was the result of having to tone down the brightness. I'm going to see if some tweaks will help with this.

On a 5% color scales pattern you can see that at 100% R, G, and B is way wrong. This is a well known side affect and the area that I decided to sacrifice in the making of CMS 2 in return for good linearity from 95% down. Per Googer I am quite sure I could drop my contrast control if I wanted and at some point 100% would fall in. I may try that quickly just as a test but would never run with a lowered contrast control.

Also CMS Setting 2 did not look quite good enough on my Disney's "Up" movie preview test (start Wall-E, watch previews). The house shown in the previous is sunlit in the sky. In THX mode the side of the house has a yellow-ish gold tone to it which I believe is correct and looks pleasant. With CMS 2 this was too washed out looking - not enough color in there.

Instead of looking yellowish/gold it looked more like pale-ish yellow. Interesting CMS Setting 1 suffers from the same thing. I'm not sure what aspect of my color controls is causing this, and I could not tweak any individual CMS setting to compensate for it. With CMS 2 Yellow is spot on at 75% and a bit under saturated at 100%. With CMS 1 Yellow is spot on at 100% and a bit over saturated at 75%. Yet both CMS 1 and 2 show this scene the same way.

The only way I could compensate for this was to raise the color control from -10 to somewhere around -4ish and as high as 0. With the color control now up, this scene looked virtually identical in THX mode as it did in CMS 2 and 1.

I did not check it, but I am quite sure that raising the color control like this on CMS 2 is causing some clipping that is not there with color at -10. But once again, this is a sacrifice needed to get all the tones in line.

Comparing CMS 1 to CMS 2 is very interesting. Overall preliminarily I like CMS 2 with color between -4 and 0 better than I like CMS 1 with color -4. CMS 2 w/ color provides a good approximation of Rec 709 with some push and vividness, which is what I was going for. There will likely be some scenes where green looks a little to neon-ish, but hopefully those will be fairly rare. If not I may take green back down a bit.

CMS 2 just looked a bit more "pure" than CMS 1. Meaning a bit more THX-like, while still retaining some vividness.

Next Steps
Once again, this calibration (including redoing the 5-100% grayscale that had drifted quite a bit) took up all the time and it was too late to experiment with much real world material.

At this point I am done creating new CMS settings and feel I have exhausted all options and combinations, short of the possibility of a custom gamma that perhaps can help. So the good news at least as I am done with the extremely time intensive task of trying to balance everything out and can finally get down to some viewing.

I have CMS 1, 2 and default OOTB color set to my User 1/2/3 preset, so it makes A/B testing very easy. If I happen to see a scene that has a significant error, I'll check it in my other CMS setting. I'll repeat this taking note of which setting has less and more tolerable errors than the other and pick that. I will then post my final CMS settings so you can try this at home.

Conclusion
The reality is clear. This is not a full featured CMS, and while it does offer some flexibility and better than nothing, using it requires careful balance and deciding what type of errors you will deal with and making the best of things overall given what we have to work with.

Although my opinion may change after some extensive real material testing across my modes, currently my recommendation is to calibrate the gamut you want at 75% using color of -10 and get it right at the proper xyY, then balance this by measuring with 100% and ensuring you do not grossly oversaturate any colors (red in particular is easy to throw way out of wack) at 100%, and balance to compensate but weight things so that 75% is favored. Then watch some material and increase the color control from -10 to as much as 0 depending on your preference (but keep in mind you will be clipping).

Comments and questions always welcome. I'll keep you posted on how the testing goes.

Here are my measurements with my lux meter.

278 centimeter wide screen
5 meters from Lens->Screen
vertical position of HD750 somewhere in upper third of screen mounted,
thus lenshshift down
Lamp mode high

Iris 0 : 33000:1 186 Lux 807 Lumen
Iris -8 : 40600:1 137 Lux 595 Lumen
Iris -15: 45200:1 97 Lux 421 Lumen

Objektive on Tele/minimum Zoom: image width 176 cm

Iris 0 : 40000:1 381 Lux 662 Lumen
Iris -8 : 47600:1 275 Lux 478 Lumen
Iris -15: 51500:1 183 Lux 318 Lumen

Objektive maximum zoom: image width 357cm

Iris 0 : 30000:1 125 Lux 896 Lumen
Iris -8 : 37200:1 98 Lux 702 Lumen
Iris -15: 42000:1 72 Lux 516 Lumen

Personally I use iris 0 all the time, as I don't feel the gain in black level outweighs the massive loss in brightness. I mean even with iris -15 and with 45.000:1 the black level is still so high I easily spot the difference between screen and masking. Maybe I got owl eyes who knows ;-)
On the other hand 97 Lux or roughly 9 footLambert for whitelevel approaches the point where I miss the punch in daylight scenes.
With >180 Lux I have enough headroom for lamp aging.

Before the HD750 I owned a Dreambee/HD1. On its mounting position I got roughly 650 lumens and 14400:1 on/off, on max zoom I got 730 lumens and on tele position up to nearly 20.000:1 contrast.

So compared to it, I got now 150 lumens more light output, half the blacklevel, a subjectively great step forward in sharpness and much better customisation possibilities with regards to gamma and CMS.

My unit came with precalibrated color temperature(s) on Custom 1 & Custom 2 settings. For lamp high mode the red and blue offset were on +2, I quickly noticed that raising blue from +1 to +2 visibly heightened the black level. Roughly I can cut my on/off by one third to around 22000:1.

After watching some movie scenes with offsets dialed back to 0, I noticed that skin tones in dark scenes looked not as good as with the custom setting.
So I dialed in the compromise of red+1,green-1,blue+1. To the naked eye this does not have a relevant impact on blacklevel but I'll have to remeasure this.

For color I dialed in main color setting to -15 and dropped the saturation of red,yellow and green to -10. Watched both of the Narnia movies yesterday and skin tones looked very good to me.

My custom gamma settings came already prebumped at 5% from my dealer and I like the better shadow detail quite a lot. The drawback is with material that has heightened black level, as the raised lower IREs enhance the poor mastering. For that I have another custom setting with a standard gamma of 2.2 for a quick remedy.

Other nice enhancements over the Dreambee are the missing bright corners and the blue halo around the screen. Also it is much quieter.
This time I also payed attention to the convergence of my machine before I packed it up at my dealers place and it is within .5 pixels. My Dreambee had a while pixel offset on the right side, which means I could not correct it. The manufacturers (DreamVision) statement was that they saw no issue with the convergence. :-(

Another thing I noticed is the lesser red/green fringing on moving high contrast objects. It is still there but to a lesser extent. Some scenes that looked awful on the Dreambee are completely fine now.

All in all this is a fantastic machine, and because I made a swap-in with my dealer for my Dreambee it wasn't even that expensive.

regards,
Gregor

I have the DVE Basics (don't have a standalone HD-DVD player, only on HTPC) and I can confirm there are no color ramps unfortunately...

I had a look at my 4th attempt settings using the internal ramps of the 750 (only greyscale and primaries, 32 segments). It looks okay for blue and red, but green (surprise, surprise) completely clips about 7 segments from the high end of the scale and ends with a paler segment (as if it lacked brightness). It doesn't clip at all in THX (although the greyscale ramp is quite obviously off, but it looks like my settings also slightly off now (bit too red, my lamp has probably drifted a bit since my last calibration)

This is the first objective flaw I can report regarding my settings. I'm sure more are to come (although I watched chapter 4/5 in Spiderman 3 and couldn't spot the moment described by LovingDVD). Any chance you could provide a timecode LDVD?

I knew I shouldn't have looked for it. Ignorance is bliss.

Great recap and great work LDVD.

I'm probably going to try your approach as it looks like the right compromise for me at this stage.

That is once I have resolved my contrast mystery.

It might be a bit later on in the movie. I can't recall exactly its all a blur at this point!

There is some scene where it starts out with the camera panning across various masks, then you see the villain step out of a glass chamber with green bellowing smoke.

Anyway to be clear, I reported noticing this with my then-current CMS settings, which I then later redid in such a way that eliminated the issue. So depending on how you did the CMS, you may or may not experience this. It is an easy mistake for an RS20 CMS beginner to make so it is a good way to see at a glance whether a calibration is too bright in green. Of course a check of xyY will do too.

BTW I've found several good sources to look for problems with color brightness and color errors in general. One is that SpiderMan scene. Another is the preview for the movie "Up" at the beginning of Wall-E. Lots of beautifully colored balloons you can oohh and ahhh at. Another is the show Sunrise Earth on discovery. I have it on my DVR so I can easily check it. They show a lot of sunlit grass and trees, skies and other such colors good to check. Also the food channel in HD since you can easily recognize lots of objects and know what the color should look like (bottle of ketchup/mustard, fruits etc). Also video games like Little Big Planet use a lot of cyan and magenta etc. Happy hunting!

I found the answer to my question here http://www.avsforum.com/avs-vb/attac...8&d=1225648932 (I had missed this link which describes the content of the disc. See end of doc for notes re %).

So the 75% color patterns on the AVS rec709 dvd are 75% stimulus, for those who may be interested.

Thanks very much, I've seen the scene (indeed beginning of chapter 4 or 5, blur for me as well), but nothing hit me (I may have experienced beginner's luck with my settings). I'll watch it again, along with the UP scene, and I'll report.

Thank you very much for that informative reply Gregor!

So far the ratings for the light out put seem to be holding up pretty well in the new
JVC models.

Thanks!

Thanks, I suspected as much. My screen is 1.3 gain

Good news for everyone without an HD-DVD Full DVE version.

Here is Alluringreality's answer to one of my questions re ramps on the AVS REC709 patterns disc. It looks like V1.2 - to be released within a week - will include full patters with 5% steps for all primaries and secondaries.

http://www.avsforum.com/avs-vb/showt...8#post15502388

@ Lovingdvd:
could you attach your .chc files with the last calibrations?

Good point as always. Further to this remark, have you noticed that when you switch between the possibly hardcoded modes and the others, there is a noise as well as the pause, which suggests something is going on in the optical path as well.

this made me think that maybe there is a motorised system that puts a FILTER in and out of the way, to further correct the gamut and bring evrything within reach.

This means that if cinema1 has such filter in place AND gives full access to greyscale, it main indeed contain the answer...

I'm trying it right away!

I remembered the noise from ootb, and it is indeed the iris, I've just checked that. When iris is the same, there is no noise.

Hoewever, there is no pause either when moving to/from cinema1/thx and the other modes, can you check your iris settings are the same?

Also, the CMS is not accessible in cinema1, so even if the gamut is closer, how would his help us to get it right?

I've set iris to -15 on all the modes to make sure this doesn't play, and it turns out that on my unit, the mode with a pause (and access to CMS) is cinema2, not 1... Could you confirm if it's the same on yours?

This is very interesting.

Has anyone measured the Cinema 1 gamut?

I suspect that some software program exists that gives true access via serial cable to the THX and perhaps Cinema 1 modes.

I have, and unfortunately it's completely off. Could you confirm if you can access the CMS in cinema1? I only can in cinema2 (which is also off). I've check linearity at 75% and 100% in both, and there is no improvement.

I'm still experimenting and will post more info later...

I am in the office so not until tonight.

Are you saying that neither Cinema 1 nor Cinema 2 track well?

Yes

Thanks for clarifying. I will post more info later tonight (including preset tests numbers) but I may be onto something re another lead and I need a bit more time to write down my thoughts. In a nutshell re these questions I think we experience the same behaviour on both our PJs (pause with THX and cinema1), I was just confused because I thought you had access to the CMS in cinema1 (which I don't have, only to cinema2, just like yourself). The gamut is defintely off in all modes but THX, and none ofthose I have tested track at 75% and 100%. I have done a general reset of my PJ in service mode before testing to be sure it was as close as possible in ootb state. More info in couple of hours...

Yes I can confirm this is the behavior with mine as well (at least for THX mode, haven't ever tried cinema).

One thing that may be causing some confusion among posts regarding this behavior is that the "blanking" only occurs when going OUT of THX mode back to one of the other user modes. When you switch from a user mode to THX its not quite as smooth a going between user modes but certainly doesn't blank like coming out of THX.

At any rate, it is a virtual certainty at this point that THX mode is accomplished using special internal processing and parameters that we do not have access to through the regular menus o,r the service menu we have access to...

After making some optimizations to my CMS Settings 1 and 2 last night and performing extensive A/B testing between these two settings, THX mode, and OOTB colors I am happy to report that I have what I believe is a winning combination! And the winner is - CMS Setting 2. For an explanation of how these settings were deprived, how Setting 1 and 2 differ, and what the point of all this is see my previous posts in this thread.

As previously discussed, the Rec Ldvd CMS settings are designed for those who find Rec 709 (THX mode) too pale for their tastes, yet find the RS20 OOTB settings way too oversaturated.

I'm now getting great results (albeit with some unavoidable yet seemingly subtle errors), at least as it applies to my particular tastes. YMMV. With Rec Ldvd the colors look VERY similar (almost identical) to THX mode in many scenes, especially scenes that do not use strong green, red or blue.

In scenes that use strong green, red or blue, you will see some slight push in blue, and a bit more push than that in red. This is intentional. Scenes that use strong green will experience a modest amount of push. This amount unfortunately is a bit more than I would like, but was unavoidable to in order to keep green from completely falling apart at 100%. I may try tweaking this further down the line if some refinements are possible, but would anticipate any improvement to be barely noticeable (without having to destroy 75%) so this is low priority ATM.

In summary, on a saturation scale of 0 to 10, with 0 being the saturation level of THX mode, and 10 being the saturation with OOTB colors, I'd put Rec Ldvd at about a 3.5 which was my goal all along.

Essentially you should find that Rec Ldvd provides excellent color rendition and tracking for everything except the when colors are at their brightest (near 100% level). Colors at those levels will either clip or show as undersaturated.

I watched nearly 4 hours of various material and clips last night to QA the results. This includes cable TV, Blu-ray's including animated and non-animated movies, and PS3 gaming with lots of A/B testing.

In all this time I only came across a couple of scenes where the color errors were visible. The first was in the preview for the movie Aliens Vs. Monsters available when starting up Kung Foo Panda.

At the beginning of the clip a guy is getting scanned by a green light beam (I looked at the 3rd scan - the one of his elbow). In THX and OOTB modes, when paused you can see that the light bean consists of a glowing rod of green light, and an inner, brighter green light. When watching in Rec Ldvd the inner bright green light was not visible - instead this area looked a bit washed out and blended with the surrounding green glowing light. Clearly this green was at or near 100% and it was understandable and unavoidable with this set up (and a trade off that Rec Ldvd is based around).

The other scene I saw was on cable - it was a close up of a very bright explosion that had lots of yellow and red tones and nearly blinding white light dominated much of the picture.

In this scene, when paused in THX or OOTB mode you can see a bright yellow halo-like thick edge/border around the outer area of the large white explosion. With Rec Ldvd settings the yellow was still visible but was very washed out looking - again, consistent with errors at or near 100%.

I'm sure there were some errors like this in the other material I was watching, but most importantly nothing jumped out at me. And I only noticed these because I specifically paused to stare at it. So in general although there will be color errors and some clipping in certain scenes, these will largely go unrecognized. Some detail may be lost due to clipping occasionally as well, but again, if I can't spot it then it won't bother me.

One other thing I should mention is that the green shown on "The following preview is rated..." type of screens before a movie trailer will show green more oversaturated than I'd like. It is still better than the OOTB setting, both in terms of hue and saturation, but it is a little strong relative to Rec 709. Fortunately green for the most part in most program material looks much better, and in many cases only looks a bit more vivid when compared to THX mode.

Also I am happy to report that in no cases did I see any color that was just plain wrong. In other words no light/off-blue looked purple-ish etc when comparing Rec Ldvd to THX mode. The same of course cannot be said when comparing THX mode to OOTB, which goes back to why I do not care for the OOTB colors and the reason for Rec Ldvd (some mild oversaturation, but without going too far).

OK, so without further adieu, here are the settings:

- First, make sure you have a flat grayscale. These CMS settings were derived using a grayscale that tracked with a dE of <3 for 5-100%.

- I am not sure how relevant it is to the CMS, but I did this work with a custom gamma curve with the gamma correction set to 2.3 and values tweaked along the curve as needed to fix the tracking below 20% as well as to improve 70-95%. This measures to a 2.2 gamma. If you do not have the capabilities to set this up do not worry about it as it may make only a marginal difference (tho I can't be sure).

- Set the Color control to 0

- Set the CMS settings as follows:

Color, Hue, Saturation, Brightness
R: -2, -15, 9
Y: 3, -25, 15
G: -10, -25, 13
C: -1, -30, 6
B: 0, -10, 15
M: -1, -12, 10

- Reduce the color control to taste, if desired. I like it right at 0. However if you find Rec Ldvd to be a bit more over saturated than you'd like, drop color down a bit. This will also help remove clipping (generally most clipping is gone by -4ish) and help tame green a bit more, but keep an eye on your sky-blues and yellows (in particular) which will start to flatten out with a reduced Color setting to actually be undersaturated compared to THX mode.

These settings are the work of approximately 60 hours of experimentation, redos, and tweaking to wee hours of the morning over and over, and is based on the input, theories and feedback that many here have shared for which I am very thankful. I hope you enjoy these and only ask that if you repost these settings elsewhere you provide credit to AVS forum members, and that one day Chris takes me for a spin in his toy. Of course if I had to do it over it would take a lot less time with the hindsight of everything learned, but surely I hope I never have to do so!

We can't be sure at this point that CMS settings are transferable from machine to machine, but unlike grayscale tracking, I anticipate it should be (as an example I think we all have the same CMS settings for preset modes like Natural etc, although I don't think we ever verified this). If not, then I previously outlined the approach I used, and keep in mind the approach for CMS Setting 2 was the clear winner.

Also I should close by saying that while I am finally happy with the colors, it is not a substitute for a fix and/or better implementation to the CMS which is still sorely needed. I view this as a temporary (hopefully) work around.

The fact that I like colors a bit over saturated played well into the balancing act made possible with these settings, but for those looking to dial things in right at Rec 709, I think it will be even harder to find a good balance.

Please be sure to post your impressions on how these settings look on your RS20 and whether you change the color control or prefer it at 0. I look forward to your feedback and suggested tweaks to the settings are of course very welcome and encouraged. Enjoy!!

Lets hope the CMS gets at least a decent upgrade.
Has anyone viewed the RS20 corrected through a Radiance? It would be good hearing impressions about this to understand how much better the RS20 can become with a better CMS.