Official JVC RS20 / HD750 Calibration and CMS thread - Page 25 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #721 of 1782 Old 01-06-2009, 11:35 AM
AVS Special Member
 
JeffY's Avatar
 
Join Date: Mar 2001
Location: England
Posts: 3,667
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Lovingdvd, looking at the chart I think you have yours pretty close to the way mine is setup now.
JeffY is offline  
Sponsored Links
Advertisement
 
post #722 of 1782 Old 01-06-2009, 11:48 AM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by TomHuffman View Post

Thanks for these numbers. This is a textbook example of why I think CIELUV is a poor metric for pri/sec color grading. If you use CIE94 to evaluate these exact same values you get.

Code:
                   
       R      G         B        Y        C         M       W
x     0.6620   0.2960   0.1480   0.4370   0.2150   0.3190
y     0.3310   0.6630   0.0620   0.5360   0.3320   0.1480
Luma   53.9    167.1    21.3     237.7    200.4    80.5     311.6
CIE94  6.0     10.3     1.5      8.4       7.1     2.5
L     -8.6%   -10.8%   -2.7%    -7.4%    -7.7%    -4.0%
S     7.2%    14.3%    -1.2%    12.7%    11.3%    4.7%
H     0.2%    0.3%     -1.2%    0.5%     -2.7%    -1.1%
These results are almost exactly what one would expect from an uncorrected RS1/2. Red, Green, Yellow, and Cyan are all grossly oversaturated. This is masked in CIELUV because the lightness is lowered across the board. Because CIE94 scales lower, you have to more than triple the values to get a sense of what the error is relative to CIELUV.

Thanks for the feedback Tom, this is all still very much WIP as we figure out how to deal with the CMS as it currently is. I do think you overstated how far off the colors now are though - while still oversaturated to a moderate degree, they're pulled much further in than OOTB settings, and the bit of lowered lightness that's associated with my current settings subjectively mitigate the still too-high saturation. Going back and forth between my settings and the THX preset, in many real scenes it's actually very difficult to see a difference at all, other than some instances of red appearing a bit stronger with my settings and greens being neutral with my settings as opposed to visibly yellow with THX (which I posted about earlier). More importantly, however, I haven't seen any instances of colors looking grossly out-of-whack at any brightness with these settings, which is still rather unique from what anyone else has found with custom CMS settings where they've calibrated to absolutely minimize error at a single level, and found out after the fact that doing so introduces huge errors at others.
Googer is offline  
post #723 of 1782 Old 01-06-2009, 12:19 PM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by lovingdvd View Post

Has anyone spent some time playing around with the Sharpness and Detailed Enhancer settings that are under the Advanced setting? If so have you found these helpful? How do these two controls differ from each other? Any optimal settings you've found?

As a side note I find it interesting FWIW that the OOTB config has Sharpness set to 15 and Detail Enhancer set to 5, whereas the default setting in the CUstom1/2/3/ has Sharpness at 5 and Detail at 0.

They appear to be very similar to one another but operate mostly on different frequencies - Sharpness mostly on lower frequencies and Detail Enhance mostly on the highest frequencies (i.e., single-pixel). The effects can be clearly seen if you turn them to their upper ends. Sharpness introduces large (3-4 pixel-wide) halos around objects, and detail enhancer is much finer, working more on the order of single-pixel sharpness enhancement. As such, I think Sharpness is best left at 0 for basically everyone, but mild use of Detail Enhance (i.e., no higher than 15 or so) could be seen as desirable by some. YMMV of course.
Googer is offline  
post #724 of 1782 Old 01-06-2009, 12:24 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by lovingdvd View Post

I had a chance to play with this exact approach last night. I'm happy to say that, preliminarily, the results looed very good (also as you have found).

Glad you're happy with your settings! As I said, mine are not perfect number-wise, but I'll only know they can be better when I'll find a way to improve them (ie nothing really annoys me in the picture, whcih I can't say for OOTB or THX untweaked.

I will definitely try your approach (try to get as close as you can without touching the color control, which mean there is no need to overcompensate with brightness). Although I plan to try "tweaked THX" as well, I am not convinced I'll be happy either with my greyscale using only color temp adjustments (unless I'm as lucky as JeffY), which means I won't be able to leave gamma adjustments behind.

Did you by any chance try to measure at 75%? As you're not touching the general color setting, and are probably not overcompensating brightness, you may have a better linearity than I (or LG) got when we were trying to hit rec709 perfect with a lower color setting.

Could you, just out of interest, post your settings for the gamut you've posted? I'd like to see if I like the slight oversaturation of "recLDvd", plus some others who have tried mine or LG's settings may want to try yours and see which they prefer (more choice is always good!).
Manni01 is online now  
post #725 of 1782 Old 01-06-2009, 12:26 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,357
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 15 Post(s)
Liked: 204
Quote:
Originally Posted by Googer View Post

I do think you overstated how far off the colors now are though - while still oversaturated to a moderate degree, they're pulled much further in than OOTB settings, and the bit of lowered lightness that's associated with my current settings subjectively mitigate the still too-high saturation.

I wasn't offering any opinions on this, or at least I didn't intend to. I simply made the observation that these numbers are objectively virtually identical to what one would get with a RS1/2, especially with the Color control turned down a little: saturation down only a little, luminance down substantially. If you subjectively like what you see with this, then you would have liked what you would have seen with the RS1/2, color-wise.

My point was that if you leave the saturation control alone and count on the main Color control to lower the luminance of the colors to hide the oversaturation (robbing Peter to pay Paul it seems to me, but that's another story), then the CMS becomes pretty much a moot point.

Tom Huffman
ChromaPure Software/AccuPel Video Signal Generators
ISF/THX Calibrations
Springfield, MO

TomHuffman is offline  
post #726 of 1782 Old 01-06-2009, 12:33 PM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by TomHuffman View Post

I wasn't offering any opinions on this, or at least I didn't intend to. I simply made the observation that these numbers are objectively virtually identical to what one would get with a RS1/2, especially with the Color control turned down a little: saturation down only a little, luminance down substantially. If you subjectively like what you see with this, then you would have liked what you would have seen with the RS1/2, color-wise.

My point was that if you leave the saturation control alone and count on the main Color control to lower the luminance of the colors to hide the oversaturation (robbing Peter to pay Paul it seems to me, but that's another story), then the CMS becomes pretty much a moot point.

It's a tradeoff that's looking to be relatively close to the best one can do with the CMS. Yes I can pull in the points more by turning down the Saturation controls more, but this pulls down the lightness so much that it then becomes necessary to (substantially) increase the CMS Brightness, which as we've already found, causes color tracking across different % stimuli to completely fall apart.

Anyway, current settings for these results for anyone that feels like trying them (before I go in and play around some more ):

Color: -10
Tint: 0

CMS (H, S, B):

Red: 4, -10, 0
Yellow: 2, -17, 0
Green: 0, -20, 0
Cyan: 0, -19, 10
Blue: 12, 0, 0
Magenta: -1, 0, 0
Googer is offline  
post #727 of 1782 Old 01-06-2009, 12:36 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by Googer View Post

More importantly, however, I haven't seen any instances of colors looking grossly out-of-whack at any brightness with these settings, which is still rather unique from what anyone else has found with custom CMS settings where they've calibrated to absolutely minimize error at a single level, and found out after the fact that doing so introduces huge errors at others.

Sorry but I have to slightly disagree with you here. Although there are obvious errors in the numbers at 75% stimulus when calibrating at 100% stimulus, I have nothing to report visually from my own settings as "grossly out of whack", neither has LovingDVD in his last attempt, and the last time Lawguy mentioned his picture after is own calibration, he was subjectively very happy with it (please correct me LG if I have missed a negative report from you regarding your calibrated picture, from a subjective point of view).

We all agree the numbers are bad when our calibrated gamuts are checked at a different stimulus level when calibrated at 100%, but we haven't reported any grossly out of whack visual errors.

Plus you haven't published (unless I missed them) your readings at 100% and 75% stimulus. Whatever the results, we still don't know if your calibrating method is better re pure numbers, which is where most of the problem lies, not regarding the ability to get a visually mostly satisfying picture using the CMS (which is subjective anyway)...
Manni01 is online now  
post #728 of 1782 Old 01-06-2009, 01:20 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Googer View Post

...On both RS20's this has gotten me to on the order of < 2 dE for 15% gray and higher in all of 5 minutes of work, and then fighting with the gamma controls (which don't like to let you make fine adjustments due to some smoothing functions JVC seems to be applying as you try to change them) to finish the fine-tuning below that.

Right. The key difference is only going for 15-100%. If you feel up for it you may want to try getting 5-100% under a dE of 3. If you can do that in under 5 minutes still I will be impressed!
lovingdvd is offline  
post #729 of 1782 Old 01-06-2009, 01:25 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Googer View Post

They appear to be very similar to one another but operate mostly on different frequencies - Sharpness mostly on lower frequencies and Detail Enhance mostly on the highest frequencies (i.e., single-pixel). The effects can be clearly seen if you turn them to their upper ends. Sharpness introduces large (3-4 pixel-wide) halos around objects, and detail enhancer is much finer, working more on the order of single-pixel sharpness enhancement. As such, I think Sharpness is best left at 0 for basically everyone, but mild use of Detail Enhance (i.e., no higher than 15 or so) could be seen as desirable by some. YMMV of course.

Thank you. You helped me realize something else as well. Some of you may recall me reporting "light icicles" as I called them and some "glowing" around bright objects with my first RS20 which I found distracting.

On my replacement unit I thought it was interesting that I was not observing this phenomenon. Well, with my first unit it didn't focus that well and had bad MC so I used a lot of the sharpness control (30-ish IIRC). Whereas on my replacement I use very little if any. So it likely is the sharpness control that causes this affect, based on what you wrote above.

Doesn't sound like Sharpness had much practical application. But I will give the detail enhancer a try to see how I like it.
lovingdvd is offline  
post #730 of 1782 Old 01-06-2009, 01:28 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Manni01 View Post

...Did you by any chance try to measure at 75%? As you're not touching the general color setting, and are probably not overcompensating brightness, you may have a better linearity than I (or LG) got when we were trying to hit rec709 perfect with a lower color setting.

No, I didn't even bother measuring it, as I've done so a hundred times with previous calibration attempts like this. For instance I previously calibrated my unit using this same exact approach, but it measured awful at 75%. I have no reason to expect this would measure any differently.

However, the key difference is that last time I just bagged the calibration after seeing what 75% looked like, without really trying it, thinking it was a waste. But after Manni said he did this and it looked good, I decided to start over and try it again. And low and behold it does seem to hold up fairly well, at least so far.
lovingdvd is offline  
post #731 of 1782 Old 01-06-2009, 01:32 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Manni01 View Post

Sorry but I have to slightly disagree with you here. Although there are obvious errors in the numbers at 75% stimulus when calibrating at 100% stimulus, I have nothing to report visually from my own settings as "grossly out of whack", neither has LovingDVD in his last attempt...

Actually Manni I did report on obvious color error I've seen so far - see what I said in that post about the green smoke in Spiderman 3. I'm hopeful it will be far and few scenes where it rears its head, but I haven't watched enough with it to have a good feel for that yet.

Quote:
Plus you haven't published (unless I missed them) you're readings at 100% and 75% stimulus, whatever the results, we still don't know if your calibrating method is better re pure numbers, which is where most of the problem lies, not regarding the ability to get a visually mostly satisfying picture using the CMS (which is subjective anyway)...

Yes I am eagerly awaiting Googer's numbers to see how they measure out at 75%. If they measure out better than what I got by using Color 0 and 100% patterns, then I will have reason to believe that the approach he is using is superior.
lovingdvd is offline  
post #732 of 1782 Old 01-06-2009, 01:43 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by Lawguy View Post

I haven't sent anything on. Rather, I have asked people to look at the thread.

Write something up and I'll put it in the first post.

Sorry LG, I misunderstood your post re "feelers" sent to JVC. I understand now what it meant. I'll post my suggestion in the firmware thread.
Manni01 is online now  
post #733 of 1782 Old 01-06-2009, 01:44 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Googer View Post

It's a tradeoff that's looking to be relatively close to the best one can do with the CMS. Yes I can pull in the points more by turning down the Saturation controls more, but this pulls down the lightness so much that it then becomes necessary to (substantially) increase the CMS Brightness, which as we've already found, causes color tracking across different % stimuli to completely fall apart.

Anyway, current settings for these results for anyone that feels like trying them (before I go in and play around some more ):

Color: -10
Tint: 0

CMS (H, S, B):

Red: 4, -10, 0
Yellow: 2, -17, 0
Green: 0, -20, 0
Cyan: 0, -19, 10
Blue: 12, 0, 0
Magenta: -1, 0, 0

Googer - I very much like the track you are on with this. Let me see if I have the theory straight...

Essentially the idea is that cranking brightness is the culprit. So by using the Color control to lower things overall to the point where the Saturation control alone can dial in the gamut, we hope that will stay linear. Is that it in a nutshell?

BTW I have measured the affect of changing the Color control on the THX gamut. Essentially the Color control works like a Brightness setting and only affected the Y in the xyY for primaries and secondaries (xy did not move).

So if I have this straight, essentially by lowering the Color control that is equivalent to lowering the Brightness of each color. So then I guess that provides enough granularity to the Saturation controls so that they alone can be used to hit their targets without touching Brightness.

I have to say that on paper this sounds like it has good potential, especially when accompanied by your report that it seems to be working great with no errors.

Just to confirm, you calibrated using Color at -10 and 100% patterns, correct? And you were able to get to the right xyY point and even the right brightness by using just the Saturation control? Looks like it from your data, but hard to imagine how you would have been able to fine tune Y without touching the brightness control?

Anyway it would be great if you can post your measured xyY at 100% and 75%. Thanks again.
lovingdvd is offline  
post #734 of 1782 Old 01-06-2009, 01:45 PM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by lovingdvd View Post

Right. The key difference is only going for 15-100%. If you feel up for it you may want to try getting 5-100% under a dE of 3. If you can do that in under 5 minutes still I will be impressed!

I do have dE < 3 for all levels that I can get reasonable readings from my EyeOne for, but I used the gamma controls to get the low end. Even with the finicky gamma controls though, an entire grayscale flattening is only taking me around 15-20 minutes start-to-finish; it's definitely not an hour or hours-long process. For the truly lowest-end tracking, I mostly trust my eye at many very low levels (i.e., 1%, 2%, ....), but I also for jollies checked that with my EyeOne at a distance of around 6 inches from the RS20's lens so even at those low levels it should be receiving enough light to give possibly reasonable readings, and sure enough, even those are tracking down in the <5 dE range with gamma-scale tweaking.
Googer is offline  
post #735 of 1782 Old 01-06-2009, 01:58 PM
AVS Addicted Member
 
millerwill's Avatar
 
Join Date: Apr 2004
Location: Berkeley, CA
Posts: 11,356
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 27
Quote:
Originally Posted by lovingdvd View Post

Googer - I very much like the track you are on with this. Let me see if I have the theory straight...

Essentially the idea is that cranking brightness is the culprit. So by using the Color control to lower things overall to the point where the Saturation control alone can dial in the gamut, we hope that will stay linear. Is that it in a nutshell?

BTW I have measured the affect of changing the Color control on the THX gamut. Essentially the Color control works like a Brightness setting and only affected the Y in the xyY for primaries and secondaries (xy did not move).

So if I have this straight, essentially by lowering the Color control that is equivalent to lowering the Brightness of each color. So then I guess that provides enough granularity to the Saturation controls so that they alone can be used to hit their targets without touching Brightness.

I have to say that on paper this sounds like it has good potential, especially when accompanied by your report that it seems to be working great with no errors.

Just to confirm, you calibrated using Color at -10 and 100% patterns, correct? And you were able to get to the right xyY point and even the right brightness by using just the Saturation control? Looks like it from your data, but hard to imagine how you would have been able to fine tune Y without touching the brightness control?

Anyway it would be great if you can post your measured xyY at 100% and 75%. Thanks again.

This is reminescent of Jason's comments at the very beginning, namely having to lower Color a great deal, even to -20 or so, and then go from there.
millerwill is offline  
post #736 of 1782 Old 01-06-2009, 01:59 PM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by lovingdvd View Post

Googer - I very much like the track you are on with this. Let me see if I have the theory straight...

Essentially the idea is that cranking brightness is the culprit. So by using the Color control to lower things overall to the point where the Saturation control alone can dial in the gamut, we hope that will stay linear. Is that it in a nutshell?

Yep, that's the gist of it. As far as I can tell response remains at least mostly linear as long as you don't mess with the CMS brightness setting.

Quote:
BTW I have measured the affect of changing the Color control on the THX gamut. Essentially the Color control works like a Brightness setting and only affected the Y in the xyY for primaries and secondaries (xy did not move).

So if I have this straight, essentially by lowering the Color control that is equivalent to lowering the Brightness of each color. So then I guess that provides enough granularity to the Saturation controls so that they alone can be used to hit their targets without touching Brightness.

Right; one thing I haven't yet done once lowering the Saturation on each color to help pull them in is see if I can get away with putting the global Color setting back up a bit higher to restore some of the lost brightness, which in turn may mean lowering the saturation more for the colors that really need it could be done. I only dropped saturation by as much as I did for each color because at the point I currently have it at, dropping saturation by any more then started to drop the luminance too much.

Quote:
I have to say that on paper this sounds like it has good potential, especially when accompanied by your report that it seems to be working great with no errors.

Just to confirm, you calibrated using Color at -10 and 100% patterns, correct? And you were able to get to the right xyY point and even the right brightness by using just the Saturation control? Looks like it from your data, but hard to imagine how you would have been able to fine tune Y without touching the brightness control?

Anyway it would be great if you can post your measured xyY at 100% and 75%. Thanks again.

The starting point was to completely zero out the CMS and then find the global Color / Tint settings that minimized dE to start with, because my thinking was that the less work the CMS has to do, the better. From there I then adjusted each color individually, first doing any hue correction as necessary since errors there are the most noticeable of all, then pulling in saturation as allowed to further reduce dE. Once dE started to actually climb more as a result of reducing saturation (due to how much luminance was dropping), I stopped. The sole exception to this was cyan since in my first CMS attempt, I'd found that increasing brightness for it (at least as long as you don't go overboard on it) doesn't perceptibly (by eye anyway) cause the nonlinear response that it so visibly does on all other colors.

Anyway, I intend to take formal measurements at different % stimuli sometime later tonight to appease those that are insistent I do so. My guess is that the numbers should basically bear out what I've been saying about response basically remaining linear though; even if not perfectly so, it's definitely better than what is plainly obvious anytime I've tried playing with the CMS brightness settings.
Googer is offline  
post #737 of 1782 Old 01-06-2009, 02:04 PM
AVS Special Member
 
mdrew's Avatar
 
Join Date: Dec 2004
Location: Up north
Posts: 1,067
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by mdrew View Post

Is anyone using a VP50pro?

I was playing a bit more last night and noticed that I can not get the blacker than black bars to show when using the Pro's brightness / contrast test pattern. I know I am supposed to set B / C to where I can get the 1 and 2 IRE above blacks to show, but this has me wondering if I don't something set correctly. I should be able to see the blacker than black bar, right?? It doesn't matter if the Iris is wide open or closed. All I get is the 1 and 2 IRE bars to show (I can also get them to disappear).

I was also playing around with the Pro's Color Space Output a bit. If I change it from 4:2:2 to RGB (HDMI output), the image looks washed out. My last project was not affected by changing this nearly to this degree.

If anyone is using a VP50pro or other VP, I'd appreciate if you could tell me how you have its output set up. I'm as busy as a show pony for the next several days and do not have time to fiddle. IE: output format, color space, etc I don't need gamma or color settings yet. I don't want to even begin thinking about that yet, or reading the calibration thread.. I'll never get sleep if I start down that road till I can get a few days off.

I posted this in the owner's thread, but didn't get a response. I suspect my question is too ambiguous to answer.

I tried changing the RS20's color space to manual 4:2:2 but that didn't do any good. I still can't get the blacker than black bar to show.

Anyone have an idea what I'm doing wrong? My PT1000 had no problem showing all three bars.

If it means anything, Jason did calibrate this machine.
mdrew is offline  
post #738 of 1782 Old 01-06-2009, 02:20 PM
AVS Club Gold
 
mrlittlejeans's Avatar
 
Join Date: Jan 2004
Location: the area formerly known as Black Forest
Posts: 4,714
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 192
I wasn't able to get BTB to show either using a PS3 and sending YPrPb. I haven't tried the HD DVD DVE yet but did try the THX optimizer on the Toshiba A35 and it wasn't passing BTB there either. I know the RS1 showed it with both so maybe the RS20 doesn't display BTB.

Absence of evidence is not evidence of absence

mrlittlejeans is offline  
post #739 of 1782 Old 01-06-2009, 02:25 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Googer View Post

I do have dE < 3 for all levels that I can get reasonable readings from my EyeOne for, but I used the gamma controls to get the low end. Even with the finicky gamma controls though, an entire grayscale flattening is only taking me around 15-20 minutes start-to-finish; it's definitely not an hour or hours-long process.

We used the same exact approach. In my case I used red offset of -3 and left green and blue at 0. This gave me what I needed for 25-45% without having to compensate for that in the gamma. I figure the less correction in the custom gamma, the better - provided that the offsets are not ever raised above 0 - that is a big no no for this pj.
[/quote]

Can you describe the process you use to dial this in so quickly? Although it sounds like we may have taken the same approach, I spent close to an hour on this IIRC. How do you deal with the fact that the RS20 takes your numbers at x% and changes them at surrounding points? I had to do this over and over as I mentioned earlier in this thread. Maybe the key difference is that I used instrumentation to ensure 5-20 stayed in tack and keeping 5% happy was not easy.

BTW I don't know what calibration software you are using but if it supports meter training you may want to pick up a Spyder 2 or 3. I use a Spyder 2 which I trained to my EyeOne. This gives me the color accuracy of the EyeOne with consistent and accurate the low level light reading down to 5% (perhaps even lower, I never bothered measuring below that). Plus its much easier to work with since you don't have to keep stopping for dark readings etc.
lovingdvd is offline  
post #740 of 1782 Old 01-06-2009, 02:33 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by lovingdvd View Post

So if I have this straight, essentially by lowering the Color control that is equivalent to lowering the Brightness of each color. So then I guess that provides enough granularity to the Saturation controls so that they alone can be used to hit their targets without touching Brightness.

I'm not sure I follow you, LovingDVD. From what I see at this stage from Googer's results, yes, lowering the Color control is equivalent to lowering the brightness of each color (we already knew that) but also their saturation (this is why we did it in the first place).

Googer (in his last attempt) seems to be using a bit of saturation control (although more than in his first attempt) to bring things closer, and he ends up with colors which are still oversaturated. But because he uses more color control AND doesn't use the CMS brightness control at all to compensate for what is lost through the use of the color control, his colors lack luminance as well (so xy is wrong, Y is wrong too).

Finally, we don't know exactly what this compromise brings to the table as we are not able to see how his settings track at different levels of stimulus, which is the reason why he isn't using the CMS brightness control (a very interesting idea, which goes with LG's recent findings, but we need the data!)...

As you know I'm always ready to get excited about a new possible solution (including those already suggested by Googer), but here I am a bit (it may be because I don't understand, so forgine me if it's the case, it wouldn't be the first time).

At this stage, I see as much potential in your own attempt with a color setting at zero which ends up in slightly oversaturated colors, as it may track better at different levels of stimulus (using less color control, you need less CMS brightness to compensate so you may generate less linearity problems), and it may also be more accurate despite the slight oversaturation, because you may be closer re Y.

I'm not knocking down Googer's attempts, just saying we lack enough objective data to see what is better or worse. Again, if we speak visually or subjectively, we have already found settings which are mostly satisfying. The question is more is there a way to get a customised gamut that tracks as well as the THX preset at different levels of stimulus...

We won't be able to find out from an objective point of view until you both post your results at 75% and 100% stimulus with your last settings...

If you do, I promise I'll do it too for my next attempt

(I guess it would be good if we were all using the same patterns though, as there seems to be a variation between those available, so please tell us which one you use and I'll try to use the same).

EDIT: sorry I missed Googer's reply while composing this, so thanks Googer, looking forward to these readings!
Manni01 is online now  
post #741 of 1782 Old 01-06-2009, 02:36 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Googer View Post

Right; one thing I haven't yet done once lowering the Saturation on each color to help pull them in is see if I can get away with putting the global Color setting back up a bit higher to restore some of the lost brightness, which in turn may mean lowering the saturation more for the colors that really need it could be done. I only dropped saturation by as much as I did for each color because at the point I currently have it at, dropping saturation by any more then started to drop the luminance too much.

I'm a bit confused. Isn't there a set point for which you must lower saturation to make it correct? Not sure how you can just stop at a point where a control runs out and still wind up with a valid result.

Quote:
The starting point was to completely zero out the CMS and then find the global Color / Tint settings that minimized dE to start with, because my thinking was that the less work the CMS has to do, the better.

So basically you just watched dE and xyY and the only thing that should have been changing is Y.

From there I then adjusted each color individually, first doing any hue correction as necessary since errors there are the most noticeable of all, then pulling in saturation as allowed to further reduce dE. Once dE started to actually climb more as a result of reducing saturation (due to how much luminance was dropping), I stopped. The sole exception to this was cyan since in my first CMS attempt, I'd found that increasing brightness for it (at least as long as you don't go overboard on it) doesn't perceptibly (by eye anyway) cause the nonlinear response that it so visibly does on all other colors.
[/quote]

OK but I am again confused how you can just stop once dE worsens yet still have accurate results.

Quote:
Anyway, I intend to take formal measurements at different % stimuli sometime later tonight to appease those that are insistent I do so. My guess is that the numbers should basically bear out what I've been saying about response basically remaining linear though; even if not perfectly so, it's definitely better than what is plainly obvious anytime I've tried playing with the CMS brightness settings.

Thanks. I hope you understand it is not that we doubt you, but rather that we want to compare the data to see if it yields results that come out better than with the other approaches we've been taking. Hopefully this solves the linearity issue while at the same time yielding accurate colors. Its a great lead you've been following and hopefully with the data we can assess whether this is a viable work around for some people.
lovingdvd is offline  
post #742 of 1782 Old 01-06-2009, 02:37 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Googer - one clarification please. You are saying that with Color at -10 you were able to dial in xy and Y correctly just using the Saturation control?
lovingdvd is offline  
post #743 of 1782 Old 01-06-2009, 02:39 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Someone above requested I post my CMS settings so they can try Rec Ldvd for themselves. I'll be happy to. Where gathering the data for the post I realized that one color was off more in brightness than I though so I am going to tweak this tonight. As soon as I have the refined data I'll post it, along with updated impressions from viewing tonight.
lovingdvd is offline  
post #744 of 1782 Old 01-06-2009, 02:39 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by mdrew View Post

I posted this in the owner's thread, but didn't get a response. I suspect my question is too ambiguous to answer.

I tried changing the RS20's color space to manual 4:2:2 but that didn't do any good. I still can't get the blacker than black bar to show.

Anyone have an idea what I'm doing wrong? My PT1000 had no problem showing all three bars.

If it means anything, Jason did calibrate this machine.

Did you try to set the HDMI input type (standard or enhanced) manually? It's not detected automatically with some sources when set on auto.
Manni01 is online now  
post #745 of 1782 Old 01-06-2009, 02:46 PM
AVS Special Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 8,818
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 23
Quote:
Originally Posted by Manni01 View Post

...At this stage, I see as much potential in your own attempt with a color setting at zero which ends up in slightly oversaturated colors, as it may track better at different levels of stimulus (using less color control, you need less CMS brightness to compensate so you may generate less linearity problems), and it may also be more accurate despite the slight oversaturation, because you may be closer re Y.

Unfortunately I can tell you definitively (due to hours and hours of experimentation) that even using my current approach of color 0 and calibrating to 100% patterns, it definitely does not track well. Essentially cyan, green and red (IIRC) all pop back to their original OOTB gamut positions for xy and also have significant Y errors. As I said I didn't measure this specific calibration but I did for another calibration just like this one and I don't expect these results would be any different.

That being said, what I am optimistic about, is that despite the very wrong tracking at 75%, for real world viewing so far it seems to be doing the trick, with an occasional blowout of a color (florescent green for instance) to be expected. I haven't watched enough to know whether this will be very rare and perfectly acceptable, or whether it will occur more often and be more annoying than anything (forcing us back to THX +color mode with no custom gamma ). Man this is frustrating, especially considering they gave us the proper tool for the job but its just not working as needed.
lovingdvd is offline  
post #746 of 1782 Old 01-06-2009, 03:12 PM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by lovingdvd View Post

We used the same exact approach. In my case I used red offset of -3 and left green and blue at 0. This gave me what I needed for 25-45% without having to compensate for that in the gamma. I figure the less correction in the custom gamma, the better - provided that the offsets are not ever raised above 0 - that is a big no no for this pj.

Can you describe the process you use to dial this in so quickly? Although it sounds like we may have taken the same approach, I spent close to an hour on this IIRC. How do you deal with the fact that the RS20 takes your numbers at x% and changes them at surrounding points? I had to do this over and over as I mentioned earlier in this thread. Maybe the key difference is that I used instrumentation to ensure 5-20 stayed in tack and keeping 5% happy was not easy.

Basic 5-10 minute method (this works on all display types I've ever calibrated so long as the grayscale controls work as expected - CRT, LCD, plasma, RPTV of various flavors, FP LCOS):

1. Take measure of 20-step grayscale.
2. Look at resulting RGB levels plot to see how R, G, and B track across all stimuli.
3. Make changes to RGB gains (and cuts) to attempt to bring the RGB lines closer to the target 100% level across the board, not ever targeting any single % gray level.
4. Repeat from step 1 until you're happy with the results.

On the RS20, I can get dE < 2 from 15% on up using just the RGB gains and never have to touch the offsets even.

Step 2 for the RS20 - gamma tweaking for the low-end - you should notice this process seems familiar.

1. Take measure of 20-step grayscale.
2. Look at resulting RGB levels plot to see how R, G, B track across all stimuli. Notice they're all very close to 100% except for the bottom end.
3. Make changes to the individual RGB gamma values (most likely only at the 5% and 10% input levels) to try to bring in the values.
4. Repeat from step 1 until plot says you're very close (perfect isn't necessary though due to more refinement coming up).
5. Look at very low % gray levels by eye (I like single % increments from 0-15 or so) and make adjustments to the individual RGB gamma levels to get them to looking as close as possible to neutral as possible. Since the higher end of this range should be close to D65 and your eye is pretty good as seeing relative color differences, doing this strictly by eye shouldn't be all that hard to do.
6. Take measure of 20-step grayscale; notice you should have dE < 3 across the board if you did a good job, < 2 if you did a great job.

Quote:
BTW I don't know what calibration software you are using but if it supports meter training you may want to pick up a Spyder 2 or 3. I use a Spyder 2 which I trained to my EyeOne. This gives me the color accuracy of the EyeOne with consistent and accurate the low level light reading down to 5% (perhaps even lower, I never bothered measuring below that). Plus its much easier to work with since you don't have to keep stopping for dark readings etc.

As I've said before, I'm using HCFR right now, but I've also used in the past umr's own i1 Pro DCS (most likely the version I have is out-of-date since he's probably continued to update it internally but I don't know that he's made newer versions since v1.7 available to those of us that bought it from back when he was selling it ). I probably should pick up a cheap Spyder or other lower-end colorimeter at some point and train it as you suggest, since the constant recalibrating of the EyeOne (every ~10 minutes) is bothersome, as is how much it struggles at low input levels.
Googer is offline  
post #747 of 1782 Old 01-06-2009, 03:12 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by Googer View Post

Yep, that's the gist of it. As far as I can tell response remains at least mostly linear as long as you don't mess with the CMS brightness setting.

Right; one thing I haven't yet done once lowering the Saturation on each color to help pull them in is see if I can get away with putting the global Color setting back up a bit higher to restore some of the lost brightness, which in turn may mean lowering the saturation more for the colors that really need it could be done. I only dropped saturation by as much as I did for each color because at the point I currently have it at, dropping saturation by any more then started to drop the luminance too much.

The starting point was to completely zero out the CMS and then find the global Color / Tint settings that minimized dE to start with, because my thinking was that the less work the CMS has to do, the better. From there I then adjusted each color individually, first doing any hue correction as necessary since errors there are the most noticeable of all, then pulling in saturation as allowed to further reduce dE. Once dE started to actually climb more as a result of reducing saturation (due to how much luminance was dropping), I stopped. The sole exception to this was cyan since in my first CMS attempt, I'd found that increasing brightness for it (at least as long as you don't go overboard on it) doesn't perceptibly (by eye anyway) cause the nonlinear response that it so visibly does on all other colors.

Anyway, I intend to take formal measurements at different % stimuli sometime later tonight to appease those that are insistent I do so. My guess is that the numbers should basically bear out what I've been saying about response basically remaining linear though; even if not perfectly so, it's definitely better than what is plainly obvious anytime I've tried playing with the CMS brightness settings.

Googer, thanks for clarifying your strategy, it is very useful.

Be careful though, because if you push color up a little bit, you may end up with -8/-7 which is what JeffY, LG and myself found more than 15 pages ago as offering the best compromise to get xy within reach of REC709, but if you are not using brigtness to compensate, you may end up with a nice looking gamut tracking well at different levels of stimulus, but it will still be way off regarding Y, even if you limit your use of the saturation control in order not to increase dE.

In a way, because you and LovingDVD (in your last respective attempts) end up with more oversaturated gamuts than we did, it looks like you may have more linearity (no use of brightness) but a bigger error in luminance, while LovingDVD is closer in luminance (no use of color) but has less linearity.

Both these strategies may be better, depending on which trade-off you want to compromise (linearity vs brightness vs rec709 accuracy).

Personally, I found my first attempts visually very dim with real material (lack of luminance, as I was not correcting brightness), which is why I persevered until I reached my 4th attempt. With these last settings, linearity is no doubt bad, but I've never seen how it affects the picture adversely (unlike the lack of brightness in my first attempts).

I am curious to try the scene found by LovingDVD, as it may give me the motivation I need for my 5th attempt.

I'm also curious to see if I like Rec LDvd, as it may be a better compromise for me (better linearity, better brightness, less rec709 accuracy) until JVC fix this CMS or we find a way to crack it.
Manni01 is online now  
post #748 of 1782 Old 01-06-2009, 03:19 PM
AVS Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 2,220
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 48 Post(s)
Liked: 93
Quote:
Originally Posted by Googer View Post

Basic 5-10 minute method (this works on all display types I've ever calibrated so long as the grayscale controls work as expected - CRT, LCD, plasma, RPTV of various flavors, FP LCOS):

1. Take measure of 20-step grayscale.
2. Look at resulting RGB levels plot to see how R, G, and B track across all stimuli.
3. Make changes to RGB gains (and cuts) to attempt to bring the RGB lines closer to the target 100% level across the board, not ever targeting any single % gray level.
4. Repeat from step 1 until you're happy with the results.

On the RS20, I can get dE < 2 from 15% on up using just the RGB gains and never have to touch the offsets even.

Step 2 for the RS20 - gamma tweaking for the low-end - you should notice this process seems familiar.

1. Take measure of 20-step grayscale.
2. Look at resulting RGB levels plot to see how R, G, B track across all stimuli. Notice they're all very close to 100% except for the bottom end.
3. Make changes to the individual RGB gamma values (most likely only at the 5% and 10% input levels) to try to bring in the values.
4. Repeat from step 1 until plot says you're very close (perfect isn't necessary though due to more refinement coming up).
5. Look at very low % gray levels by eye (I like single % increments from 0-15 or so) and make adjustments to the individual RGB gamma levels to get them to looking as close as possible to neutral as possible. Since the higher end of this range should be close to D65 and your eye is pretty good as seeing relative color differences, doing this strictly by eye shouldn't be all that hard to do.
6. Take measure of 20-step grayscale; notice you should have dE < 3 across the board if you did a good job, < 2 if you did a great job.

As I've said before, I'm using HCFR right now, but I've also used in the past umr's own i1 Pro DCS (most likely the version I have is out-of-date since he's probably continued to update it internally but I don't know that he's made newer versions since v1.7 available to those of us that bought it from back when he was selling it ). I probably should pick up a cheap Spyder or other lower-end colorimeter at some point and train it as you suggest, since the constant recalibrating of the EyeOne (every ~10 minutes) is bothersome, as is how much it struggles at low input levels.

Thanks very much for this, I'll try a combination of yours and Mark's guidelines (with a zest of JeffY) for my next attempt! Like LovingDVD, I've always found the greyscale controls tricky (although not as tricky as the CMS). Hope this will lead to the flat greyscale (and better contrast) I'm aspiring to...
Manni01 is online now  
post #749 of 1782 Old 01-06-2009, 03:40 PM
AVS Special Member
 
JeffY's Avatar
 
Join Date: Mar 2001
Location: England
Posts: 3,667
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I think what I have done with CMS is somewhat in between what Googer has done and what others have done. I can see that the colour brightness controls have a negative effect on linearity, but some of the other controls are hardly perfect either. What I have done it to use the controls with moderation and then rather than attempt to get the lumiance values perfect using probe, I used colour filters that come with DVE to balance the colours out. Given the broken nature of the CMS I'm very happy with the end results. I can look at test screens and real material and nothing is really broken or out of whack.
JeffY is offline  
post #750 of 1782 Old 01-06-2009, 03:51 PM
Advanced Member
 
Googer's Avatar
 
Join Date: Jan 2002
Location: Pittsburgh, PA, USA
Posts: 674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 15
Quote:
Originally Posted by Manni01 View Post

Thanks very much for this, I'll try a combination of yours and Mark's guidelines (with a zest of JeffY) for my next attempt! Like LovingDVD, I've always found the greyscale controls tricky (although not as tricky as the CMS). Hope this will lead to the flat greyscale (and better contrast) I'm aspiring to...

Oh BTW, I left out one bit in my steps specific to the RS20, which you probably already know but I'll remind you nonetheless - if you want to target some specific gamma - say, 2.3 or 2.4, load up a custom gamma ramp with that before you begin any grayscale modification (meaning the gains / offsets even), since the overall untweaked gamma has a minor effect the grayscale tracking.
Googer is offline  
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off