[Timothy91]

Hi all,


The AVS709 color-contrast-clipping pattern I noticed that Red, Green & Blue all clip at different points. Green has a blinking bar at 235 while Red & Blue are barely blinking at 233. Also, the 3 primary colors all clip at different points than the grayscale white clipping patterns. Is this a normal result or does it indicate an incorrect color balance?


Second. After going into the menu and adjusting both the main color setting and color CMS section (where I also made a minor blue tint adjustment to fade the bar at 235 to appear invisible), I was able to get all the primary colors to clip at the same point as the white contrast grayscale. I noticed little to no change on the results on the primary color/tint pattern setting after making these adjustments. They weren't dramatic changes, just minor 'bumps' to change the point at which each color clips and also matching the precise point that white clips.


I swapped back and forth between my old setting on ISF1 & the new adjustments on ISF2. I don't notice a direct A/B difference but when watching the new settings I would swear that some blue and purple colors are ever so slightly 'off' but again, when I go to A/B compare it to the old setting, I honest can't see a difference. Perhaps my instincts are noticing the difference or perhaps I have psychologically convinced myself that I should see a difference and therefore see what I want to see (ie, subjective bias is messing with my objective observations).


There is definitely a difference but can't say for sure if I like it. When watching HD Cable TV, the old setting looks normal but the new setting seems to really show any noise or distortions more glaringly. I have adjusted the TV to the sharpness pattern so strictly already that I know if there is noise, it's because the source video just plain sucks.


I am evaluating the new adjustments still but wanted to run this by the calibrators for an opinion on what I've done. Could my adjustments to get the white & color clipping patterns to match up without effecting the main color test result give a just barely noticable downgrade of the color quality? Or have I done something worth keeping and maybe just aren't completely used to it?

 

[Doug Blackburn]

It is somewhat common for 1 or 2 colors to "run out of gas" before the 3rd color (meaning... at the top end of the luminance scale, 1 or 2 colors may not be able to achieve the proper level). That does indicate a color problem since the 3 colors are not all "equal" (they are never really equal when producing shades of gray or white.... 72% of the light is green, 21% is red, and 7% is blue light for neutral gray or white).


There are 2 reasons for the colors being imbalanced at 100% white...


- Contrast set too high

- Calibration problem


If you reduce the contrast setting and then find all 3 colors are "equal" at 235, then your TV has an internal problem that limits how high you can set the contrast control without getting into non-linear operation (color-wise). Once you have Contrast set low enough that the 3 colors remain balanced at 100%, you can increase Contrast 1 click at a time until you can see the 3 colors are no longer staying together. When you hit that point you will have determined how far you can increase Contrast without causing a discoloration at the top end. This may (or not) be brighter than you want Contrast set for a dark room anyway. So the real purpose for this check is to determine the highest you can set contrast without having it introduce a color problem at 100% white. The problem is, without a meter, you don't know if that Contrast setting is 100 fL or 20 fL... you want about 35 fL for a dark room, and probably 55+ for a room with some daylight.


If the relationship of the 3 primary colors never changes, then the TV is remaining linear at all Contrast settings and you have a calibration problem... and it won't be easy to fix with the controls in most TVs. For example, if 90% white is correct (about 212 digital) and 100% is "off", if you make an adjustment with typical high/low controls to fix 100%, you'll end up with an error at 90% that you didn't have before. Samsung and some LG TVs have 10 or 20 adjustment points that can fix 100% while not changing 90%.

 

[Timothy91]



Quote:
Originally Posted by Doug Blackburn /forum/post/20808987


It is somewhat common for 1 or 2 colors to "run out of gas" before the 3rd color (meaning... at the top end of the luminance scale, 1 or 2 colors may not be able to achieve the proper level). That does indicate a color problem since the 3 colors are not all "equal" (they are never really equal when producing shades of gray or white.... 72% of the light is green, 21% is red, and 7% is blue light for neutral gray or white).


There are 2 reasons for the colors being imbalanced at 100% white...


- Contrast set too high

- Calibration problem

I'm confused here. Last night I was playing with the settings again. On the color-contrast pattern, when contrast is turned "down", the result is the bars become visible past 235. When adjusting pure contrast white follows that behavior as well on the standard grayscale contrast pattern. When I adjust the main "Color" control while looking at the color-contrast pattern, all 3 colors change on the scale. When I reduce color, the higher number bars are revealed. When I increase color, the bars start disappearing.


I went into the more find CMS color control and changed the primary blue/red/green colors until all 3 colors on the pattern "matched" the behavior of the color white on the grayscale pattern. (ie, now after my adjustments white is set to clip right at 235 and all 3 colors now clip at 235). After the fine tuning was done and the main color control was readjusted, the blue-filter color test shows the color is set correctly. However, I need to test red and green still to be sure they aren't a tad bit off.


When setting the main color setting, (blue-filter color test), "48" is the proper setting when no adjustments were done in the CMS menu. After the very slight CMS adjustments to match all 3 colors to the behavior of 'white', the only setting color setting that allows me to 'pass' the blue filter test and match white's behavior on the color-contrast pattern is now "47". It took a while to go back and forth with the different patterns and play with the adjustments until I reached an equilibrium. Now, I want to perform a red and green filter test to verify those colors are still in proper balance.

Quote:
If you reduce the contrast setting and then find all 3 colors are "equal" at 235, then your TV has an internal problem that limits how high you can set the contrast control without getting into non-linear operation (color-wise). Once you have Contrast set low enough that the 3 colors remain balanced at 100%, you can increase Contrast 1 click at a time until you can see the 3 colors are no longer staying together. When you hit that point you will have determined how far you can increase Contrast without causing a discoloration at the top end. This may (or not) be brighter than you want Contrast set for a dark room anyway. So the real purpose for this check is to determine the highest you can set contrast without having it introduce a color problem at 100% white. The problem is, without a meter, you don't know if that Contrast setting is 100 fL or 20 fL... you want about 35 fL for a dark room, and probably 55+ for a room with some daylight.

Do you have any "ballpark" backlight number where the LG 55LH90 hits a certain FTL? I could at least try out what it might look like post-calibration.

Quote:
If the relationship of the 3 primary colors never changes, then the TV is remaining linear at all Contrast settings and you have a calibration problem... and it won't be easy to fix with the controls in most TVs. For example, if 90% white is correct (about 212 digital) and 100% is "off", if you make an adjustment with typical high/low controls to fix 100%, you'll end up with an error at 90% that you didn't have before. Samsung and some LG TVs have 10 or 20 adjustment points that can fix 100% while not changing 90%.

The only way I can think of to test this is to turn down the contrast and watch the 3 colors on the color-contrast pattern to see if they lose 'sync' at some point while the bars above 235 appear. This will give me an idea of where I stand. This TV (LG 55LH90) is considered a quality TV, so I would have to think it will be pretty consistent. I'll find out for sure tonight.

 

[Timothy91]

Ok,


I tinkered around last night some more. I pulled up the color-contrast pattern and lowered the contrast revealing the bars slowly above 235 and the Red/Green/Blue consistently lit up new bars in unison all the way until 243 where blue and red ran out of gas and only green would give a faint bar at 243. Lowering the contrast any further below around "70" changed nothing more.


So, it seems like all 3 primary colors have very consistent output/performance with one another, which IMO it BETTER because I paid extra to have a THX rated TV that was the best LCD on the market in 2009.


I ran the primary color bars test from AVS709 and my TV has a built-in blue/green/red filter that I can turn on and off. Only the Red pattern showed 2 out of the 8 or so different bars as being a visibly different shade.


I turned off the filter and noticed the two colors of the bars were magenta and yellow. So, I went back into CMS and adjusted the magenta and yellow colors until I had the boxes really close to matching the overall shade. I subsequently tested the Green and Blue primary colors one more time and they were still virtually indistinguishable bars.


So, in essence, I have verified my TV's color stays very consistent on the 3 primary colors across various intensities and while I don't have the equipment to test the variances at low intensities, I have a strong suspicion that the 3 primary colors will still be consistent there as well.


I have verified that my CMS color adjustments have corrected a color imbalance in the 3 primary colors where green was always pushed above red and blue. I also further corrected the inaccuracies in the red color for both yellow and magenta. All 3 color bar patterns are very close to perfect now and all 3 primary colors behave EXACTLY like white on the contrast test patterns. The gray/color intensity match pattern however shows a slightly darker "Red" than the matching backround for gray. So, it's my opinion that Red is not perfect but it's pretty darn close and the perception of that on the TV's picture is negligible. Reds look very nice on content to me. I have no complaint.


So, now that I have a rather nicely balanced color and the factory THX grayscale settings plugged into the ISF section, I'm confident that my ISF settings should be in the ballpark if my contrast is set to "90" to match the THX setting. I'm hoping that puts my gamma curve in a nice "ballpark".


The only thing now which perplexes me is, the backlight setting. Will the backlight setting effect gamma? What should the backlight setting be to achieve a night time FTL of 35? I have a gut feeling that I will want it to be brighter than most other people would want it in a dark room, so I will want to factor that into my settings so I get a picture that I really like.


Anyone know what setting range the backlight for the LG 55LH90 is set at to achieve a 35 FTL output? I have my backlight currently set on 62 for night viewing. It's bright, but it's the intensity that looks most natural to me on daylight movie scenes in a dark room.

 

[Doug Blackburn]

You can NEVER trust the blue filter method of setting color and tint. If the red, green, and blue points aren't accurate, the blue filter method won't give accurate results... even if just 1 or 2 colors are "off" from the Rec. 709 standard coordinates, the blue filter method won't give accurate results. Furthermore, the inexpensive blue filters in cardboard holders are not necessarily "ideal" filters and error can creep in there also. There is no way for you to tell if your particular blue filter is good or not so good. There's also no way (without instrumentation) for you to tell how accurate your red, green, and blue points are. In fact, every properly written set of instructions for using the blue filter to set color and tint tells you to view content after setting the controls and if anything looks "off" color-wise to further adjust color and tint controls by eye until images look natural/normal. That's because the blue filter method isn't reliable. You can't just assume the color and tint settings you get using the blue filter are accurate.


And... you can NEVER... and I mean NEVER trust your eyes to see accurately. Because... your eyes will assume that the BLUE-EST shade of white is really the accurate white and accurate shades of white will appear to be too yellow when they are NOT too yellow. Other "tints" of color produce additional perceptual errors. You cannot stop these perceptual errors from happening. No matter HOW aware you are of their existence, the human visual system is always tricked into seeing the wrong things. This is why instrumentation is used for calibration. Instrumentation can't be fooled. All the adjustments you did by eye could easily have made the display LESS accurate rather than MORE accurate... there's probably a 75% chance the adjustments you made did not improve the accuracy of images (or made accuracy worse) and a 25% chance that you did improve images overall.


All this applies to looking at ramp patterns (stripes of shades of white from 80% to 110%) trying to use them to make a display more accurate by eye. You cannot change how human visual perception works even when you KNOW you can be fooled.

 

[PlasmaPZ80U]



Quote:
Originally Posted by Timothy91 /forum/post/20806801


Hi all,


The AVS709 color-contrast-clipping pattern I noticed that Red, Green & Blue all clip at different points. Green has a blinking bar at 235 while Red & Blue are barely blinking at 233. Also, the 3 primary colors all clip at different points than the grayscale white clipping patterns. Is this a normal result or does it indicate an incorrect color balance?


Second. After going into the menu and adjusting both the main color setting and color CMS section (where I also made a minor blue tint adjustment to fade the bar at 235 to appear invisible), I was able to get all the primary colors to clip at the same point as the white contrast grayscale. I noticed little to no change on the results on the primary color/tint pattern setting after making these adjustments. They weren't dramatic changes, just minor 'bumps' to change the point at which each color clips and also matching the precise point that white clips.


I swapped back and forth between my old setting on ISF1 & the new adjustments on ISF2. I don't notice a direct A/B difference but when watching the new settings I would swear that some blue and purple colors are ever so slightly 'off' but again, when I go to A/B compare it to the old setting, I honest can't see a difference. Perhaps my instincts are noticing the difference or perhaps I have psychologically convinced myself that I should see a difference and therefore see what I want to see (ie, subjective bias is messing with my objective observations).


There is definitely a difference but can't say for sure if I like it. When watching HD Cable TV, the old setting looks normal but the new setting seems to really show any noise or distortions more glaringly. I have adjusted the TV to the sharpness pattern so strictly already that I know if there is noise, it's because the source video just plain sucks.


I am evaluating the new adjustments still but wanted to run this by the calibrators for an opinion on what I've done. Could my adjustments to get the white & color clipping patterns to match up without effecting the main color test result give a just barely noticable downgrade of the color quality? Or have I done something worth keeping and maybe just aren't completely used to it?

This is from the pattern manual that goes with the AVS Disc:


"This clipping pattern is similar to White Clipping, but it splits the

vertical bars into red, green, and blue. The pattern allows you to

observe how white-level may affect colors on digital displays. Flashing

gray is included behind the lettering that labels the digital values. The

bars for 235 have been labeled as red, green, and blue to remind the

user that 235-251 are allowed to blend together or clip. You can use

this pattern to ensure that red, green, and blue are not blending

together for 219-233. If the range from 219-233 does not flash, you

may need to lower white-level so the entire range of colors will

display. Using this pattern you may find that red, green, and blue do

not necessarily clip at exactly the same levels, but you simply want to

make sure levels lower than 235 flash. Nearby levels can be difficult to

tell apart, so if the display never shows flashing for 235 or higher it

may be difficult to spot flashing at 233."


You don't need to balance them or attempt to use the Color or CMS settings to do so. The only setting this pattern is intended for is Contrast and you want 233 and below to flash for each primary.

 

[GeorgeAB]



Quote:
Originally Posted by Doug Blackburn /forum/post/20813632


You can NEVER trust the blue filter method of setting color and tint. If the red, green, and blue points aren't accurate, the blue filter method won't give accurate results... even if just 1 or 2 colors are "off" from the Rec. 709 standard coordinates, the blue filter method won't give accurate results. Furthermore, the inexpensive blue filters in cardboard holders are not necessarily "ideal" filters and error can creep in there also. There is no way for you to tell if your particular blue filter is good or not so good. There's also no way (without instrumentation) for you to tell how accurate your red, green, and blue points are. In fact, every properly written set of instructions for using the blue filter to set color and tint tells you to view content after setting the controls and if anything looks "off" color-wise to further adjust color and tint controls by eye until images look natural/normal. That's because the blue filter method isn't reliable. You can't just assume the color and tint settings you get using the blue filter are accurate.


And... you can NEVER... and I mean NEVER trust your eyes to see accurately. Because... your eyes will assume that the BLUE-EST shade of white is really the accurate white and accurate shades of white will appear to be too yellow when they are NOT too yellow. Other "tints" of color produce additional perceptual errors. You cannot stop these perceptual errors from happening. No matter HOW aware you are of their existence, the human visual system is always tricked into seeing the wrong things. This is why instrumentation is used for calibration. Instrumentation can't be fooled. All the adjustments you did by eye could easily have made the display LESS accurate rather than MORE accurate... there's probably a 75% chance the adjustments you made did not improve the accuracy of images (or made accuracy worse) and a 25% chance that you did improve images overall.


All this applies to looking at ramp patterns (stripes of shades of white from 80% to 110%) trying to use them to make a display more accurate by eye. You cannot change how human visual perception works even when you KNOW you can be fooled.

Sweet clarity!

 

[HDTVChallenged]



Quote:
Originally Posted by Doug Blackburn /forum/post/20813632


You can NEVER trust the blue filter method of setting color and tint. If the red, green, and blue points aren't accurate, the blue filter method won't give accurate results... even if just 1 or 2 colors are "off" from the Rec. 709 standard coordinates, the blue filter method won't give accurate results. Furthermore, the inexpensive blue filters in cardboard holders are not necessarily "ideal" filters and error can creep in there also. There is no way for you to tell if your particular blue filter is good or not so good. There's also no way (without instrumentation) for you to tell how accurate your red, green, and blue points are. In fact, every properly written set of instructions for using the blue filter to set color and tint tells you to view content after setting the controls and if anything looks "off" color-wise to further adjust color and tint controls by eye until images look natural/normal. That's because the blue filter method isn't reliable. You can't just assume the color and tint settings you get using the blue filter are accurate.

OTOH, let's not conflate "Color Decoding" setup with "Gamut Adjustment" ... Recently, it's been my experience that blindly trusting one's instrumentation might cause your color/tint/CMS setup to go horribly, horribly wrong ... then your eyes will scream at you for days that it's wrong and you still won't believe them because the instrumentation says it's setup "right."

With a proper pattern and usable color filters (namely ones built into the display) your eyes are a far better "tool." Granted, the "Blue" method *alone* is not sufficient ... you need to check all three colors.

 

[PlasmaPZ80U]



Quote:
Originally Posted by HDTVChallenged /forum/post/20813856


OTOH, let's not conflate "Color Decoding" setup with "Gamut Adjustment" ... Recently, it's been my experience that blindly trusting one's instrumentation might cause your color/tint/CMS setup to go horribly, horribly wrong ... then your eyes will scream at you for days that it's wrong and you still won't believe them because the instrumentation says it's setup "right."

With a proper pattern and usable color filters (namely ones built into the display) your eyes are a far better "tool." Granted, the "Blue" method *alone* is not sufficient ... you need to check all three colors.

...but you can't do CMS work without a meter and the proper patterns and the OP is misusing a pattern designed to visually set contrast to try to set color and CMS (which is completely wrong).

 

[HDTVChallenged]



Quote:
Originally Posted by PlasmaPZ80U /forum/post/20814414


...but you can't do CMS work without a meter and the proper patterns and the OP is misusing a pattern designed to visually set contrast to try to set color and CMS (which is completely wrong).

... And so is trying to make Y(luma) of red exactly 21.26% of the Y(luma) of White (etc, etc). That method only works if all three primaries of the display are exactly on the BT709 spec ... and the white point is *exactly* (0.3127,0.329)


If your primaries are off spec, you will have a different (off BT709 spec) mix of them to achieve D65 ... and all other colors will need to be translated by the same amount to get to the proper color (x,y.) A proper color decoder pattern + filters + (non colorblind) eyes takes care of that translation auto-magically.


You can still use a lux meter or colorimeter to assist, but you need to either target (x,y) points exclusively, *or* you need to figure out the precise mix (lumas) of the R,B,G primaries that makes up D65-white on your specific display (and I don't know anyway to do that without using color filters applied to the white target.) Edit: I suppose there may be a way to figure that out using one of the alternate coordinate systems.


What you *can't* do is simply measure Yw of white, then set Yr of red to 21.26% of Yw, Yg of green to 71.52% of Yw and Yb of blue to 7.22% of Yw and expect to have accurate colors.


PS: I realize that plasmas with aggressive ABL probably create an extra level (or two) of challenges for both human and solid state eyes.

 

[Doug Blackburn]



Quote:
Originally Posted by HDTVChallenged /forum/post/20813856


OTOH, let's not conflate "Color Decoding" setup with "Gamut Adjustment" ... Recently, it's been my experience that blindly trusting one's instrumentation might cause your color/tint/CMS setup to go horribly, horribly wrong ... then your eyes will scream at you for days that it's wrong and you still won't believe them because the instrumentation says it's setup "right."

With a proper pattern and usable color filters (namely ones built into the display) your eyes are a far better "tool." Granted, the "Blue" method *alone* is not sufficient ... you need to check all three colors.

Sorry, I'm not in agreement with ANY of this. If you have a big problem after using instrumentation you either don't know what you are doing or you haven't measured the right thing(s) to see what is causing the error. If you do the RIGHT measurements with a meter that is working properly, you will find EVERY error. If you use a single point for CMS adjustments.... say 75% window patterns, you will never see big CMS problems at 100% or 25% because you never MEASURED those other stimulus levels (and calibration software typically won't do this, you have to manually set the software to measure levels other than 75% (or whichever level you selected). Likewise, if your software is measuring 10 grayscale points and the results look good, but there's still a visible grayscale problem, more than likely you would FIND that problem with measurements that have more than 10 points measured... like 20 points.


The meter will ALWAYS produce the proper results if you use it properly (so that visible problems aren't missed) and measure the right things. And the meter has to be CAPABLE of accurate grayscale AND color measurements (some low-cost meters are essentially unusable for color measurements, or they may be OK with one type of display but not other types).


If you get bad results when using a meter, it is only because you haven't measured the right thing(s) yet to discover the error(s).


On the other hand, your eyes are WORTHLESS for "calibration" because they are so easily fooled. The internet is LOADED with black/white and color optical illusions that illustrate quite well why you CANNOT trust your eyes. These are not "subtle" illusions either -- one of them has a yellow feature that appears gray until you block everything but the yellow area, only then can you see that it really is yellow. These illusions NEVER fool a meter... but you do have to measure the right thing to determine that there is an illusion.

 

[HDTVChallenged]



Quote:
Originally Posted by Doug Blackburn /forum/post/20817077


Sorry, I'm not in agreement with ANY of this. If you have a big problem after using instrumentation you either don't know what you are doing or you haven't measured the right thing(s) to see what is causing the error.

Please read the post immediately above yours, before you assume that I don't know what I'm talking about.

I agree that human eyes are completely useless to achieve D65 (without a reference, such as an optical comparator.) OTOH, they're pretty darn good at setting color/tint, with the proper assistance ... and way faster than using a colorimeter.

PS: Yes I've run the full suite of measurements (including 0-100% color saturation "curves") on my set ... and guess what, my eyes + built-in "filters"(R,B and G) + color decoder pattern were 100% correct (within the limitations of the set's available adjustment range.)

PPS: Yes I admit that I *originally* was measuring the the right parameter the wrong way ... I suspect that many, many folks have made the exact same mistake. Hence, my cautionary post. Capish??