[RyanHomsey]

On my LG 42LD520 I calibrated for very tight deltaE's in the primary/secondary readings... got the delta luma very tight as well.


Thought I was done until I ran across the calibration aid spreadsheet here and plotted my saturation points. They were WAY off, deltaE's around 50.


Obviously whatever the LG CMS system does (color and tint controls for primaries/secondaries), when I bring those primaries/secondaries in tight it is whacking out the saturation points.


It seems I have to decide between tight primary/secondary DeltaE's or tight saturation point DeltaE's. My current calibration balanced the two, all saturation point DeltaE's are generally 2 or less... and the primary/secondary deltaE's are fairly good. I'm not satisfied with Red or Blue however, both have deltaE's around 5. It just seems like when I get those in tighter, especially the deltaLuma, it throws the saturation points off.


So do I get those primary/secondary deltaE's tighter at the expense of saturation points? Or do I leave it where its at with a good balance. Isn't a deltaE of 5 for red not very good?


Primary/Secondary Chart for current calibration with very good saturation points:

CIE:

And take a look at how far off the saturation points got when I adjusted for tight delta luma primary/secondary:

 

[RyanHomsey]

I'd also like to note that I probably have 15+ hours in calibrating this television... it's been a journey of learning for sure. But I'm started to go in circles and get frustrated at this point. I very much want to sit back and enjoy the beautiful picture I've achieved. Just not sure if I'm happy with the DeltaE's on Red/Blue/Magenta and whether or not I should be worrying so much about that saturation points.

 

[alluringreality]

I prefer delta E '94 calcuations for color, and I generally consider the color delta E numbers from ColorHCFR somewhat worthless. Anyway, I'm not sure you're going to get a clear response on this one. Most people don't go to the level of detail in using something like the saturation measurements. Honestly I wasn't aware that some displays might perform that way. If the closer saturation measurements are an improvement over the default settings I would probably lean in that direction. My take is simply that there's a certain amount of subjectivity in the items that generally fall under the umbrella of calibration, so without seeing the default operation the measurements look reasonable.

 

[RyanHomsey]

My research has seemed to indicate that, like you said, people dont typically get down into the saturation points. I just happened to plug them in on a whim when I found the excel spreadsheet and was shocked to see how off they were visually on the CIE chart. I took a look at the saturation DeltaE's in HCFR and they revealed the same. I'm certainly glad I looked at them and adjusted to get things back in check.


One thing I've seen others note is that the 100% saturation points seem to be locked down on the LG. So, it appears that when I adjusted the 'color' setting (which seems to adjust luminance+saturation) using 100% saturation boxes to get the deltaLuma in check it was moving the lower saturation points (luminance+saturation) and only changing luminance for the 100% saturation. I guess, if thats the case, then the CMS on this LG is mostly useless for adjusting luminance . I tried adjusting using 75% saturation and that made things FAR worse.


My final approach was to make minute adjustments to the 'color' setting to bring the primaries closer to the desired luminance, basically going as close as I can before it began to significantly affect the lower saturation points. Secondaries were set (luminance) using 'color' based on the 100% white luminance and then balancing using the 'tint' setting. (Just an FYI for other LG owners looking for help when using the CMS).


Thanks for your input alluringreality.

 

[Doug Blackburn]

uvL is STILL the most perceptually uniform color space and using dEuv (in CalMAN, can't say wether this is offered in other calibration software or not. uvL is hue, saturation, Luminance. dEs can be 1 dimensional (1 of the coordinates only), 2 dimensional (2 coordinates with the 3rd ignored), or 3 dimensional (errors for all 3 color coordinates combined into a single dE value. I find looking at single-coordinate dEs (like luminance or saturation) unnecessarily complicates the whole calibration process. When you look at dE for color, having ALL errors in ALL coordinates combined into a single number (as dEuv does in CalMAN... I can't say if other software works the same way) stops you from trying to get low dEs for 3 separate dE calculations. The whole point is to get the errors as low as possible. A dEuv below 2 is going to give errors so small for moving video images that they can't be seen (certain stationary test targets might reveal measurable errors using a meter, but the errors should be so small they can't be detected while viewing a movie).


That said, grayscale dEs can also be expressed in dEuv however CalMAN really uses ONLY the u and v coordinates for this calculation for grayscale. The Luminance coordinate comes into play when you work on finding the most linear gamma. This is an appropriate way to deal with grayscale for many video displays. But when you are using a Lumagen Radiance processor that has adjustable luminance for each step as well as RGB balance for each step you could easily get away with a single dE number using all 3 coordinates if the grayscale dEs were based off the 100% white measurement.


So in CalMAN you can see a graph of saturation errors (dEv), hue errors (dEu), and Luminance errors (dEL). Or you could display a graph showing all 3 errors in a single calculated number. You could look at the 3 graphs and try to make each color as low as possible... but it gets annoying because one error might go down while the other 2 errors might go up and you have to be able to keep all that in your head. It gets to be a pain in the butt. With just a single dEuvL for each color, you are assured when you get the lowest dE number, you have found the spot with the lowest overall errors for all 3 coordinates. Simplifies the whole calibration process. Some TV controls fiddle with more than one of the coordinates so you can't watch just a single-coordinate for any adjustment (say, the TV has a control that adjusts saturation up and down... it might move hue or luminance a little bit while it is moving saturation so you ALWAYS have to be aware of what happened in all 3 graphs if they are single-coordinate errors).

 

[PlasmaPZ80U]

I think this article describes why dE94 is best for gamut work.

 

[PlasmaPZ80U]



Quote:
Originally Posted by Doug Blackburn /forum/post/20853122



That said, grayscale dEs can also be expressed in dEuv however CalMAN really uses ONLY the u and v coordinates for this calculation for grayscale. The Luminance coordinate comes into play when you work on finding the most linear gamma. This is an appropriate way to deal with grayscale for many video displays. But when you are using a Lumagen Radiance processor that has adjustable luminance for each step as well as RGB balance for each step you could easily get away with a single dE number using all 3 coordinates if the grayscale dEs were based off the 100% white measurement.

Actually, dE76 in CalMAN is LUV and there is also an option under miscellaneous settings to factor in whatever target gamma is specified.

 

[Doug Blackburn]



Quote:
Originally Posted by PlasmaPZ80U /forum/post/20853341


Actually, dE76 in CalMAN is LUV and there is also an option under miscellaneous settings to factor in whatever target gamma is specified.

????


dE76 uses unique calculations (as determined by the CIE) to express dE... it's not the same as dEuv or dEuvL. dEuv doesn't apply the CIE formulas to dE calculations. they are 2 different things and CalMAN does have dEuv as a selectable option which does NOT include the CIE's "spin" on perceptual calculations for errors in uv/uvL space. CalMAN's dEuv option uses uv to calculate grayscale errors (gamma measured separately) and uvL to calculate color gamut dEs (as a single number for all 3 coordinates - without CIE's "spin" on what errors are more or less visible). dEs calculated by dE76 and dEuv are not the same. CalMAN offers both choices, I prefer dEuv based on decades of work on professional imaging systems.

 

[PlasmaPZ80U]



Quote:
Originally Posted by Doug Blackburn /forum/post/20858342


????


dE76 uses unique calculations (as determined by the CIE) to express dE... it's not the same as dEuv or dEuvL. dEuv doesn't apply the CIE formulas to dE calculations. they are 2 different things and CalMAN does have dEuv as a selectable option which does NOT include the CIE's "spin" on perceptual calculations for errors in uv/uvL space. CalMAN's dEuv option uses uv to calculate grayscale errors (gamma measured separately) and uvL to calculate color gamut dEs (as a single number for all 3 coordinates - without CIE's "spin" on what errors are more or less visible). dEs calculated by dE76 and dEuv are not the same. CalMAN offers both choices, I prefer dEuv based on decades of work on professional imaging systems.

I am referring to this:

http://www.spectracal.com/forum/view...&t=2401#p15055


Also, I agree dEuv is best for grayscale since it treats all color errors equally regardless of % stimulus. dE76 is biased toward the high end.