Methods to Optimize Color and Tint Settings without CMS Controls by Using deltaE’s (d - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #1 of 62 Old 10-16-2009, 08:23 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Methods to Optimize Color and Tint Settings without CMS Controls by Using deltaE's (dE)

I have been following this forum for over a year and read many calibration posts and articles. The question I have is related to calibrating the COLOR and TINT user settings. I have tried unsuccessfully to use the color filters (included in either DVE or AVIA standand DVDs) when setting both Color and Tint. My complete video system and settings for each component are described below.

What methods or tradeoffs are taken into account when attempting to set both Color and Tint with TV's that do not have complete CMS controls? In my particular case, I found that by minimizing the deltaE's for RED in HCFR gave me the best results for COLOR=33 setting. Initially, I left TINT=0(Default) for many weeks, then I decided to look at more closely at possible errors with the HCFR program and the EyeOne Display meter. This resulted in the data plotted below for the limited range of TINT that I thought would include the proper Tint setting.

Let me explain how the data graph was created. All menu settings were held constant except for the TV's User Menu Tint setting. This parameter was varied between TINT=0 and TINT=-10 in 1 increments. A HCFR Color test was run for each of the 11 different TINT settings. After all the data files were taken I extracted the Luminance and deltaE data and plotted the combined data of all 11 Color test runs into a single Excel plot that is displayed below. So the data plot shows the effect of changing Tint settings with respect to deltaE (blue/green bars) and luminance (red/orange bars) for each primary and secondary color. The RED number labeled deltaE bars are my old (Tint=0=default) setting. The new Tint=-5 setting is labeled in BLACK numbers. So, the delatE's for Primary Colors => R=5.3 to 5.9, G=18.2 to 15.6 (slight decrease) and B=34.5 to 35.2 remained basically constant. The deltaE's for the Secondary Colors Yellow=7.6 to 16.9, Cyan=14.7 to 17.0 and Magenta=35.9 to 26.4 change either increasing or decreasing with the tint changes.

Conclusions from studying the combined color data plot are:
1) Luminance for R, G, B and Y are Constant over the Tint change range.
2) Luminance for Cyan and Magenta change slightly for the changing Tint.
3) deltaE for R,G and B are basically unchanged with changing Tint (as it should be).
4) deltaE for Y,C and M do change with changing Tint (as expected).
5) deltaE for Y and C increased with decreasing Tint (more Red).
6) deltaE for M decreased with decreasing Tint (more Red).

Is it better to have a lower deltaE for Yellow and Cyan and accept the higher deltaE error for Magenta (the RED labeled bars) or try to reduce Magenta's error by increasing both Yellow and Cyan's errors (the Black labeled bars)?

This is a rather long and detailed question but I hope some of the more experienced professional calibrators and serious videophiles might provide feedback and rationales as to whether this method or results are more accurate or could be improved over the color filter method with the equipment available and limitations of my system.



Plasma54321 is offline  
Sponsored Links
Advertisement
 
post #2 of 62 Old 10-16-2009, 08:47 AM
AVS Club Gold
 
Bear5k's Avatar
 
Join Date: Feb 2006
Location: Houston, TX
Posts: 2,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
You are using HCFR, so I'm not going to get into too much detail. But here is a foundational set of questions:

Do you know what the gamut of your display looks like? In other words, have you measured the locations of your primaries? Are you using your actual gamut to create your targets?

Since you are showing positive color error in your measurements of your primaries, my guess is that the answers to my above questions are "no". If so, then your methodology is going to be challenged going this route. Your goal is to minimize your overall color error (dE), but using the decoder controls (color/tint) to move around virtual primaries is not the right way to go about it. Measure your primary locations first, then create a custom gamut using your actual physical primaries. Once you have that, then use your decoder controls to minimize the color error in your secondaries. Chances are that your decoder is fine.

Bill

Color accuracy evangelist and CalMAN insider
Bear5k is offline  
post #3 of 62 Old 10-16-2009, 09:23 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I'm not a pro calibrator just a home tinkerer. So you may have lost me already, but I'll try.

Do you know what the gamut of your display looks like? In other words, have you measured the locations of your primaries? Are you using your actual gamut to create your targets?

Are you saying where are my primaries located on the CIE diagram?

Am I using actual RGB locations to get the Secondaries? No. How would I do that? What effect would that have on PQ? How do I check the present location for accuracy as you described.
Plasma54321 is offline  
post #4 of 62 Old 10-16-2009, 09:42 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Additional thought/question regarding your comment.

Since you are showing positive color error in your measurements of your primaries

I was not aware that deltaE could be anything other than a POSITIVE value. Can it be NEGATIVE?
Plasma54321 is offline  
post #5 of 62 Old 10-16-2009, 12:30 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I was just rereading "Basic Guide to Color Calibration" by Tom Huffman. In it he states that different color errors are perceived differently depending on the color. For example RED errors are worse than BLUE errors.

Is there a ordered list of what colors should have the lowest deltaE?

Then you could optimize within the limitions of your components the Color and Tint settings.
Plasma54321 is offline  
post #6 of 62 Old 10-16-2009, 12:37 PM
AVS Club Gold
 
Bear5k's Avatar
 
Join Date: Feb 2006
Location: Houston, TX
Posts: 2,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Plasma54321 View Post

Are you saying where are my primaries located on the CIE diagram?

Yes.

Quote:


Am I using actual RGB locations to get the Secondaries? No. How would I do that? What effect would that have on PQ? How do I check the present location for accuracy as you described.

It would not have an effect on PQ, but it will have an effect on things like this. There are two schools of thought on setting the color decoder:

School 1: Make the decoder accurate - This is where I fall. The goal of the decoder is to convert YCbCr into RGB, so you want to use your actual primary locations when trying to measure and minimize the color error in your secondaries.

School 2: Make the colors "accurate" - This is a school that advocates making the primaries and secondaries as accurate as possible (minimizing dE overall).

As for the mechanics of how you do this, etc., you will need to get help from the HCFR community for that. You may want to download the CalMAN demo app and take a look at the help files in it. You may find they help you out a fair bit.

Bill

Color accuracy evangelist and CalMAN insider
Bear5k is offline  
post #7 of 62 Old 10-16-2009, 02:41 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
You can use the accupel calculator to find out what your secondaries should be.
http://www.accupel.com/Display%20Cal...20%28PC%29.zip

You can feed it the information off HCFR and then do your primaries and secondaries.

After spending hours doing that figuring out my TV, I can safely say using calman w/ the built-in target editor will feel like quite the luxury.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #8 of 62 Old 10-16-2009, 02:57 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
The problem that I have with looking at the CIE Diagram that it seems that it can mislead one, because it does not include luminance data. Therefore, if you align to the x and y coordinates - that will not necessarily give you a minimum deltaE for the color. So I tend to ignore it and have been only looking at deltaE's for either Primary or Secondaries.

If I get what you are saying about School 1: Wherever the best Primaries are located the Secondaries should be located with respect to the actual Primaries and not the Reference Secondary locations.

For Shool 2: Minimizing deltaE's withinn the limitations of the TV being calibrated creates a dilema. Which color is more important? Is there an order for colors that we "see" more error or are less sensitive to.
Plasma54321 is offline  
post #9 of 62 Old 10-16-2009, 03:07 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,640
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 175
You're absolutely correct about #1.

But the thought process there is that by calculating a new gamut off of the physical location of your primaries, you'll get a more accurate luminance target for where your primaries actually are.

So you may find that you can do step 1 minimize DE for the rec.709 primaries, step 2 use those values to calculate what you current primary luminance target should be and what where your secondaries should be. Then you can repeat step two until you get as uniform a response as you can.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #10 of 62 Old 10-16-2009, 03:41 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I'll have to download and try out the Color calculator that you suggest, to see if the values of TINT that I have selected is in/on the actual Primary color triangle.

Still I'm wondering whether there is a color hierarchy to aim for? One that matches what the eyes are actual capable of detecting. In other words, what color is the eye most sensitive to or least sensitive.
Plasma54321 is offline  
post #11 of 62 Old 10-17-2009, 10:12 AM
AVS Special Member
 
alluringreality's Avatar
 
Join Date: Jul 2006
Posts: 3,203
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 46 Post(s)
Liked: 40
The Color Corrector spreadsheet page from http://www.avsforum.com/avs-vb/showt...ighlight=excel is easier to copy and paste with ColorHCFR, rather than having to enter information in the Accupel calculator.
anta1974 likes this.
alluringreality is offline  
post #12 of 62 Old 10-17-2009, 12:54 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by alluringreality View Post

The Color Corrector spreadsheet page from http://www.avsforum.com/avs-vb/showt...ighlight=excel is easier to copy and paste with ColorHCFR, rather than having to enter information in the Accupel calculator.

I'll check it out.
Plasma54321 is offline  
post #13 of 62 Old 10-17-2009, 02:50 PM
AVS Club Gold
 
Bear5k's Avatar
 
Join Date: Feb 2006
Location: Houston, TX
Posts: 2,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by sotti View Post

But the thought process there is that by calculating a new gamut off of the physical location of your primaries, you'll get a more accurate luminance target for where your primaries actually are.

Even more simplistically, if you use the physical gamut as the basis for your targets, and you have a good grayscale calibration, then if your secondaries are aligned with where they are supposed to be, then you know that the decoder is accurate in the technical sense (school #1). What you do from there is up to you.

Color accuracy evangelist and CalMAN insider
Bear5k is offline  
post #14 of 62 Old 10-17-2009, 05:30 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I looked at the Excel Color Corrector Calculator Tab Sheet. If I understand the method here, I would take my TINT=0 readings and then calculate the modified/new Secondary targets for Yellow, Cyan and Magenta. I then could check all the data previously taken to match the modified target x and y data to get the best fit for the TINT setting. So referencing the acutal data below for closest target values (in RED) for each color:

Optimized Yellow => TINT= 0
Optimized Cyan => TINT= -8
Optimized Magenta => TINT = -2

Then once again, back to that question which color should be optimized over the others because in this particular case all can not be accomidated.

I guess I'll try the TINT= -2 because Magenta has RED and I would want to minimize that color's error. Any opinions out there?

So my actual x,y,Y data for TINT=0 is:
HCFR RED Green Blue Yellow Cyan Magenta White
x 0.665 0.266 0.145 0.431 0.199 0.362 0.319
y 0.325 0.633 0.068 0.505 0.329 0.174 0.342
Y 5.409 20.494 1.415 27.455 20.378 6.445 31.530

New Calculated Secondary Targets:
Yellow Cyan Magenta
0.426 0.205 0.350
0.510 0.348 0.170
29.148 24.098 9.813

All other TINT data from [-1..-10]
TINT = -1
0.665 0.268 0.146 0.434 0.200 0.356 0.319
0.325 0.633 0.068 0.503 0.334 0.172 0.342
5.409 20.184 1.401 27.196 20.684 6.450 31.530

TINT = -2
0.663 0.271 0.146 0.435 0.201 0.352 0.320
0.326 0.634 0.068 0.501 0.336 0.170 0.342
5.460 20.132 1.396 27.454 20.837 6.349 31.478

TINT = -3
0.666 0.271 0.146 0.438 0.202 0.348 0.319
0.323 0.636 0.068 0.499 0.338 0.167 0.342
5.461 20.183 1.391 27.609 21.198 6.301 31.426
TINT = -4
0.663 0.272 0.145 0.439 0.201 0.347 0.319
0.325 0.635 0.068 0.498 0.339 0.167 0.342
5.512 20.234 1.393 27.764 21.251 6.302 31.479
TINT = -5
0.664 0.272 0.146 0.440 0.203 0.340 0.319
0.323 0.637 0.068 0.498 0.342 0.164 0.342
5.512 19.975 1.401 27.454 21.245 6.309 31.530
TINT = -6
0.662 0.271 0.146 0.443 0.202 0.338 0.319
0.325 0.638 0.068 0.495 0.342 0.163 0.342
5.564 20.027 1.398 27.454 21.298 6.205 31.479
TINT = - 7
0.660 0.272 0.146 0.444 0.203 0.328 0.319
0.326 0.638 0.068 0.494 0.344 0.158 0.342
5.461 19.975 1.390 27.609 21.242 6.003 31.479
TINT = - 8
0.663 0.271 0.146 0.445 0.204 0.321 0.319
0.321 0.642 0.068 0.493 0.347 0.154 0.342
5.461 19.974 1.389 27.609 21.238 5.906 31.479
TINT = - 9
0.662 0.271 0.146 0.449 0.205 0.313 0.319
0.321 0.644 0.069 0.490 0.357 0.150 0.341
5.462 20.025 1.390 27.351 21.746 5.810 31.427
TINT = -10
0.660 0.272 0.146 0.452 0.205 0.304 0.319
0.321 0.645 0.069 0.488 0.359 0.147 0.341
5.513 19.972 1.394 27.402 21.742 5.818 31.427
Plasma54321 is offline  
post #15 of 62 Old 10-18-2009, 02:23 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
After thinking some more on this whole COLOR/TINT optimization experiment, I thought I would summarize everything so far. I basically started this process soon after getting my Panasonic TH-42PZ85U HDTV, Oppo DV980H DVD Player, Avia and VE Calibration DVDs last year. I tried a number of times to adjust both color and tint with the color filters from both of the test discs mentioned and never got what I considered good results. Red never seemed right on faces and colors seemed overblown.

So I perused many posts and articles on calibration - a lot of information but I did not find what I needed to get the skin tones right for all sources - cable TV, DVDs and later BluRay. This eventually led to getting more analytical with an X-rite EyeOne Display LT calorimeter and downloading HCFR software (from reading about them on AVS forums.) After many grayscale/color data runs and adjustment attempts I now had a ton of data to look at but digesting and understanding it all was confusing and frustrating.

I finally decided that I was most interested in getting the flesh tones right by trying to minimize the RED deltaE readings over all the others. Remember, Panasonic 42PZ85/80 TVs do not have a complete set of CMS controls, probably the same situation as most popular TVs purchased. Therefore, the color adjustments are limited to only COLOR and TINT settings in the User Menus. So I found the best COLOR setting at my calibrated Brightness and Contrast setting was at minimum deltaE for RED. I got to that point by using HCFR and varying the color settings from COLOR=0 to 50 by 5's and plotting all the data. This is the same method as I have done with the TINT Change Study Plot already posted in this thread.

However, now I had the TINT setting to adjust or optimize. This is why I posted this thread. So far it seems that some have suggested setting the Secondary colors at modified targets based on the actual data readings for RGB (Primaries) with my present finalized COLOR=33 setting. This if I understand the suggestion will minimize the decoder errors and give better results than the actual absolute reference targets. This results in a TINT=-2 setting based on the previously taken x, y color data (included above). I am now evaluating this setting with actual program material.

I originally tried leaving the TINT=0 (default) for a long time and did not find anything objectionable PQ issues, however I wondered if another setting would be more accurate for my system. This led me to plot the TINT range data and arbitrarily choose to equalize the Secondary deltaEs - thereby increasing Yellow and Cyan errors in an attempt to decrease the Magenta deltaE error. This resulted in the TINT=-5 setting indentified in the TINT data plot.

The conclusion that I am reaching seems to be a skin tone color optimization process. This involves minimizing RED deltaE error with an initial TINT = 0 setting, then minimize other flesh color errors like yellow then magenta (possibly with the modified Secondary targets). The remaining colors - blue, green and cyan will then fall where they will. This method then gives preference to natural skin colors as the best total color accuracy within the limitations of the system and controls to adjust it.
Plasma54321 is offline  
post #16 of 62 Old 10-18-2009, 05:01 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
If you don't want to use dE, and your unmodified primaries measure

x y
R 0.665 0.325
G 0.266 0.633
B 0.145 0.068

Then the brightness targets are

R 0.235
G 0.681
B 0.084
Y 0.916
C 0.765
M 0.319

and the secondary targets are

x y
Y 0.427 0.507
C 0.201 0.331
M 0.338 0.162

Use the Color control to get as close to those brightness targets as you can and use the Tint control to get as close to those secondary xy targets as you can.

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

TomHuffman is offline  
post #17 of 62 Old 10-19-2009, 02:40 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by TomHuffman View Post

Then the brightness targets are

R 0.235
G 0.681
B 0.084
Y 0.916
C 0.765
M 0.319

and the secondary targets are

x y
Y 0.427 0.507
C 0.201 0.331
M 0.338 0.162

I'm not quite sure how the brightness target values would be used?

The YWhite=31.53 from my TINT=0 data table values.

It looks like the modified Secondary x,y locations are close to the calculations that I did. They seem to match within 0.01 for either x or y.
Plasma54321 is offline  
post #18 of 62 Old 10-19-2009, 03:35 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
Quote:
Originally Posted by Plasma54321 View Post

I'm not quite sure how the brightness target values would be used?

The YWhite=31.53 from my TINT=0 data table values.

Multiply 31.53 with those values, e.g., red = 7.41.

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

TomHuffman is offline  
post #19 of 62 Old 10-20-2009, 11:25 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Tom, I have reread your "Basic Guide to Color Calibration" for the nth time. Each time I read it I understand something a little more. It seems to me that in your discussion of Color and it's characteristics - luminance or brightness level is important to getting both a measured accuracte color and a visually perceived accuracy when viewing with your eyeballs. Did I get is right?

I have been looking at my data with dE Calculator from ChromaPure that outputs all the various deltaEs. In my case the base starting point for TINT=0 had following

Lightness Errors:
Red=-9%, Green=-3.7%, Blue=-21.9%,
Yellow=-2.4%, Cyan=-7.5% and Magenta= -13.2%.

Does anything in this data indicate a "red push" or other anomaly that is built into the TV?

Why is your preference for deltaE calculations the CIE94 method? What does it catch as errors that the others do not? What would you see subjectively in program material?

PS. Thanks for creating and making available the Basic Calibration DVD. I have been using it almost exclusively for use with HCFR software. It's much faster and less error prone than the others.
Plasma54321 is offline  
post #20 of 62 Old 10-20-2009, 01:13 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
Quote:
Originally Posted by Plasma54321 View Post

It seems to me that in your discussion of Color and it's characteristics - luminance or brightness level is important to getting both a measured accuracte color and a visually perceived accuracy when viewing with your eyeballs. Did I get is right?

Yes. The brightness of the color is one of its characteristics, just like hue and and saturation.

Quote:
Originally Posted by Plasma54321 View Post

I have been looking at my data with dE Calculator from ChromaPure that outputs all the various deltaEs. In my case the base starting point for TINT=0 had following

Lightness Errors:
Red=-9%, Green=-3.7%, Blue=-21.9%,
Yellow=-2.4%, Cyan=-7.5% and Magenta= -13.2%.

Does anything in this data indicate a "red push" or other anomaly that is built into the TV?

Several things:

1. The brightness of the color will be affected by the Color control, not the Tint control.

2. The % numbers you list above are relative to a selected standard gamut in the spreadsheet, which has fixed values. The targets I gave in the previous post were relative to your own measured gamut, which is not equal to any standard gamut. Since you have no CMS that would allow you to achieve a standard gamut, that's the targets I would use.

3. Those numbers certainly don't show "red push". They show red down 9%, but again that's relative to a standard gamut. Using the gamut you gave me and the white measurement, the red target is 7.41. In fact, all of your primaries are too dim. Raise the Color control.

Quote:
Originally Posted by Plasma54321 View Post

Why is your preference for deltaE calculations the CIE94 method? What does it catch as errors that the others do not? What would you see subjectively in program material.

Experience. It does a better job of predicting what are to my eyes a more accurate mix of saturation and brightness than either of the 1976 formulas. The 2000 formula is absurdly complicated, which makes it suspect in my mind, and has never been widely adopted outside of the textile industry (probably just because it is so complicated to implement).

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

TomHuffman is offline  
post #21 of 62 Old 10-20-2009, 03:06 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by TomHuffman View Post

Yes. The brightness of the color is one of its characteristics, just like hue and and saturation.

I guess I wasn't clear. I'm trying to get a sense for what factors and what relative importance they have to each other for optimization. For instance the order of importance that I would place on these parameters based on your post:

1) Brightness = L => Most Important
2) Hue = H
3) Saturation = S => Least Important
Plasma54321 is offline  
post #22 of 62 Old 10-20-2009, 03:24 PM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by TomHuffman View Post

1. The brightness of the color will be affected by the Color control, not the Tint control.

2. The % numbers you list above are relative to a selected standard gamut in the spreadsheet, which has fixed values. The targets I gave in the previous post were relative to your own measured gamut, which is not equal to any standard gamut. Since you have no CMS that would allow you to achieve a standard gamut, that's the targets I would use.

3. Those numbers certainly don't show "red push". They show red down 9%, but again that's relative to a standard gamut. Using the gamut you gave me and the white measurement, the red target is 7.41. In fact, all of your primaries are too dim. Raise the Color control.

1) The TINT=0 data set was my final COLOR=33 setting used as a starting point for all the tint changes. The included chart shows that the Primary Color Luminance= brightness is CONSTANT for all the Tint settings (0 to -10 by 1).

2) I don't think I understand how I would use your modified targets for the Primary Colors. The TINT=0 starting point is for COLOR=33 fixed. At this point I was only changing TINT values (Secondary colors) trying to find an optimum.

3) I sort of was backtracking with this question. In other words, the starting point for the TINT changes was where I set the COLOR=33 for minimum deltaE. This resulted in the "low luminance" values for all colors including RED which was the color that I was monitoring to minimize it's deltaE.
Plasma54321 is offline  
post #23 of 62 Old 10-20-2009, 04:49 PM
AVS Special Member
 
alluringreality's Avatar
 
Join Date: Jul 2006
Posts: 3,203
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 46 Post(s)
Liked: 40
Quote:
Originally Posted by Plasma54321 View Post

to minimize it's deltaE.

If your primaries are very far off standard I don't think it makes sense to use the '76 delta E calculation in this way of looking for a minimum delta E. For example the green on my SXRD measured around x=0.29 and y=0.70, so basically the tint of green was about right but the primary was clearly oversaturated. If you use the dE calc for color from ColorHCFR in this manner it will suggest a minimum dE happens if Y is around 0.49, so in order to compensate for the oversaturation it's simply suggesting a much lower Y. Personally I found the suggested low Y to have little resemblance at all to a display with primaries closer to Rec 709. Like commented in the thread, my take is that it makes more sense to either target a color space based on your measured primaries (spreadsheet or Accupel calculator) or else possibly use dE '94 which doesn't deal with Y in the same manner.
alluringreality is offline  
post #24 of 62 Old 10-20-2009, 05:00 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
Quote:
Originally Posted by Plasma54321 View Post

1) The TINT=0 data set was my final COLOR=33 setting used as a starting point for all the tint changes. The included chart shows that the Primary Color Luminance= brightness is CONSTANT for all the Tint settings (0 to -10 by 1).

2) I don't think I understand how I would use your modified targets for the Primary Colors. The TINT=0 starting point is for COLOR=33 fixed. At this point I was only changing TINT values (Secondary colors) trying to find an optimum.

3) I sort of was backtracking with this question. In other words, the starting point for the TINT changes was where I set the COLOR=33 for minimum deltaE. This resulted in the "low luminance" values for all colors including RED which was the color that I was monitoring to minimize it's deltaE.

If your question is how to get the lowest dE values using just the color/tint controls, that's an empirical question: adjust both until your get the lowest average dE. I would weigh the dE for red higher than green or blue. I have no opinion about how to weigh the secondaries. Of course, what setting yields the lowest dE depends upon which dE formula you use. CIE94 can yield radically different results in just the type of case you present where you lower brightness to compensate for oversaturation.

If you are asking how you can use the color/tint controls to yield what may be the most natural color, irrespective of dE, then use the targets I provided.

Other than that, I guess I am no longer entirely sure what you are asking.

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

TomHuffman is offline  
post #25 of 62 Old 10-22-2009, 11:16 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Well, after much thought and a headache or two studying both Tom Huffman and alluringReality's comments. Please, consider I've only used the HCFR software and EyeOne Display LT meter on exactly 3 LCD monitors for computers and one HDTV the Panasonic 42Z85 I have been referring to.

Here's my best shot at both trying to understand, interpret and apply your suggestions regarding getting a more color accurate TV setting with only COLOR and TINT controls available. Up to this point I had considered the minimizing of RED and flesh tone colors (yellow and magenta) deltas the best route.

Now I'm evaluating your suggestions regarding setting the RED luminance to a modified target. What I did was change the COLOR control from 33 to 40 while trying to set the Luminance Error = 0% for RED. This resulted in the data below.

Did I interpret your suggestions correctly? Is this data more color accurate than before? Or did I miss the boat?

Plasma54321 is offline  
post #26 of 62 Old 10-22-2009, 01:30 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
Quote:
Originally Posted by Plasma54321 View Post

Did I interpret your suggestions correctly? Is this data more color accurate than before? Or did I miss the boat?

What do the numbers tell you? If I look at CIE94, I see

Before After
R 5.4 3.7
G 5.7 5.2
B 8.8 3.0
Y 3.2 4.1
C 7.8 6.3
M 9.6 3.6

Green is slightly better, yellow is slightly worse, but red, blue, and magenta are dramatically better. Now adjust the white point and Tint to improve the secondary hues and you'll have about as good as you can get on this display.

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

TomHuffman is offline  
post #27 of 62 Old 10-23-2009, 07:34 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
To summarize the COLOR data to this point (tables below) are HCFR data and
ChromaPure deltaEs. Both the starting point COLOR=33 TINT=0 and the RED
Luminance=0% optimization setting of COLOR=40. Notice there are differences in the calculation results for the CIELUV values in HCFR and the ChromaPure calculators. Inparticular,

HCFR Yellow BEFORE dE=7.6 and
chromaPure Yellow Before (top below) dE=8.1

It seems to me these should agree better than this unless they are calculating the standard differently.

Today, I am attempting to create another TINT Range chart for COLOR=40 setting for comparison with the previous chart in this thread (above).

BEFORE => Baseline Color=33 Tint=0 CMS=On 10-12-2009
-------------------------------------------------------------
HCFR RED Green Blue Yellow Cyan Magenta White
x 0.665 0.266 0.145 0.431 0.199 0.362 0.319
y 0.325 0.633 0.068 0.505 0.329 0.174 0.342
Y 5.409 20.494 1.415 27.455 20.378 6.445 31.530
deltaE 5.3 18.2 34.5 7.6 14.7 35.9
deltaxy 0.025 0.047 0.009 0.011 0.025 0.046
dLuma -19.3 % -9.1 % -37.8 % -6.2 % -17.9 % -28.2 %


AFTER => Luminance Modification R and Y Color=40 Tint=0 CMS=On 10-21-2009
--------------------------------------------------------------------------
HCFR RED Green Blue Yellow Cyan Magenta White
x 0.667 0.266 0.145 0.439 0.197 0.344 0.320
y 0.324 0.661 0.067 0.518 0.326 0.164 0.341
Y 6.491 20.434 2.210 27.804 21.219 8.004 30.594
deltaE 20.1 24.3 7.0 11.1 17.9 17.9
deltaxy 0.027 0.070 0.008 0.024 0.028 0.025
dLuma -0.2 % -6.6 % = -2.0 % -11.9 % -8.1 %

chromaPure deltaEs (top=BEFORE and bottom=AFTER)

Plasma54321 is offline  
post #28 of 62 Old 10-23-2009, 09:27 AM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
Actually, a difference of 0.5 is negligible. It is entirely due to rounding differences. You typed 0.431, 0.505. for yellow, but the actual figure could be +- .0004, which would account for a 0.5 dE difference. You can see the actual value in HCFR by hovering your mouse over the cell in edit mode.

BTW, I found a small error in the spreadsheet that affected CIE94 calculations only. So, for example, in the original version the CIE94 Before value for green was 5.1. I updated the spreadsheet, so now it is 5.7, which is the correct value.

Quote:
Originally Posted by Plasma54321 View Post

Notice there are differences in the calculation results for the CIELUV values in HCFR and the ChromaPure calculators. Inparticular,

HCFR Yellow BEFORE dE=7.6 and
chromaPure Yellow Before (top below) dE=8.1

It seems to me these should agree better than this unless they are calculating the standard differently.

Today, I am attempting to create another TINT Range chart for COLOR=40 setting for comparison with the previous chart in this thread (above).

BEFORE => Baseline Color=33 Tint=0 CMS=On 10-12-2009
-------------------------------------------------------------
HCFR RED Green Blue Yellow Cyan Magenta White
x 0.665 0.266 0.145 0.431 0.199 0.362 0.319
y 0.325 0.633 0.068 0.505 0.329 0.174 0.342
Y 5.409 20.494 1.415 27.455 20.378 6.445 31.530
deltaE 5.3 18.2 34.5 7.6 14.7 35.9
deltaxy 0.025 0.047 0.009 0.011 0.025 0.046
dLuma -19.3 % -9.1 % -37.8 % -6.2 % -17.9 % -28.2 %
AFTER => Luminance Modification R and Y Color=40 Tint=0 CMS=On 10-21-2009
--------------------------------------------------------------------------
HCFR RED Green Blue Yellow Cyan Magenta White
x 0.667 0.266 0.145 0.439 0.197 0.344 0.320
y 0.324 0.661 0.067 0.518 0.326 0.164 0.341
Y 6.491 20.434 2.210 27.804 21.219 8.004 30.594
deltaE 20.1 24.3 7.0 11.1 17.9 17.9
deltaxy 0.027 0.070 0.008 0.024 0.028 0.025
dLuma -0.2 % -6.6 % = -2.0 % -11.9 % -8.1 %
chromaPure deltaEs (top=BEFORE and bottom=AFTER)


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

TomHuffman is offline  
post #29 of 62 Old 10-23-2009, 09:48 AM - Thread Starter
Member
 
Plasma54321's Avatar
 
Join Date: Sep 2008
Location: Florida
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Where might I find the updated deltaE calculator? Is the link available? Thanks.
Plasma54321 is offline  
post #30 of 62 Old 10-23-2009, 10:52 AM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,478
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 80 Post(s)
Liked: 249
Quote:
Originally Posted by Plasma54321 View Post

Where might I find the updated deltaE calculator? Is the link available? Thanks.

Same as before on our website. I just updated the file.

http://www.chromapure.com/colorscience-calculators.asp

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

TomHuffman is offline  
Reply Display Calibration

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