I'm having trouble getting my color Delta-Es lower than 2 on my Samsung LN40B750.

Since the saturation of the primaries can't be increased, what's the correct strategy to get the overall error as low as possible?

I've been told to use the Accupel calculator, but I don't know how it works along with HCFR.

Hopefully somebody who has experience using Samsung's CMS can explain what I seem to be missing?

Delta E isn't the ideal way to calibrate a CMS. While Delta E values can be a useful tool, there are many variables that can make a reliable reading very difficult. One being that there is no standard Delta E formula currently agreed upon.
The calculator will be your best bet. There is much information within this forum regarding how to use it. There is also much information about CMS calibration that Tom Huffman, among others, has shared with forum members.
I would suggest getting saturation as close as you can and then work on getting luminance levels as close as possible. Also remember, you can't increase saturation if the color points are already at the maximum level. If you use a larger gamut, you can bring those color points down to their correct coordinates.
Good luck!

The calculator will be your best bet. There is much information within this forum regarding how to use it.
Thanks Lee for your reply. I'm still a relative newbie to calibration and I apologize in advance for needing hand-holding to get through this.

I've searched the calibration forum and haven't found concise information on how to use the calculator. Is there a specific thread you had in mind?

There is also much information about CMS calibration that Tom Huffman, among others, has shared with forum members.
I've read all of Tom's CMS thread but there's little that specifically applies to Samsung's CMS.

Good luck!
Thanks..I need it!

I've read all of Tom's CMS thread but there's little that specifically applies to Samsung's CMS.

It's in there: http://www.avsforum.com/avs-vb/showpost.php?p=15296174&postcount=959

There's also an example here: http://www.avsforum.com/avs-vb/showpost.php?p=17121503&postcount=44

Tom's thread also gives information on how to use the calculator. I believe he also has the spreadsheet he uses posted in that thread.

It's in there: http://www.avsforum.com/avs-vb/showpost.php?p=15296174&postcount=959
Many thanks ToBeFrank for your help. I understand the theory of what the controls do, it's the step-by-step procedure I take to get the best out of the CMS is what I'm looking for.

Is there a way to describe the Samsung CMS calibration process the same way the greyscale calibration is laid out at the GREYSCALE & COLOUR CALIBRATION FOR DUMMIES (http://www.curtpalme.com/forum/viewtopic.php?t=10457) website?

Do you primaries as close to REc.709 as you can.

Load your pirmaries into the accupel calculator, or you can remeasure the color targets in CalMAN. CalMAN really does a good job of automating the process.

Once you recalculated your primaries, doulbe check their lumanience values, then do your secondaries.

There should be very little interaction between the individual colors or greyscale/gamma.

The threshold of human perception is either 3 or 4 depending on who is writing about it. If you don't have any color measurements above dE=4, I wouldn't worry about it. And furthermore, there are many ways to derive dE numbers and what looks low using the 1976 method/space, could be better or worse using a later or earlier calculation method/space. In otherwords, you could have all dEs below 3 using 1 method and they could all be higher than 3 using a different method or color space.

Next... your meter isn't particularly accurate measuring color... it's sort of OK for primaries, but definitely not accurate for secondary colors. You need at least an i1Pro or Chroma5 with some correction tables in order to be able to work with color with some reasonable accuracy. Just because the meter provides a reading doesn't mean the reading is accurate.

There is nothing tricky or special about using the Accupel calculator with any software or meter. You simply take a reading and manually type the numbers into the Accupel calculator. It's not pretty and it's really annoying if you have to do it many many times, but it gets the job done. There is a page of the calculator for color and another page for grayscale. You enter xyY numbers for color (or grayscale steps if you are doing grayscale) and the calculator will calculate the dEs based on the color space (for color) and gamma target (for grayscale) that you select.

Thanks sotti and Doug for your replies! You've helped quite a lot.

The threshold of human perception is either 3 or 4 depending on who is writing about it. If you don't have any color measurements above dE=4, I wouldn't worry about it. And furthermore, there are many ways to derive dE numbers and what looks low using the 1976 method/space, could be better or worse using a later or earlier calculation method/space. In otherwords, you could have all dEs below 3 using 1 method and they could all be higher than 3 using a different method or color space.

Next... your meter isn't particularly accurate measuring color... it's sort of OK for primaries, but definitely not accurate for secondary colors. You need at least an i1Pro or Chroma5 with some correction tables in order to be able to work with color with some reasonable accuracy. Just because the meter provides a reading doesn't mean the reading is accurate.

There is nothing tricky or special about using the Accupel calculator with any software or meter. You simply take a reading and manually type the numbers into the Accupel calculator. It's not pretty and it's really annoying if you have to do it many many times, but it gets the job done. There is a page of the calculator for color and another page for grayscale. You enter xyY numbers for color (or grayscale steps if you are doing grayscale) and the calculator will calculate the dEs based on the color space (for color) and gamma target (for grayscale) that you select.

You take the readings and manually type the numbers in ... THEN WHAT? I mean if your set doesn't have on-spec Rec709 primaries and you cannot adjust the primaries to Rec709 spec, what do you then do?

If you have somewhat of a CMS (like the Samsungs) that lets you adjust luminance (Y) of primaries and secondaries over a fairly wide range, but with minimal, if any, of the needed adjustments of xy coordinates of the primaries and perhaps a bit more potential adjustments of xy coordinates of the secondaries, what do you do with the data you have entered into Accupel when you don't really have a Rec709 TV?

You take the readings and manually type the numbers in ... THEN WHAT? I mean if your set doesn't have on-spec Rec709 primaries and you cannot adjust the primaries to Rec709 spec, what do you then do?

If you have somewhat of a CMS (like the Samsungs) that lets you adjust luminance (Y) of primaries and secondaries over a fairly wide range, but with minimal, if any, of the needed adjustments of xy coordinates of the primaries and perhaps a bit more potential adjustments of xy coordinates of the secondaries, what do you do with the data you have entered into Accupel when you don't really have a Rec709 TV?

Get the primaries as close to 709 as you can, then when you've reached the limits you plug those measured primaries in to calculate your target secondaries, then you calibrate your secondaries to those calculated results.

Get the primaries as close to 709 as you can, then when you've reached the limits you plug those measured primaries in to calculate your target secondaries, then you calibrate your secondaries to those calculated results.

Thanks for the reply. There seems to be two schools of thought on this subject. Or maybe I remain confused over a year later. I've been pretty happy with the calibration I stumbled upon with my Samsung 750 and have been working on sound for a year.

I recently bought the CalMAN deal with the calibrated Spyder3, and my measurements are different than they were with ColorHCFR ... or maybe a couple thousand hours of operation have changed my TV. I entered the results into the Accupel calculator and got to thinking about all of this again.

Look at this thread, and in particular look at posts of gregr (Greg Rogers, who created the Accupel calculator), the first is post #34, and perhaps the most important one is post #73 where he talks of a six-axis CMS. You'll see a lot of dumb questions from me :o

http://www.avsforum.com/avs-vb/showthread.php?t=1005676&page=3

Greg clearly recommends getting secondary xy values as close to Rec709 as possible, even if the primaries are not Rec709, and I believe he recommends minimizing dE by trial and error using measured Y values and then increasing/decreasing Y value entries in the calculator until the minimum dE is found. Then the Y values that result in min dE become Y targets. These Y values are different than the ones generated by the calculator from your real measurements.

So if I have a handle on this we are to use the calculator to find out what Y values to use with our non-standard, real life, color gamut after we have adjusted xy values as close as we can to standard (Rec709) coordinates.

According to my understand/recollection of what Greg Rogers said, that is.

Considering the OP has a B750 LCD with a full CMS, wouldn't he be able to get primaries right on top of Rec. 709 targets and thus not need to use the accupel calculator for non-standard primaries?

Glad to see this thread still getting replies and major props to Chris and kjgarrison for your posts.

I thought I had searched this forum for 'calibration calculator' but I may have used 'Accupel'. Anyway, the thread I needed was the one posted by kjg.

I do have another question that relates to grey scale. If I follow Tom Huffman's guide, he suggests getting 20% and 80% as close to the correct values as possible and hopefully everything else falls into place. If I do that, this is the result:

RGB Levels.jpg (http://farm3.static.flickr.com/2637/3938121067_a5d3b36328_b.jpg)

Since blue errors aren't as perceptually significant as red and green, is it better to leave blue alone at this point instead of trying to fix the hump from 20% to 70%? If I try and fix the hump, blue starts going below red and green at 20% and I was told that this is looks worse than the blue hump. Is that correct or should I try for overall dE's of 3 or less from 20-100 and disregard the lack of blue?

Also, since I use a HTPC as a source, would it be better to let i1Match create an ICC profile or stick with HCFR? I know it can't help with my other sources, but there's at least one B750 user who claims to get better measured results using i1Match.

Considering the OP has a B750 LCD with a full CMS, wouldn't he be able to get primaries right on top of Rec. 709 targets and thus not need to use the accupel calculator for non-standard primaries?That's the problem, I can't get primaries right on top of 709 because I can't increase saturation with Samsung's CMS.

I measured the baseline (default values) of the custom color settings and this was the dE results:

R - 4.1
G - 2.6
B - 5.5
Y - 1.9
C - 2.8
M - 5.3

Since my meter isn't all that accurate with color and all my dE's are below 6, should I just leave things as they are and not bother getting better results?

Thanks for the reply. There seems to be two schools of thought on this subject. Or maybe I remain confused over a year later. I've been pretty happy with the calibration I stumbled upon with my Samsung 750 and have been working on sound for a year.

I recently bought the CalMAN deal with the calibrated Spyder3, and my measurements are different than they were with ColorHCFR ... or maybe a couple thousand hours of operation have changed my TV. I entered the results into the Accupel calculator and got to thinking about all of this again.

Look at this thread, and in particular look at posts of gregr (Greg Rogers, who created the Accupel calculator), the first is post #34, and perhaps the most important one is post #73 where he talks of a six-axis CMS. You'll see a lot of dumb questions from me :o

http://www.avsforum.com/avs-vb/showthread.php?t=1005676&page=3

Greg clearly recommends getting secondary xy values as close to Rec709 as possible, even if the primaries are not Rec709, and I believe he recommends minimizing dE by trial and error using measured Y values and then increasing/decreasing Y value entries in the calculator until the minimum dE is found. Then the Y values that result in min dE become Y targets. These Y values are different than the ones generated by the calculator from your real measurements.

So if I have a handle on this we are to use the calculator to find out what Y values to use with our non-standard, real life, color gamut after we have adjusted xy values as close as we can to standard (Rec709) coordinates.

According to my understand/recollection of what Greg Rogers said, that is.

That works too.

By the way, what is your meter? I fail to see why you can't hit 709 with the samsung LCD...

By the way, what is your meter?If that question was directed at me, I'm using a X-Rite Eye-One Display LT.

You take the readings and manually type the numbers in ... THEN WHAT? I mean if your set doesn't have on-spec Rec709 primaries and you cannot adjust the primaries to Rec709 spec, what do you then do?

Watch the TV and forget about it... or return it and purchase something with more adjustments that work correctly (or that is factory adjusted to be reasonably accurate)... or complain to the manufacturer (though that will never fix the TV you have).


If you have somewhat of a CMS (like the Samsungs) that lets you adjust luminance (Y) of primaries and secondaries over a fairly wide range, but with minimal, if any, of the needed adjustments of xy coordinates of the primaries and perhaps a bit more potential adjustments of xy coordinates of the secondaries, what do you do with the data you have entered into Accupel when you don't really have a Rec709 TV?

I wouldn't call Samsung's CMS "somewhat of a CMS" - in fact, it is the best CMS you can buy in a TV from a "major" brand, by far. I consistently adjust Samsung (plasma) primaries and secondaries to dEs of 3 or less using their CMS. The only thing I've ever calibrated that is better are the Lumagen Radiance XD ($5000) and XE ($6000) and those are only video processors, you still need a video display. You have full control over x y and Y -- if your not seeing that, your meter may be incapable of reading the real changes that are happening when you adjust the controls.

NO CMS, not even the Lumagen Radiance, can move a measured primary color that is inside the triangle toward the outside of the triangle (assumung the controls weren't set to move the point inside the triangle in the first place. In otherwords, if any of the displays "native" primary points is/are inside the Rec 709 triangle, there's nothing you can do with any CMS to move the points outwards. You can move Red (for example) towards green or blue and you can make red brighter or darker. In the case of Samsung plasmas, their Green and Red are both a bit outside the triangle, blue is usually pretty close and the complimentary colors can be completely controlled for x y and Y also (Y is controlled by moving 2 sliders up and down at the same time by the same amount... for cyan (as an example), to make it brighter, you could add 3 to blue and 3 to green and get a brigher cyan without moving it off the ideal xy coordinates.

Considering the OP has a B750 LCD with a full CMS, wouldn't he be able to get primaries right on top of Rec. 709 targets and thus not need to use the accupel calculator for non-standard primaries?

I don't know what you mean by a FULL CMS, but I can tell you that the Samsungs' CMS can not correct for undersaturation and can't do much with xy coordinates of primaries without introducing undersaturation. So it is a limited CMS, although apparently a lot more robust than many sets have (and it is in the User Menu at that!)

EDIT: I posted this before I saw Doug Blackburn's great post above. Looks like Sammy's CMS is as "full" as they come.

Watch the TV and forget about it... or return it and purchase something with more adjustments that work correctly (or that is factory adjusted to be reasonably accurate)... or complain to the manufacturer (though that will never fix the TV you have).



I wouldn't call Samsung's CMS "somewhat of a CMS" - in fact, it is the best CMS you can buy in a TV from a "major" brand, by far. I consistently adjust Samsung (plasma) primaries and secondaries to dEs of 3 or less using their CMS. The only thing I've ever calibrated that is better are the Lumagen Radiance XD ($5000) and XE ($6000) and those are only video processors, you still need a video display. You have full control over x y and Y -- if your not seeing that, your meter may be incapable of reading the real changes that are happening when you adjust the controls.

NO CMS, not even the Lumagen Radiance, can move a measured primary color that is inside the triangle toward the outside of the triangle (assumung the controls weren't set to move the point inside the triangle in the first place. In otherwords, if any of the displays "native" primary points is/are inside the Rec 709 triangle, there's nothing you can do with any CMS to move the points outwards. You can move Red (for example) towards green or blue and you can make red brighter or darker. In the case of Samsung plasmas, their Green and Red are both a bit outside the triangle, blue is usually pretty close and the complimentary colors can be completely controlled for x y and Y also (Y is controlled by moving 2 sliders up and down at the same time by the same amount... for cyan (as an example), to make it brighter, you could add 3 to blue and 3 to green and get a brigher cyan without moving it off the ideal xy coordinates.

Thank you, Doug, for a very illustrative post. If I was complaining about Samsung's CMS, I no longer am! And to think that they put all these controls in the User Menu, too!

With some trepidation I venture to disagree with your statement that the secondaries can be completely controlled for xyY. In my set this is not true for yellow. The green primary is off to the left of the Rec709 green target (x too low, and y about right). Consequently the part of the line between red and green where yellow resides passes inside of the Rec709 triangle, and yellow is on that line. So "Samsung yellow" is undersaturated just a little, and it can not be pushed away from the white point toward Rec709 yellow.

All Y values, primary and complimentary, can be completely controlled, as you stated. The challenge is in figuring out what Y values to aim for.

By the way, what is your meter? I fail to see why you can't hit 709 with the samsung LCD...

If this was for me, it is the CalMAN calibrated Spyder 3, and as has been stated you can only move any primary towards the other two primaries or any secondary towards the 3 primaries.

Glad to see this thread still getting replies and major props to Chris and kjgarrison for your posts.

I thought I had searched this forum for 'calibration calculator' but I may have used 'Accupel'. Anyway, the thread I needed was the one posted by kjg.

I do have another question that relates to grey scale. If I follow Tom Huffman's guide, he suggests getting 20% and 80% as close to the correct values as possible and hopefully everything else falls into place. If I do that, this is the result:

RGB Levels.jpg (http://farm3.static.flickr.com/2637/3938121067_a5d3b36328_b.jpg)

Since blue errors aren't as perceptually significant as red and green, is it better to leave blue alone at this point instead of trying to fix the hump from 20% to 70%? If I try and fix the hump, blue starts going below red and green at 20% and I was told that this is looks worse than the blue hump. Is that correct or should I try for overall dE's of 3 or less from 20-100 and disregard the lack of blue?

Also, since I use a HTPC as a source, would it be better to let i1Match create an ICC profile or stick with HCFR? I know it can't help with my other sources, but there's at least one B750 user who claims to get better measured results using i1Match.

I really don't know enough to answer your question about which error is preferable, but personally I would rather have my error at IRE20 than IRE30,40, 50 and 60. Lowering your blue offset will not be an isolated change. This will require changes in red offset and blue gain at the very least. I wouldn't think fixing blue would mess up red, since you can adjust red and get those lines crawling back and forth over each other. Just leave green at default.

That's the problem, I can't get primaries right on top of 709 because I can't increase saturation with Samsung's CMS.

I measured the baseline (default values) of the custom color settings and this was the dE results:

R - 4.1
G - 2.6
B - 5.5
Y - 1.9
C - 2.8
M - 5.3

Since my meter isn't all that accurate with color and all my dE's are below 6, should I just leave things as they are and not bother getting better results?

You might be able to increase the color control first to oversaturate the primaries and then rein them in better with the CMS. I don't currently have a Samsung with a CMS, but I would imagine that would be the only way to raise saturation.

Thank you, Doug, for a very illustrative post. If I was complaining about Samsung's CMS, I no longer am! And to think that they put all these controls in the User Menu, too!

With some trepidation I venture to disagree with your statement that the secondaries can be completely controlled for xyY. In my set this is not true for yellow. The green primary is off to the left of the Rec709 green target (x too low, and y about right). Consequently the part of the line between red and green where yellow resides passes inside of the Rec709 triangle, and yellow is on that line. So "Samsung yellow" is undersaturated just a little, and it can not be pushed away from the white point toward Rec709 yellow.

All Y values, primary and complimentary, can be completely controlled, as you stated. The challenge is in figuring out what Y values to aim for.

If you have green in that position, and your Blue slider is at 0, you can't do anything about it. If your blue slider (for green) is at some value other than 0, moving the slider towards zero will move green farther from blue (up). When you get green above the correct green reference point, increasing the red slider (for green) should move the green point down towards red and put it in exactly the right spot. Of course, if your blue slider (for green) is already at 0 and your green point is in that position, you can't fix it - that's the way the TV is made and you can't do anything about it... of course there is another consideration...

Spyder 3s are NOT good for measuring color - they just AREN'T. You get readings, of course, they just don't mean much. It's like asking a 3 year old math questions like "What's 2 plus 3" and getting a random number answer from the kid - the kid might guess right once in a while given enough chances, but mostly you'll get the wrong answers. You can't get enough "tech" in a meter that inexpensive to make them worth much for measuring color. The Spyder meters are reasonable for measuring grayscale where there's a wide spectrum of red, green, and blue light reaching the meter all at the same time. But try to measure red spectrum alone or blue spectrum alone or green spectrum alone and the errors can be very significant. I've never measured a Samsung plasma or LCD with green in the position you describe (with sliders in the default factory positions) - except when I used a Spyder 2.

I don't know what you mean by a FULL CMS, but I can tell you that the Samsungs' CMS can not correct for undersaturation and can't do much with xy coordinates of primaries without introducing undersaturation. So it is a limited CMS, although apparently a lot more robust than many sets have (and it is in the User Menu at that!)

EDIT: I posted this before I saw Doug Blackburn's great post above. Looks like Sammy's CMS is as "full" as they come.

I mentioned in my post that if the primary was inside the Rec 709 triangle, that no CMS could "fix" that problem - that's the same thing you are saying in your post. If a primary is inside the Rec 709 triangle, it is undersaturated. That would typically mean the filters or phosphors used by the manufacturer caused the point to be inside the triangle and it just is not possible to move that point farther out (increase the saturation) with any CMS - not even a Lumagen Radiance.

Now... if the point is inside the triangle because the manufacturer whacked the settings in the CMS and their default positions (let's say we are talking about green again) for the red and blue sliders are what is causing green to be inside the triangle, you should be able to back off on the red and blue sliders to move the green point outward (increase saturation).

Another possibility is... the Color control on some TVs may cause the measured points for the primaries to shrink and grow (make the triangle larger or smaller) as you adjust the Color control up and down. If the color control had been set much too low, the primaries might be inside the triangle, but if you increase the color control, you MAY (big MAY) move the primaries outward towards points that are not undersaturated so you CAN use the CMS controls. But this doesn't work for all TVs and as mentioned in another post, you need a meter that is capable of measuring color with reasonable accuracy... the least expensive meter I know of that does that is the i1-pro which is around $750-$800 +/- or the C5 maybe (no experience with that one, but it looks like it may be OK, circa $500). Nobody who has a Spyder, or lower-end i1 meter should be spending a lot of time getting wrapped around the color axle with a meter that's not really up to the job.

There was a post that implied using the CMS to move the primaries to the reference xy coordinates for each primary was all there is to getting color right... I'd like to correct that notion (perhaps it was unintentional, but some reading that post may not understand).

Color is 3-dimensional. If you are talking about RGB... that's 3 dimensions. If you are talking about YCbCr... that's also 3-dimensions but they are just encoded differently. uvL is a color space alternative to RGB and uvL is also 3-dimensional (hue, saturation, Luminance (brightness)). xyY is a 3-D color space also... xy are only 2 of the dimensions. Y (luminance or brightness) is the 3rd dimension and it should be addressed in the CMS. When you see a CIE color chart in either xy or uv, you are seeing only 2 dimensions of color. The chart's 3rd dimension would exist above and below the chart you see on the page but it's hard to really see a 3-D color space on a flat printed page. The 3-D RGB color space isn't too bad to visualize because it is a cube... one corner is black, one corner is white one corner is red, one green, one blue, one cyan, one magenta, one yellow... so there are 8 points to the cube and the RGB coordinates locate any point within the cube. xyY and uvL color spaces are more irregularly shaped (when we are discussing humanly visible color).

Anyway, once you set (use green as an example) the coordinates so your measured green point matches the reference green point, you still need to get the Y or luminance coordinate correct. If your software doesn't support CMS, you need the Accupel Calibration Calculator... because your green Y reference point is going to depend on what your measured WHITE point is. So you need your grayscale setup with accuracy and the best gamma before you do CMS. Anyway, you measure your white point and the Accupel calculator will tell you what your red, green, blue, cyan, magenta, and yellow Y coordinates should be. Change your white measurement and your references for Y will change. If you measure 100% white, you have to measure 100% red, 100% green, etc. If you use 75% white as your measurement point (and you should for a plasma), you also have to measure 75% green, 75% red, etc. Because your 75% white Y becomes the reference for each of your color measurements. If you measure 100% white, then measure 75% colors, you will get royally messed up.

Bottom line - there are constant, never-changing, reference points for red, green, blue, yellow, cyan, magenta, and yellow and yes, also for white. But these are 2-dimensional reference points. Y is a variable that depends on what your white reference is. If you setup the TV for 35 fL peak white in a dark room... that 35fL is your 100% white reference level for Y calculations for your colors -- but only if your colors are also measured at 100%. 100% Green will need to be rougly 70% of white, red will need to be roughly 20% of white, and blue will need to be roughly 10% of white... as far as luminance goes. So the 3rd dimension of color is a variable. If you are doing a plasma and using 75% windows, and your 75% white window measures 20 fL, your 75% green should be around 70% of 20 fL or 14 fL (the Accupel Calibration Calculator uses the precise correct fractions for each color, my whole percentages here are just rounded-off estimates for illustration). If your 75% white level measured 40 fL, your 75% green Y should be more like 28 fL -- so you see the 3rd Dimension (Y) is variable depending on how you have the video display setup.

If you are using a lower-end meter like the Spyder or i1-Display - you can use the meter to get your 100% white or 75% white measurement and you can use the Accupel calculator to get your target Y values for primaries and complimentary colors. But that's as far as you can really go because your meter won't reliably tell you what the xy and Y readings are for the primaries. These meters are a bit better for the complimentary colors since you have 2 colors of light getting to the meter - though the low-end meters are optimized to read white light made from the spectrum of 3 colors (red green blue)). So these low end meters will be maybe 50% more accurate measureing complimentary colors than they are at measuring primary colors. But they are best when reading white and shades of gray.

Great post. Most of you pros are pretty tight lipped about all of this. Understandably so.

The challenge is in figuring out what Y values to aim for.What problem are you having in this regard? If the primaries are accurate, then just use the values that are part of the definition of your target gamut. If the primaries are not accurate, then you can use Greg's Display Calculator to figure out the desired targets. I also posted a spreadsheet that does the same thing. BTW, the differences between the standard targets and the custom targets is not huge. Start with:

R: 21%
G: 71%
B: 8%

of reference white and then adjust from there. The theoretically correct values will be very close to these.

What problem are you having in this regard? If the primaries are accurate, then just use the values that are part of the definition of your target gamut. If the primaries are not accurate, then you can use Greg's Display Calculator to figure out the desired targets. I also posted a spreadsheet that does the same thing. BTW, the differences between the standard targets and the custom targets is not huge. Start with:

R: 21%
G: 71%
B: 8%

of reference white and then adjust from there. The theoretically correct values will be very close to these.

The primaries are not accurate, particularly green.

I had just earlier today found your spreadsheet, but haven't had time to try to figure out how to use it.

The Accupel calculator is confusing to me, because there are two places to enter data and no matter where you enter stuff hitting "Calculate" always gives answers.

My current (lacking confidence) understanding is to:

Get the greyscale as good as possible
Adjust the primaries and secondaries xy coordinates as close as possible to Rec709 target spots by entering actual xyY data in the lower boxes and moving xy with the CMS for spots that are off and have an adjustment that works in the right direction.
Re-enter (lower boxes) step-by-step adjustments and Calculate until the step is discovered that reveals minimum dE based on xy.
Choose Rec709 in the Accupel
Run color measurements, enter actual white xyY in the UPPER box and hit Calculate. This gives the target Ys for all the colors which will be slightly different than Rec709 ratios.
Then enter the actual measured xyY data in the bottom boxes and hit Calculate. This gives dE for each color.
Do trial and error with each color's Y value in the bottom box and hitting Calculate until the minimum dE is found.
Then go to the CMS and adjust the Ys to these new targets.


Is this right? I am a bit fuzzy on finding minimum dE with regards to primaries and secondaries xy coordinates.

The primaries are not accurate, particularly green. . . .

Do trial and error with each color's Y value in the bottom box and hitting Calculate until the minimum dE is found.
Then go to the CMS and adjust the Ys to these new targets.


Is this right? I am a bit fuzzy on finding minimum dE with regards to primaries and secondaries xy coordinates.Unfortunately, you have entered an area where even the pros disagree.

For example, I would not recommend using trial and error with the Y value to find the lowest dE. The problem with this is that there are different dE formulas and they will give radically different recommendations in this area.

I would simply get as close as possible to the Y values and xy secondary values that are the calculated targets from your measured primaries, even if other values would yield a lower dEs.

Unfortunately, you have entered an area where even the pros disagree.

For example, I would not recommend using trial and error with the Y value to find the lowest dE. The problem with this is that there are different dE formulas and they will give radically different recommendations in this area.

I would simply get as close as possible to the Y values and xy secondary values that are the calculated targets from your measured primaries, even if other values would yield a lower dEs.

Ah! So you are saying don't try to hit the Rec 709 spots with the secondaries, but rather to to the calculated xy spots? I'm sure you know that's not what Greg Rogers said to do with his calculator.

There it is in black and white!

And although different dE formulas will give different results, will they also disagree on which xyY data results in minimum dE values?

Thank you very much, by the way. I look forward to playing with your spreadsheet.

Ah! So you are saying don't try to hit the Rec 709 spots with the secondaries, but rather to to the calculated xy spots? I'm sure you know that's not what Greg Rogers said to do with his calculator.

There it is in black and white!

And although different dE formulas will give different results, will they also disagree on which xyY data results in minimum dE values?

Thank you very much, by the way. I look forward to playing with your spreadsheet.I think that you will find that there is not a huge difference between the calculated secondary spots and the Rec. 709 targets. In any case, I am of the opinion that either strategy is an acceptable approach. Since the primaries are off in this scenario, anything you do is going to be less than ideal.

My only point is that using the Y value that provides the lowest dE places too much faith in the dE model you choose. CIE94 and the 1976 formulas treat the relationship between Y and xy differently, so unless xy is very close to the gamut's spec they can give radically different recommendations as to what value of Y is the best.

I think that you will find that there is not a huge difference between the calculated secondary spots and the Rec. 709 targets. In any case, I am of the opinion that either strategy is an acceptable approach. Since the primaries are off in this scenario, anything you do is going to be less than ideal.

My only point is that using the Y value that provides the lowest dE places too much faith in the dE model you choose. CIE94 and the 1976 formulas treat the relationship between Y and xy differently, so unless xy is very close to the gamut's spec they can give radically different recommendations as to what value of Y is the best.

Thanks. And if I may ask ...

If one is willing to do the trial and error method of dE minimization for Y, which formula would you recommend be used? Would there be different recommended formula for xy than for Y?

And does your spreadsheet offer these choices?

And are you willing to comment on the formula used in the Accupel calculator?

If one is willing to do the trial and error method of dE minimization for Y, which formula would you recommend be used? Would there be different recommended formula for xy than for Y?

And does your spreadsheet offer these choices?

And are you willing to comment on the formula used in the Accupel calculator?I use CIE94 but others prefer the 1976 formulas. The formulas are similar with respect to xy. I have several spreadsheets on my web site (http://www.chromapure.com/colorscience-calculators.asp). One of them offers a dE analysis using all 4 of the major formulas.

Greg's app uses CIELUV.

Thank you VERY much.

Thanks to everyone for your very informative posts.

After many tries, I'm able to get the primaries and secondaries right on top of 709 except for red (undersaturated). According to Tom, my meter (i1LT) is fairly accurate for red.

Should I just be happy with what I've got (it looks good to me), or should I still use the Accupel calculator? The dE measurements aren't as low as before, but x,y is as close as I can get for every color but red.

What about Y? The Y measurements I get in HCFR seem to be way off from the values I should be seeing according to Tom's chart.

What about Y? The Y measurements I get in HCFR seem to be way off from the values I should be seeing according to Tom's chart.Then adjust them. That's what the CMS in the Samsung is for.

Then adjust them. That's what the CMS in the Samsung is for.
As an example, I'm getting the following values in HCFR for red:

x 0.635
y 0.334
Y 35.127

Y is nowhere close to what it should be according to your chart (0.2126). What am I missing here?

As an example, I'm getting the following values in HCFR for red:

x 0.635
y 0.334
Y 35.127

Y is nowhere close to what it should be according to your chart (0.2126). What am I missing here?

YredTarget = 0.2126 x YwhiteMeasured

YredTarget = 0.2126 x YwhiteMeasured
Thanks, that makes sense.

....In the case of Samsung plasmas, their Green and Red are both a bit outside the triangle, blue is usually pretty close and the complimentary colors can be completely controlled for x y and Y also (Y is controlled by moving 2 sliders up and down at the same time by the same amount... for cyan (as an example), to make it brighter, you could add 3 to blue and 3 to green and get a brigher cyan without moving it off the ideal xy coordinates.

Doug has your experience been red & green outside the triangle in all Samsungs? I have a 52" A630, and my red is certainly inside the triangle. I an using am i1 Display 2 though. Truthfully, I would love to get my set professionally cal'd but right now can't afford it! My results with CalMan are pretty darn good I'd say, but I just would like to eke out the last few drops fo a good picture!

Doug has your experience been red & green outside the triangle in all Samsungs? I have a 52" A630, and my red is certainly inside the triangle. I an using am i1 Display 2 though. Truthfully, I would love to get my set professionally cal'd but right now can't afford it! My results with CalMan are pretty darn good I'd say, but I just would like to eke out the last few drops fo a good picture!

Well, it depends on the year and model (and plasma or LCD). The one I worked on most recently (plasma) had red just outside the triangle in the x direction, but it was ever so slightly "above" the lower line of the triangle (y coordinate). Blue also happened to be ever so slightly above the lower line of the triangle (Y coordinate), so that placed Magenta ever so slightly above the lower line of the triangle also (Y coordinate). But these were very small errors.

So there are cases where you can be "outside" in one direction/coordinate, but "inside" in the other direction/coordinate.

I have found that most Samsumgs , until you get into the 8 series, have an undersaturated red. As stated above there is really nothing you can do about that.

For all those with trouble getting calculations right, I recommend ponying up the cash and getting CALMAN. You have a wealth of options, and the color cal includes getting it right in 3 dimensions. You have choices of xyY uvL, and several delta E standards.

There is certainly a bit of a learning curve on setup, but overall, I think it is a better tool than HCFR which is where I started. YMMV