Official Samsung 71series calibration thread - Page 43 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
Reply
 
Thread Tools
post #1261 of 1365 Old 05-20-2008, 02:17 AM
Advanced Member
 
Raptor007's Avatar
 
Join Date: Jun 2006
Location: Seattle, WA, USA
Posts: 647
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by kjgarrison View Post

This is the 17th run and the best so far. I found that the My Color Control slider changes where secondaries are plotted (after grayscale correction, tint adjustment, etc, of course).

Note how the dotted lines shown from the primaries to the secondaries converge on white (D65).

Also the color gamut of Wide puts cyan out on the line between blue and green, almost exactly at the coordinates Accupel says it should be. NOTHING else is different between Auto and Wide ... only where cyan is plotted.

Color HCFR calculates dE based on the ideal Rec709 primaries, but we don't have an ideal Rec709. We have a Samsung variation.


Run #17
5/15 wide 16

HDMI Black Normal
Movie
cont 89
brightness 43
sharpness 50
color 46
tint 48/52

warm1

colorspace Wide

Red offset 18
Green offset 17
blue offset 15
Red Gain 16
Green gain 13
blue gain 25

My Color Control all 15 except white 16

Accupel dE values:
red 1.1
green 0
blue 2.8
yellow 3.1
cyan 3.3
magenta 1.7

NOTE: the secondary spots are not where they look like the should be on the CIE chart (especially magenta) but they are extremely close to where Accupel says they should be. Looks like Samsung engineers have it covered.

Hmmm, unfortunately I doubt this latest configuration will look quite right on my 1013.1 set. My blue saturation seems lacking in Movie mode, which is needed to select the Warm1 temperature. (On my set the exact same settings in Standard and Movie modes do NOT look identical.)

Samsung BD-UP5000 (1.3), Samsung 4671 (1013.1), Denon AVR-3808CI, HTD Level 3 5.1
Denon AVR-3300, Denon PMA-700V, Advent Loudspeakers
Raptor007 is offline  
Sponsored Links
Advertisement
 
post #1262 of 1365 Old 05-20-2008, 06:32 AM
Newbie
 
TB Cruzer's Avatar
 
Join Date: May 2008
Posts: 13
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Liked: 10
Hi, new here. I'm trying to use the above calibration. My HDMI Black is set to low and grayed out. I can't change it. Anyone know why and how to change it?
TB Cruzer is offline  
post #1263 of 1365 Old 05-20-2008, 10:13 AM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by skibum5000 View Post

but the thing is REC709 magenta has a certain xyY and that is what it is and what it looks like. xyY DOES NOT depend upon device, if something measures as say .319,.275,0.54 on three sets with completely different primaries it'll look the same on all three because it measures out the same, maybe you'll have to send it a different number on all three to get that if they are not calibrated but what it measures is what it is.

So when you measure your altered magenta and it comes out to not match 709 magenta, then that is saying it doesn't look like 709 magenta, it looks like the perfect complemntary color to samsung green, but that is not what we want or how the movies were encoded to expect. It will not look like what it should, for sure. And since you have D65 locked into REC 709 D65 spot I don't think the magenta at 709 spot would warp things in a bad way.

IMO, instead of having some greens a little bit off, now you are also pulling stuff near magenta out of whack too as well as a bit by the yellows as well.

Remember the set is not taking green and doing color subtraction and getting the complentary color out. The set doesn't work like that.

Sure 709 magenta xyY is not exactly the complementary color to samsung green primary, but so what? If you set magenta to measure as 709 then you get 709 magenta which is what we want. OTOH if you could enter a conversion matrix into your PS3 then you might want the altered magenta since then you'd have a very uniform color gamut with no warping and the matrix would apply well to it, but you don't have that option.

And looking how the color decoder appears to perform on these sets I don't think going for 709 secondaries will do anything bad and I don't think your eye will care that magenta no longer appears to be the 100% perfect complentary to samsung green, could you even remotely come close to setting magenta by trying to eyeball what you think the complementary color is? I doubt hardly anybody could.

Anyway, we'll see if anyone else chimes in and eventually I'll also run the test and see whether the decoder on this set puts more colors in the 709 gamut close to spec with 709 secondaries or accupel derived secondaries, I don't have access to my set at the moment though. Maybe I'll be wrong, we'll see.

I'll get back to you about the other thing.

Well, I'll admit that I don't know enough about all of this to really debate it. And as I read what you write, it all sounds good.

HOWEVER.

What you say appears to me to be at odds with what the Accupel calculator was made for. To me it doesn't make sense to insist on having spec secondaries when you can't have spec primaries.

I do know that having the secondaries with the placement and luminance the Accupel calc calls for is very close to what the TV does, indicating perhaps that Sammy engineers see it that way too. And I do know that the dE values away from Rec709 for the secondaries are pretty big, no matter what you do. I also know that saturation mapping looks best when the Accupel spots are used. Those things I know and have measured.

There could be an infinite number of color gamuts, with different primary locations and different primary luminances. Each gamut would be able to create white, but the point at which this white would be created would be different. It wouldn't be D65 .313, .329. But it would look the same as white created by a perfect Rec709 gamut. The same goes for any of the colors. To the eye they would look the same even though they are in different parts of the xy map.

But by convention, all gamuts use D65 as white. This means that the Samsung gamut has to have some fairly radical changes from Rec709 to luminances and secondary placements to get white at D65. The definition of a secondary is based on the location of it's complementary primary and the location of white, and the locations and luminances of the adjacent primaries. This definition depends on the gamut and for the one single Rec709 gamut all the software has been set to show us where those spots are. Too bad Samsung doesn't give us a Rec709 gamut to work with.

I don't think the coordinates are absolute. I think they are relative (to the gamut). Consequently I think that Rec709 magenta only applies to the Rec709 gamut primaries. Magenta and all other secondaries are defined and calculated from their complementary primary. With Rec709 gamut magenta is where the dot is on the Rec709 CIE. With Samsung gamut, magenta is where Accupel says it should be. Look at the Accupel calculator at Rec601 and PAL gamuts, and you will see that there are no unique coordinates or luminance values for magenta (or any other color). It's all relative to the gamut.

You said too bad we don't have a convesion matrix. Well I think we do. And it's our TVs!

Now I think that side by side sets, one Rec709 gamut and one Samsung gamut, even though the xyY values would be different, magenta would look the same to our eyes.

I wish we could get gregr, who created the Accupel calculator (and of course Accupel makes devices that use these same calculations (I presume) for calibrations), to comment on all of this. Maybe you could PM him and get him to post here.

Here is the thread where he announced the calculator.

https://www.avsforum.com/avs-vb/showthread.php?t=1002615

BTW, just out of curiosity, I "eyeballed" where the white spot would be if a line is drawn from the Samsung primaries to the Rec709 secondaries. This is where the 3 lines would intersect. I forget exactly what I used, but it was around x=.304 and y=.331 (instead of .312 and .329). I then set up a grayscale with these coordinates and measured.

I have to tweak a bit, but I must say that the secondaries are very close to Rec709, the 3 lines intersect at Samsung white, and the picture looks pretty good. More on this later.
kjgarrison is offline  
Sponsored Links
Advertisement
 
post #1264 of 1365 Old 05-20-2008, 11:30 AM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by TB Cruzer View Post

Hi, new here. I'm trying to use the above calibration. My HDMI Black is set to low and grayed out. I can't change it. Anyone know why and how to change it?

You are probably using a source where it is fixed.
Are you using a PC or something else?
Also, don't use Normal just because someone lists normal in their calibration, HDMI Black is 100% independant from a calibration. ALl that matters is that you have it set to whatever you have hooked up and are watching at the time is sending out.

If it is not a PC, it is probably locked into the ideal (barring a bug, which is
always possible).

If it is a PC is locked into Normal, this, with Vista SP1, latest nvidia drivers, PDVD8 would means that it is fine for everything but vista media center tv which will look washed out unless you change video sliders to something like 43,57 (but be sure to change them back when watching any type of DVD).
skibum5000 is offline  
post #1265 of 1365 Old 05-20-2008, 11:55 AM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
>What you say appears to me to be at odds with what the Accupel >calculator was made for. To me it doesn't make sense to insist on having >spec secondaries when you can't have spec primaries.

If you want an ideally uniform color gamut, this is true. But we are pluggin in things with no color correction options so I don't it matters to go for this but we should jsu ttry to force as much as possible to be close to 709.

>I do know that having the secondaries with the placement and luminance >the Accupel calc calls for is very close to what the TV does, indicating >perhaps that Sammy engineers see it that way too.

I think it just means they have a good color engine which always keeps Y in balance with wherever you move them.

> I also know that saturation mapping looks best when the Accupel spots are >used. Those things I know and have measured.

I haven't looked at that yet. It seems like you adjust Color to fix it though if you kept 709's.

>There could be an infinite number of color gamuts, with different primary >locations and different primary luminances. Each gamut would be able to >create white, but the point at which this white would be created would be >different. It wouldn't be D65 .313, .329.

But xyY are device independant, that's the whole point of such coordinates, if everything was relative there would be no way to match anything. It would take a different mix of R,G,B primaries on each color set though, but if you measure D65 it'll say .313,.329 on any (calibrated) set.

>To the eye they would look the same even though they are in different >parts of the xy map.

They would look different if they were at different xy spots.
They would look different if they used the same proportions of R,G,B guns in the TV to make it since that would lead to different spots in xy.

>The definition of a secondary is based on the location of it's complementary >primary and the location of white, and the locations and luminances of the >adjacent primaries.

True, 709 magenta is not 100% perfectly complentary to the samsung's green primary. But rec709 magenta is what a movie expects magenta to look like, it does not expect it to look like what the complement to samsung green does.

>I don't think the coordinates are absolute.

xyY, sRGB, REC709 coords are all absolute.

RGB are not. TV primaries vary so if you feed them the same number they will all give different xyY for a given raw percentage of R, G, and B.

>Consequently I think that Rec709 magenta only applies to the Rec709 >gamut primaries.

it will only be the true complenet in REC709 primary bsed gamut, but it will still look the same on any set that measures it with xyY of REC709.

>You said too bad we don't have a convesion matrix. Well I think we do. >And it's our TVs!

only if we make it as close to 709 as possible.

>Now I think that side by side sets, one Rec709 gamut and one >Samsung gamut, even though the xyY values would be different, magenta >would look the same to our eyes.

that would be true if you sent the same percentage of R and G and B to each set. That is all relative to what they end up showing. But if it measures xyY=something on both sets it will look the same. Think about the probe, it doesn't store or know anything about what type of phosphors or LCD filters, etc are used it just reports whatever color it sees.

>I wish we could get gregr, who created the Accupel calculator (and of >course Accupel makes devices that use these same calculations (I presume) >for calibrations), to comment on all of this. Maybe you could PM him and >get him to post here.

I can try.
skibum5000 is offline  
post #1266 of 1365 Old 05-20-2008, 01:25 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by skibum5000 View Post

>What you say appears to me to be at odds with what the Accupel >calculator was made for. To me it doesn't make sense to insist on having >spec secondaries when you can't have spec primaries.

If you want an ideally uniform color gamut, this is true. But we are pluggin in things with no color correction options so I don't it matters to go for this but we should jsu ttry to force as much as possible to be close to 709.

>I do know that having the secondaries with the placement and luminance >the Accupel calc calls for is very close to what the TV does, indicating >perhaps that Sammy engineers see it that way too.

I think it just means they have a good color engine which always keeps Y in balance with wherever you move them.

> I also know that saturation mapping looks best when the Accupel spots are >used. Those things I know and have measured.

I haven't looked at that yet. It seems like you adjust Color to fix it though if you kept 709's.

>There could be an infinite number of color gamuts, with different primary >locations and different primary luminances. Each gamut would be able to >create white, but the point at which this white would be created would be >different. It wouldn't be D65 .313, .329.

But xyY are device independant, that's the whole point of such coordinates, if everything was relative there would be no way to match anything. It would take a different mix of R,G,B primaries on each color set though, but if you measure D65 it'll say .313,.329 on any (calibrated) set.

>To the eye they would look the same even though they are in different >parts of the xy map.

They would look different if they were at different xy spots.
They would look different if they used the same proportions of R,G,B guns in the TV to make it since that would lead to different spots in xy.

>The definition of a secondary is based on the location of it's complementary >primary and the location of white, and the locations and luminances of the >adjacent primaries.

True, 709 magenta is not 100% perfectly complentary to the samsung's green primary. But rec709 magenta is what a movie expects magenta to look like, it does not expect it to look like what the complement to samsung green does.

>I don't think the coordinates are absolute.

xyY, sRGB, REC709 coords are all absolute.

RGB are not. TV primaries vary so if you feed them the same number they will all give different xyY for a given raw percentage of R, G, and B.

>Consequently I think that Rec709 magenta only applies to the Rec709 >gamut primaries.

it will only be the true complenet in REC709 primary bsed gamut, but it will still look the same on any set that measures it with xyY of REC709.

>You said too bad we don't have a convesion matrix. Well I think we do. >And it's our TVs!

only if we make it as close to 709 as possible.

>Now I think that side by side sets, one Rec709 gamut and one >Samsung gamut, even though the xyY values would be different, magenta >would look the same to our eyes.

that would be true if you sent the same percentage of R and G and B to each set. That is all relative to what they end up showing. But if it measures xyY=something on both sets it will look the same. Think about the probe, it doesn't store or know anything about what type of phosphors or LCD filters, etc are used it just reports whatever color it sees.

>I wish we could get gregr, who created the Accupel calculator (and of >course Accupel makes devices that use these same calculations (I presume) >for calibrations), to comment on all of this. Maybe you could PM him and >get him to post here.

I can try.

One last try, then I give up.

Re: what's bolded above. I don't think you get what I'm saying. You keep referring to the Rec709 xyY values as if that is what we have to work with. We don't. The primaries can't be moved to Rec709. It takes a CMS to do that, and even some of them don't really get it just right.

True Rec709 magenta at .321, .154 and Y=28.5% of white Y in Rec709 is what the, in your words, movie expects. But .321, .154 and Y=28.5% of white Y in the Samsung gamut that we are constrained by is not the same magenta. The closest way to get to a magenta that looks to our eyes like the magenta the movie wants (Rec709 magenta) is to change xyY to Accupel specs. The only Rec709 spec that is kept constant is white at D65.

Any 3 primary colors can create a gamut. If the coordinates of .321, .154 and Y=28.5% of white Y are in that gamut, it could look like green to us. It could look red. It could be the white spot. I guarantee you that there are an infinite number of primary color combinations where .321 and .154 would be white, or any other color. There are also an infinite number of primary color combinations where .321, .154 are the same magenta as Rec709 magenta.

If the color that looks like magenta to us is in the gamut, it would most definitely be someplace else, and if our TV could relocate and adjust the display of the color just right, it would to our eyes just like a Rec709 magenta.. All colors are relative to the gamut they are part of.

It is all relative. The only absolute is, by convention, where to put the white spot. Our Sammys don't use the same gamut, but they do use the same white spot.

So when you get a color gamut that's different like we have, you have to adjust luminance to get the white spot at D65. Green is skewing the gamut and the white spot to the left away from red, oops, make red brighter to pull the white spot back to D65. Maybe a little more blue too, maybe even a little less green. Unlimited combinations. What's a brighter red do to the relationship with blue re: magenta? Pulls it over away from where it would be if the gamut was Rec709.. The white spot could be in an infinite number of other places in that same gamut with different luminances of the primaries. That would be a nightmare, so we all settle on .313 and .329.

I'm not an engineer, but it appears that it is easier to compensate, using luminance adjustments, for an "off" color gamut than it is to create a Rec709 color gamut in the first place. And, just maybe, it doesn't really make that much of a difference.

Moving on.
kjgarrison is offline  
post #1267 of 1365 Old 05-20-2008, 02:30 PM
Advanced Member
 
CIR-Engineering's Avatar
 
Join Date: Aug 2005
Location: Chicago USA & Berlin Germany
Posts: 866
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 16
Quote:
Originally Posted by kjgarrison View Post

Yes, you're right. Changes in this MCC white control requires the grayscale to be tweaked. It is similar to what happens when color temperature settings are changed, like from warm1 to cool1, etc. BUT, different MCC white settings, after grayscale adjustment change where the secondaries go. This is especially true for magenta. I found that there is an MCC white setting that (after adjustment of the grayscale, color setting, and tint setting) results in the 3 lines from the primaries to the secondaries intersecting at D65. Furthermore the secondaries are where Accupel says they should be and with the luminance Accupel says they should have as a result of the primaries (especially green) not being on spec per Rec709.

Interesting, I'll check it out, but honestly my gut says that it can't quite work that way... I hope it does. Usually if you adjust a white balance somewhere to fix primaries, and then fix the white point with other controls, you just wind up driving in circles. Iteration after iteration usually gets you no closer.

I've got seven more 71's already scheduled for the next six weeks so I will see what I can find. I'll also let you guys know the results.

I have several meters, and the only one that seems to really work with these is my PR-650. I also have an Eye-One Beamer that I've compared and it's not really very good on the colors, but the greyscale is ok with the Eye-One...

craigr

Precision Video Calibration Craig Rounds
JETI Specbos 1211 Spectroradiometer
Klein K10-A Laboratory Grade Colorimeter
Murideo Fresco Six-G & Six-A HDMI 2.x Multimedia Generator / Analyzer
LightSpace XPS Pro & ChromaPure Pro Color Calibration Software
www.CIR-Engineering.com - [email protected]
CIR-Engineering is offline  
post #1268 of 1365 Old 05-20-2008, 02:39 PM
Advanced Member
 
CIR-Engineering's Avatar
 
Join Date: Aug 2005
Location: Chicago USA & Berlin Germany
Posts: 866
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 16
Quote:
Originally Posted by skibum5000 View Post

1st quesiton to CIR-Engineering about the above quote:
do you really think it makes sense to go for the altered secondaries?
Fact is, ok, now they match and 100% complimentary to the set's actual primaries, but it also means that they are now farther off from 709 spec and if a movie sends magenta or yellow or cyan it wants it to look like what 709 spec says and now they simply will not. Doesn't it make sense to try to keep them, and as much as you can, on 709 spots? Only green was that far off to begin with so why warp out yellow and magenta too now?

I honestly don't understand what you are saying and it seems a little contradictory? Could you try and ask in a different way, because I would be happy to try and answer whatever you are asking.

I mean, rec and SMTPE have different specs and you should try and match the spec you want to use. If you are using your set for HD sources then 709 would be the best spec to match. If you are using the set for SD and DVD then rec 601 would be best.

But the different specs have specific coordinates in 3D space (xyY) that should all be matched whenever possible. So if you use 709 then your should try and get all six colors as close to all three of their coordinetas as the set will allow.

Does that come close to answering the question?

craigr

Precision Video Calibration Craig Rounds
JETI Specbos 1211 Spectroradiometer
Klein K10-A Laboratory Grade Colorimeter
Murideo Fresco Six-G & Six-A HDMI 2.x Multimedia Generator / Analyzer
LightSpace XPS Pro & ChromaPure Pro Color Calibration Software
www.CIR-Engineering.com - [email protected]
CIR-Engineering is offline  
post #1269 of 1365 Old 05-20-2008, 02:47 PM
Advanced Member
 
CIR-Engineering's Avatar
 
Join Date: Aug 2005
Location: Chicago USA & Berlin Germany
Posts: 866
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 16
Quote:
Originally Posted by skibum5000 View Post

One more quesiton to CIR, have you done any A650/750 yet or heard anything about them? I see one guy claims a 3600:1 CR, which sounds like a big step from the 71 series. But perhaps his probe would also have given higher numbers for 71 series too.


But back to the new settings, you have a very nice gamma ramp there! And considering all you had was the set controls and couldn't use an ICC profile the gray scale is very good too! (although I suspect it would not match nearly so well to my particular set, unless it was jsut my spyder2 being that far off, which it might be, but anyway very uniform on your set!).

I have not done a single A650 or 750 yet. usually there is a delay from when they go on sale to when people start asking for calibrations. Used to be nearly a year, now it's down to just a few months before I start getting requests. However, I don't think you can get an honest CR like that off an LCD panel yet. I bet there were dynamic functions that weren't disabled for the test.

I work in the service menu on these sets and I have gotten pretty much identical final results on all the 71's I've worked on to date. I find these units are super sweet with gamma and greyscale. I mean, the gamma is so adjustable on these, I usually prompt my customer for their particular viewing requirements and set gamma where I think it will be best for that particular situation. You know the usual, 2.2 gamma for a brightly lit room all the way up to 2.5 for a dark environment.

Greyscale on these is just great. I think all the 71's are capable of landing in the same general area. If you aren't getting results like I have then it's more likely the meter being used or the skill of the operator (not a slight, but I've been doing this for a while). It's not like greyscale is very difficult though so I'd suspect the meter on that one especially since the Spider meters are pretty far down on the totem pole.

craigr

Precision Video Calibration Craig Rounds
JETI Specbos 1211 Spectroradiometer
Klein K10-A Laboratory Grade Colorimeter
Murideo Fresco Six-G & Six-A HDMI 2.x Multimedia Generator / Analyzer
LightSpace XPS Pro & ChromaPure Pro Color Calibration Software
www.CIR-Engineering.com - [email protected]
CIR-Engineering is offline  
post #1270 of 1365 Old 05-20-2008, 03:12 PM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by CIR-Engineering View Post

I honestly don't understand what you are saying and it seems a little contradictory? Could you try and ask in a different way, because I would be happy to try and answer whatever you are asking.

I mean, rec and SMTPE have different specs and you should try and match the spec you want to use. If you are using your set for HD sources then 709 would be the best spec to match. If you are using the set for SD and DVD then rec 601 would be best.

But the different specs have specific coordinates in 3D space (xyY) that should all be matched whenever possible. So if you use 709 then your should try and get all six colors as close to all three of their coordinetas as the set will allow.

Does that come close to answering the question?

craigr

I think so. It appears to agree with what I am saying.


But just to make sure:

Basically I claim that xyY are absolute coordinates and that we want every last color, shade, be it primary, secondary or anything inbetween to get measured by a probe to xyY as close to 709 in all cases.

Some others are claiming that they are either patially or completely relative and that magenta as a movie wants it to look will have a different xyY value from set to set. I claim that is not true, it has to measure to the same xyY (even if it might take different proportions of firing the R,G,B channels on a TV to get that same xyY and might no longer be the exact color complement to the actual primary on the set).

I claim that if a patch on the screen is measured by a probe to be 0.345,0.298,0.76 on four-hundred different sets then it will look the same to the eye on all those sets. I'm not sure if anyone is disagreeing with that directly. But if this is true then how can the xyY for 709 magenta need to be made to vary from set to set to look as expected as some are claiming?

I also say if xyY are relative in that fashion then how could we even measure that our primaries don't match to begin with?

Now, it is true that the exact, color complement secondaries to a given set of primaries do vary, so while the secondaries on a set are relative to the primaries of a particular gamut the desired colors are not. The way that 709 says magenta should look has the same xyY on ANY set. It's just that yeah it might not be the perfect complement to its primary, but so what? Who would ever notice that?

I say that since the movies (at least on HD DVD and bblu-ray) and ATSC HDTV are encoded expecting REC709 that we should try to place all primaries and secondaries, everything, as close to the xyY in 709 spec. We can't get green primary perfect, so we are stuck with that. But why alter the secondary placement? Some say we need to use accupel to calculate where the exact complement to each primary would be and move the secondaries to those spots. I say that while sure it would bring them into perfect color theory complement of their respective primaries as the set actually delivers the primaries it would also brings them way out of looking like what 709 says and that the latter thing is what matters.

I say that shifting magenta to accupel calculated spot makes it look NOT like what the director intends for magenta since xyY is 100% absolute. If you keep it at the xyY of 709 then it looks like what the director intended.

We simply want magenta and as many shades and colors to measure as close to 709 spec as possible, who cares if that means that cyan,yellow , magenta are not the exact complement to our actual primaries so long as they look like they are intended to look by 709 spec? All we have is some shades near green and cyan off but at least everything else is kept closer to expected.

I'm claiming that the only case where it would be better to move the secondaries to the adjusted positions would be if we were using that was going to use a simple matrix color conversion method. That would work better with a standard, uniform gamut with secondaries in derived locations, but since only PCs in certain limited cases can use this transforms, this is not the way to set the set.

anyway, hopefully that is clear.

thanks.
skibum5000 is offline  
post #1271 of 1365 Old 05-20-2008, 03:26 PM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by CIR-Engineering View Post

I have not done a single A650 or 750 yet. usually there is a delay from when they go on sale to when people start asking for calibrations. Used to be nearly a year, now it's down to just a few months before I start getting requests. However, I don't think you can get an honest CR like that off an LCD panel yet. I bet there were dynamic functions that weren't disabled for the test.

I work in the service menu on these sets and I have gotten pretty much identical final results on all the 71's I've worked on to date. I find these units are super sweet with gamma and greyscale. I mean, the gamma is so adjustable on these, I usually prompt my customer for their particular viewing requirements and set gamma where I think it will be best for that particular situation. You know the usual, 2.2 gamma for a brightly lit room all the way up to 2.5 for a dark environment.

craigr

interesting to note for dark environments going a little high on gamma might not be bad. I noticed the other guy has the gamma a little low on top end and a little high on bottom end for a dark room calibration. I guess that adds a little more pop, but is still a little bit of a judgement call as to whether you want to maximize it that way or not.

Color Eye's L* gamma I think goes the opposite way, for easier shadow and highlgith detail differntiation, not sure I like it though, I think I prefer an even 2.2 to heading a bit towards your adjusted way.

Too bad you haven't done a 650/750 yet. With some issues with my set I might have option of either board swap or swtich to 650/750. If the contrast ratio is really much better it might make sense to switch, if not, then maybe why risk going through a series of sets with bad clouds or what not. I wish I knew if he had Dyanamic Contrast or power saving or something on when he measured it, maybe he had DC on Low....

Anyway, I hear rumours the CR might be noticeably better but then others are doubtful....

I have an X-94 probe now too. The single largest two differences are how it measures greens and CR/gamma vs the spyder. With x-94 I get similar CR for my Samsung 244T pc monitor to most reviews (about 1000:1) while the spyder kept saying lower (more like 600:1) giving different Y values at extreme scale ends especially.

Of course even an X-94 is probably a good dE2 off from your more expensive probe, on avg.
skibum5000 is offline  
post #1272 of 1365 Old 05-20-2008, 03:40 PM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by kjgarrison View Post

One last try, then I give up.

Re: what's bolded above. I don't think you get what I'm saying. You keep referring to the Rec709 xyY values as if that is what we have to work with. We don't. The primaries can't be moved to Rec709. It takes a CMS to do that, and even some of them don't really get it just right.

True Rec709 magenta at .321, .154 and Y=28.5% of white Y in Rec709 is what the, in your words, movie expects. But .321, .154 and Y=28.5% of white Y in the Samsung gamut that we are constrained by is not the same magenta. The closest way to get to a magenta that looks to our eyes like the magenta the movie wants (Rec709 magenta) is to change xyY to Accupel specs. The only Rec709 spec that is kept constant is white at D65.

Any 3 primary colors can create a gamut. If the coordinates of .321, .154 and Y=28.5% of white Y are in that gamut, it could look like green to us. It could look red. It could be the white spot. I guarantee you that there are an infinite number of primary color combinations where .321 and .154 would be white, or any other color. There are also an infinite number of primary color combinations where .321, .154 are the same magenta as Rec709 magenta.

If the color that looks like magenta to us is in the gamut, it would most definitely be someplace else, and if our TV could relocate and adjust the display of the color just right, it would to our eyes just like a Rec709 magenta.. All colors are relative to the gamut they are part of.

It is all relative. The only absolute is, by convention, where to put the white spot. Our Sammys don't use the same gamut, but they do use the same white spot.

So when you get a color gamut that's different like we have, you have to adjust luminance to get the white spot at D65. Green is skewing the gamut and the white spot to the left away from red, oops, make red brighter to pull the white spot back to D65. Maybe a little more blue too, maybe even a little less green. Unlimited combinations. What's a brighter red do to the relationship with blue re: magenta? Pulls it over away from where it would be if the gamut was Rec709.. The white spot could be in an infinite number of other places in that same gamut with different luminances of the primaries. That would be a nightmare, so we all settle on .313 and .329.

I'm not an engineer, but it appears that it is easier to compensate, using luminance adjustments, for an "off" color gamut than it is to create a Rec709 color gamut in the first place. And, just maybe, it doesn't really make that much of a difference.

Moving on.


A given xyY is how a certain color in the color he CIE Gamut looks. If we measure xyY for a patch of color on 500 different sets they will look the same.

A magenta with a certain xyY measured by a probe on a set would look to us like same color on any other set measuring with the same probe the same xyY.

OTOH, setting the R,G,B functions to the same values on different sets or different gamuts would look different to us. Sending on set or one gamut 40% to R channel and 50% to green and 20% to blue will surely look different if the gamut is different.

Just as the exact color complement of say the primary green on one set would be different on another set, that varies gamut to gamut. A magenta with a certain xyY might be the exact color complement of green in one gamut but not in another. 709 xyY magenta will not be the complement of the green on our samsung TV, that would have a different xyY, but the 709xyY magenta still looks the same to us on any set, even if it no longer be the actual secondary (complement to green on that set).

The actual complement to the green on our samsungs is a magenta that is different from 709 magenta and has different xyY value BUT it also looks different to us than what a director wants magenta to look like. So my point was, why also throw off magenta, we already are stuck with green being off. Why match the offness of magenta to the offness of green unless we are prepping the set for a matrix based color gamut converting source (which are very rare).

or so i all hope
skibum5000 is offline  
post #1273 of 1365 Old 05-20-2008, 05:42 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by skibum5000 View Post

A given xyY is how a certain color in the color he CIE Gamut looks. If we measure xyY for a patch of color on 500 different sets they will look the same.

A magenta with a certain xyY measured by a probe on a set would look to us like same color on any other set measuring with the same probe the same xyY.

If you are saying that if all 500 sets have magenta at the same xyY, then they will all look the same regardless whether they have the same gamut, then no. If all 500 sets have the same color gamut and the same luminance values for all the colors, then yes. The gamut is not just the CIE coordinates; it is luminance also.

Depending on the gamut, that exact same xyY might be a color totally different. It is only the color it is on the Rec709 CIE chart because that is a Rec709 CIE chart. It's not like the CIE chart represents something absolute like wavelengths of light. They are just relative positions based on conventions.

I think maybe you are thinking two dimensionally. Maybe this is causing you to think that each coordinate in xy terms is only one single color. Don't forget that Y doesn't show up on the CIE chart. Different luminance values for any and every color change the appearance of all the other colors, including the white point. By convention, the white point has to be put back at D65, and if you can't move the primaries (and we can't) various Y adjustments are made. You see Accupel's magenta at a different xy and think it is the wrong color, but given the proper mix of (non-Rec709) Y values for the other colors, the (non Rec709) locations of the other colors, and the D65 location of white, that looks like the same magenta you would see on a perfect Rec709 set.


Quote:
Originally Posted by skibum5000 View Post


OTOH, setting the R,G,B functions to the same values on different sets or different gamuts would look different to us. Sending on set or one gamut 40% to R channel and 50% to green and 20% to blue will surely look different if the gamut is different.

Just as the exact color complement of say the primary green on one set would be different on another set, that varies gamut to gamut. A magenta with a certain xyY might be the exact color complement of green in one gamut but not in another. 709 xyY magenta will not be the complement of the green on our samsung TV, that would have a different xyY, but the 709xyY magenta still looks the same to us on any set, even if it no longer be the actual secondary (complement to green on that set).

I agree completely. Now if I can only get you to understand that the different gamuts you seem to understand in the above is no different than the different gamut Samsung delivers. So how do we get our different gamut to display magenta properly? Adjust for the gamut.

Quote:
Originally Posted by skibum5000 View Post


The actual complement to the green on our samsungs is a magenta that is different from 709 magenta and has different xyY value BUT it also looks different to us than what a director wants magenta to look like. So my point was, why also throw off magenta, we already are stuck with green being off. Why match the offness of magenta to the offness of green unless we are prepping the set for a matrix based color gamut converting source (which are very rare).

or so i all hope

Define "different". If you mean different xyY. You bet. If you mean it would look different side by side with two otherwise identical, properly calibrated TVs, one a Rec709 and the other a Samsung709ish gamut with Accupel adjustments. Nope.
kjgarrison is offline  
post #1274 of 1365 Old 05-21-2008, 11:14 AM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by kjgarrison View Post

If you are saying that if all 500 sets have magenta at the same xyY, then they will all look the same regardless whether they have the same gamut, then no. If all 500 sets have the same color gamut and the same luminance values for all the colors, then yes. The gamut is not just the CIE coordinates; it is luminance also.

Well, if xyY are absolute (and by this I mean this that a certain xyY triad IS how a certain color looks to us and that is that), and I believe they are, then if you measure some color patch to have the three coordinates xyY on 50 sets then that color patch looks the same to us on all 50 sets.

As far as I understand there is the complete visible light CIE gamut that covers the entire visible light range and each and every color on it has an xyY value (mostly the chart gets drawn as 2D with just the xy part, but in reality it does have a third dimension rising above for the Y axis). Gamuts like the 709 or 601 or AdobeRGB or ProPhotoRGB, etc. all fit within the CIE full visible spectrum Gamut volume and they all use the same xyY values as the base complete gamut. Burnt Olive-Green is a particular identical xyY or sRGB triad in all. But it is also a VARYING RGB triad in all since you need a differnt mix of primaries in each gamut to make it.

Each sub-gamut (sRGB/Rec709, AdobeRGB, Sony XBR4 primaries etc) covers a different volume of the complete CIE Visible Gamut. And each of those has a different set of primaries and secondaries (each a different xyY, for at least one coordinate of one primary otherwise they are the same gamut). To get any given xyY value you need to mix a different proportion of the primaries in any given color gamut. You might need (totally made up #'s here) to fire the channel functions at say 40,80,60 with AdobeRGB to make xyY=0.3,0.2,0.5 but you might need to fire the channel functions of 709REC at 45,83,64 to make that same xyY=0.3,0.2,0.5.

RGB coordinates are relative, you need to send say RGB digital signal of 0,34,78 one set and 3,33,95 on another to produce the same color, but that same color has the same three xyY values in these absolute gamuts such as sRGB, in sRGB the same three coordinates always look the same on any device but they will need different RGB coordinates on each devices to produce that same color.


Quote:
Originally Posted by kjgarrison View Post

Depending on the gamut, that exact same xyY might be a color totally different. It is only the color it is on the Rec709 CIE chart because that is a Rec709 CIE chart. It's not like the CIE chart represents something absolute like wavelengths of light. They are just relative positions based on conventions.

But I believe that the Rec709 CIE is overlayed onto the CIE complete visible spectrum gamut (notice how it is drawn inside it and all these sub gamuts all share the same xyY coordinate system with the same xyY values giving the same color). What differs is that gamut to gamut you need to feed the three color functions different numbers to produce a given xyY value.

Quote:
Originally Posted by kjgarrison View Post

I think maybe you are thinking two dimensionally. Maybe this is causing you to think that each coordinate in xy terms is only one single color. Don't forget that Y doesn't show up on the CIE chart.

no, I'm not thinking in just 2D. And a fully drawn CIE chart actually is drawn with the third dimension, that mostly obscures things so it is not all that often drawn that way, but I have seen pictures of the full 3D chart where the Y does show up.

Quote:
Originally Posted by kjgarrison View Post

Different luminance values for any and every color change the appearance of all the other colors, including the white point. By convention, the white point has to be put back at D65, and if you can't move the primaries (and we can't) various Y adjustments are made. You see Accupel's magenta at a different xy and think it is the wrong color, but given the proper mix of (non-Rec709) Y values for the other colors, the (non Rec709) locations of the other colors, and the D65 location of white, that looks like the same magenta you would see on a perfect Rec709 set.

Yeah you do need the proper mix of the primary channel functions and the proper mix DOES VARY from gamut to gamut, but you are confusing this and then imagining that xyY coordinates are all relative, at least aside from the primaries (but you can't have them being absolute for some colors and not for others). You need a DIFFERENT mix on each gamut (DIFFERENT color channel function input values) to produce the SAME color which will have the SAME exact xyY values on different sets.

We both agree that we need to use different mixtures of the primaries in different gamuts to produce a given color. Where we differ is the coordinates of that color. I say they are identical xyY/sRGB/etc. but different RGB and you say they are different xyY too.

Quote:
Originally Posted by kjgarrison View Post

I agree completely. Now if I can only get you to understand that the different gamuts you seem to understand in the above is no different than the different gamut Samsung delivers. So how do we get our different gamut to display magenta properly? Adjust for the gamut.

I think you are mixing up the relative color channel function inputs and absolute gamut coordinates. The same three numbers for xyY in Rec709, CIE full spectrum, AdobeRGB, ProPhotoRGB, Samsung 71 series all look the same. But the sending a picture with the screen all RGB=34,68,127 to each of those WOULD look different.

I also get that the true secondaries (meaning exact color complement) to a primary does vary in xyY coordinate from gamut to gamut. Rec709 magenta is NOT the true secondary to the primary green of a samsung 71 series. But where we disagree is that you think that the true secondary magenta to samsung green will look exactly like magenta would on a set that had the exact Rec709 primaries and I say that since true secondary magenta on samsung has a different xyY than Rec709 magenta, it will look different.

I think that the eye can't easily tell if some color is the exact complement to another one and doesn't really care and that balancing that doesn't matter too much but that it can tell Rec709 Secondary Magenta from Samsung 5271 Secondary Magenta so we should shift the 5271 one back to the Rec709 spot so it will look same to the eye (even if that does make the magenta colors no longer fall directly across from green if we plotted the chart, which I don't think the eye cares about much, but maybe I am wrong and the eye likes a picture that happens to have all sorts of colors in it to have the magentas looking complentary to greens more than having them looking exactly like the intended magenta, i tend to doubt it though.)

If xyY are completely relative then what would the meaning of saying our samsung 5271 primaries are a certain xyY even have then? How could we even know how to calculate the secondaries from them if the primaries were totally vague?

It's a different concept to say that only primaries are absolute and that secondaries are derived and relative, which is true from saying that xyY are not absolute colors.
skibum5000 is offline  
post #1275 of 1365 Old 05-22-2008, 12:27 AM
AVS Forum Special Member
 
dEEahn20's Avatar
 
Join Date: Sep 2007
Location: Orange County, CA
Posts: 1,059
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Haven't been here in a while. Wow, this thread is really over my head now. Awesome, kjgarrison has new settings. Can't wait to try them out.

LN-T4671F - FW: 2003.2
DV-980H - FW: 0B-0903
6620NP - FW: 2.10
N081URAY-Y37 - FW: 0.0
dEEahn20 is offline  
post #1276 of 1365 Old 05-22-2008, 12:50 PM
AVS Forum Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,184
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 25 Post(s)
Liked: 23
Quote:
Originally Posted by skibum5000 View Post

>I wish we could get gregr, who created the Accupel calculator (and of >course Accupel makes devices that use these same calculations (I presume) >for calibrations), to comment on all of this. Maybe you could PM him and >get him to post here.

I can try.

I'd be happy to answer a question or explain something. But I can't read this entire thread so please just state a question.

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #1277 of 1365 Old 05-22-2008, 02:37 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by gregr View Post

I'd be happy to answer a question or explain something. But I can't read this entire thread so please just state a question.

Great! Thanks. That quote was something I actually wrote.

Our question/confusion is all about how your Accupel calculator works, and whether a given secondary color's new location would "look" to our eyes the same as it would look if the gamut was Rec709 and the color was at Rec709 spec xyY. So it is really all about non-Rec709 gamuts and what adjustments to xyY coordinates are required to compensate.

In our case, the Samsung 71's have green primary off to the left, and consequently Accupel calculator shows the secondary of magenta moved quite a bit over towards red. Some of us think we should try to get magenta closer to Rec709 location xy, and some of us think we should try to get magenta where Accupel says it should be.

We wonder if the appearance of a given xyY set of coordinates will look the same for multiple different gamuts, meaning that xyY coordinates are absolute. Or are the same given xyY coordinates relative to the gamut they are part of and capable of appearing to the eye as an unlimited number of colors, including white?
kjgarrison is offline  
post #1278 of 1365 Old 05-22-2008, 04:31 PM
AVS Forum Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,184
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 25 Post(s)
Liked: 23
Quote:
Originally Posted by kjgarrison View Post

Great! Thanks. That quote was something I actually wrote.

Our question/confusion is all about how your Accupel calculator works, and whether a given secondary color's new location would "look" to our eyes the same as it would look if the gamut was Rec709 and the color was at Rec709 spec xyY. So it is really all about non-Rec709 gamuts and what adjustments to xyY coordinates are required to compensate.

In our case, the Samsung 71's have green primary off to the left, and consequently Accupel calculator shows the secondary of magenta moved quite a bit over towards red. Some of us think we should try to get magenta closer to Rec709 location xy, and some of us think we should try to get magenta where Accupel says it should be.

We wonder if the appearance of a given xyY set of coordinates will look the same for multiple different gamuts, meaning that xyY coordinates are absolute. Or are the same given xyY coordinates relative to the gamut they are part of and capable of appearing to the eye as an unlimited number of colors, including white?

I'm not sure if I totally understand your questions but here are some answers and then perhaps we can iterate to the right questions/answers.

I made the AccuPel calculator to solve several issues that had been raised on the forum. In fact it evolved with new features as new issues were raised.

1. The first question asked was how to determine the Y (luminance) values for the Rec. 601/Rec. 709 standards if the YCbCr/YPbPr to RGB color decoding worked according to the standards (or if RGB signals were used and there was no color decoding), plus some people didn't know the x,y values of the complementary colors. So you can click on the Rec 709, Rec 601, or PAL buttons and you will get the appropriate x,y,Y values for the primary and complementary colors.

2. Another question was how would the primary Y and complementary x,y,Y values shift if the Gray color wasn't at D65 (which is the default value for Rec 601/709). So after clicking the Rec 709 or Rec 601 buttons, you can then enter a different set of x,y,Y values for Gray, and then when you click the Compute button you will get the x,y,Y values for all primaries and complementary colors, again assuming the YCbCr/YPbPr color decoding is according to the standard signal type being used (Rec 601 or Rec 709) or if RGB signals are used.

3. Another more general question was where would the primary Y values and the complementary x,y,Y values be for any set of primaries and any Gray, again if the color decoding was according to standard or if RGB signals were used. You can simply enter the x,y values for the primaries and gray and then click Compute to get all of those values.

4. Then the question was asked, what is the dE deviation of a set of measured values from a set of target values. If you enter your measured values in the bottom half of the Color Gamut page and click compute, you will get the dE values (CIELUV 1976) of the measured primary and secondary colors and gray relative to the target values selected/entered in the top half the page. The dE value of gray is not affected by the measured Y value of gray, since it is used as the reference Y value for the measurement of the other Y values.

In answer to your last question, a set of x,y,Y tristimulus values defines the appearance of a color. The relative luminance mix of R,G,B primary colors will have to change to produce a particular x,y,Y color if the x,y values of the primaries change.

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #1279 of 1365 Old 05-22-2008, 08:39 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by gregr View Post

I'm not sure if I totally understand your questions but here are some answers and then perhaps we can iterate to the right questions/answers.

I made the AccuPel calculator to solve several issues that had been raised on the forum. In fact it evolved with new features as new issues were raised.

1. The first question asked was how to determine the Y (luminance) values for the Rec. 601/Rec. 709 standards if the YCbCr/YPbPr to RGB color decoding worked according to the standards (or if RGB signals were used and there was no color decoding), plus some people didn't know the x,y values of the complementary colors. So you can click on the Rec 709, Rec 601, or PAL buttons and you will get the appropriate x,y,Y values for the primary and complementary colors.

2. Another question was how would the primary Y and complementary x,y,Y values shift if the Gray color wasn't at D65 (which is the default value for Rec 601/709). So after clicking the Rec 709 or Rec 601 buttons, you can then enter a different set of x,y,Y values for Gray, and then when you click the Compute button you will get the x,y,Y values for all primaries and complementary colors, again assuming the YCbCr/YPbPr color decoding is according to the standard signal type being used (Rec 601 or Rec 709) or if RGB signals are used.

3. Another more general question was where would the primary Y values and the complementary x,y,Y values be for any set of primaries and any Gray, again if the color decoding was according to standard or if RGB signals were used. You can simply enter the x,y values for the primaries and gray and then click Compute to get all of those values.

4. Then the question was asked, what is the dE deviation of a set of measured values from a set of target values. If you enter your measured values in the bottom half of the Color Gamut page and click compute, you will get the dE values (CIELUV 1976) of the measured primary and secondary colors and gray relative to the target values selected/entered in the top half the page. The dE value of gray is not affected by the measured Y value of gray, since it is used as the reference Y value for the measurement of the other Y values.

In answer to your last question, a set of x,y,Y tristimulus values defines the appearance of a color. The relative luminance mix of R,G,B primary colors will have to change to produce a particular x,y,Y color if the x,y values of the primaries change.

Thank you! Wow!

I'm beginning to worry that I don't know enough to really ask the questions in a correct manner, but the part of your reply that I bolded above seems closest to the issue we have been debating. We accept the convention that the white point is D65, and not just "any gray".

I'm sure skibum5000 must be travelling, or he would be here. He and I have been debating whether to use the Accupel calculator as you intended it to be used. I say the xy definition of magenta is the point where a straight line from green through the white point (in this case D65) intersects a line between blue and red. That's where Accupel calc "wants" it and gives the lowest dE. He says, no, the definition of magenta is the spot shown on a proper Rec709 gamut.

I say that the new location of magenta called for by your calculator will look to our eyes like the magenta (or the closest thing to it this particular gamut can produce) that would be shown if the display's gamut is Rec709.

With an ideal Rec709 gamut, magenta xyY is .321, .154 and Y=28.5% of white Y. But with the "Samsung gamut" we have to live with primaries being off, particularly green. The Accupel calculator places magenta a lot closer to red, changes xy for the other secondaries a little, and changes Y values for everything, especially red (increase of about 25%).

Skibum5000 says that .321, .154, Y=28.5% white Y is the one and only unique definition of magenta, and that we should not use the Accupel xy location (Y is moot, we can't change relative luminance of individual colors). He says we should try to get magenta as close as possible to the spot it occupies on an accurate Rec709 gamut. He says that no matter where the primaries are located (xy) and no matter what their Y values are, if .321, .154, Y=28.5% white Y is in the gamut, it will always look the same.

I say that the locations and luminances of the primaries can make whatever is displayed at .321, .154 look to our eyes like an infinite number of colors. I say that 3 primaries could be picked with .321, .154 centered and that with the right luminances, .321, .154 could be the white point for that particular gamut. RGB are not the only colors that could be used. In fact the RGB used by our Samsungs is not the same RGB that Rec709 calls for. I say that's the principle of how the Accupel calculator works.

I note that with the Accupel adjustments the lines drawn from the primaries to the secondaries intersect nicely at the white point D65.

So at the root of all this is a disagreement as to whether the Accupel calculator is correct in relocating the secondaries the way it does (mostly magenta, since it is green that is farthest from it's Rec709 spot.) I say it is straight across from green through D65, and skibum5000 says no, it's .321, .154.
kjgarrison is offline  
post #1280 of 1365 Old 05-22-2008, 09:52 PM
AVS Forum Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,184
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 25 Post(s)
Liked: 23
Quote:
Originally Posted by kjgarrison View Post

I'm sure skibum5000 must be travelling, or he would be here. He and I have been debating whether to use the Accupel calculator as you intended it to be used. I say the xy definition of magenta is the point where a straight line from green through the white point (in this case D65) intersects a line between blue and red. That's where Accupel calc "wants" it and gives the lowest dE. He says, no, the definition of magenta is the spot shown on a proper Rec709 gamut.

The AccuPel calculator doesn't "want" magenta to be anywhere! The calculator watches no movies. The calculator tells you what color (x,y,Y) a magenta signal will produce given a specific set of primaries and a specific gray point (with standard color decoding or RGB signals). You must decide what color you want a magenta signal to produce. If it were me, I would want the magenta signal to produce either the SMPTE-C (called Rec. 601 to make it more familiar to most users) color gamut's magenta color, or the Rec. 709 color gamut's magenta color. Most of the time, I would prefer the SMPTE-C magenta because that is the color gamut used by most professional monitors when converting film to video. So I would select the SMPTE-C color gamut (click Rec. 601 on the calculator) and compute the dE values of the measured values relative to those values.

Quote:
Originally Posted by kjgarrison View Post

I say that the new location of magenta called for by your calculator will look to our eyes like the magenta (or the closest thing to it this particular gamut can produce) that would be shown if the display's gamut is Rec709.

No, what I said is that each x,y,Y value produces a specific color. If you want the magenta of Rec. 709 then you need to produce the x,y,Y value of the Rec. 709 magenta color regardless of what the primary colors are.

Quote:
Originally Posted by kjgarrison View Post

With an ideal Rec709 gamut, magenta xyY is .321, .154 and Y=28.5% of white Y. But with the "Samsung gamut" we have to live with primaries being off, particularly green. The Accupel calculator places magenta a lot closer to red, changes xy for the other secondaries a little, and changes Y values for everything, especially red (increase of about 25%).

No, you have things backwards. The calculator tells you what color you will get with a magenta signal. If you want a Rec 709 magenta from a magenta signal then you need to produce the Rec 709 x,y,Y values.

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #1281 of 1365 Old 05-22-2008, 11:29 PM
Member
 
mike08's Avatar
 
Join Date: Apr 2008
Location: saskatchewan
Posts: 61
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I recently bought an audio rack and my samsung lnt 5271 lcd cord wont reach my power conditioner(ht7300pc tripplite) so i am using a heavy extention cord to make it reach.Will this effect my lcd? any comments would be greatly appreciated.
mike08 is offline  
post #1282 of 1365 Old 05-23-2008, 06:43 AM
Member
 
saturk's Avatar
 
Join Date: Jan 2008
Posts: 62
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 11 Post(s)
Liked: 10
Personally, I would not use an extension cord. Try relocating.
saturk is offline  
post #1283 of 1365 Old 05-24-2008, 11:42 AM
Member
 
elf6c's Avatar
 
Join Date: Nov 2006
Posts: 20
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
kjgarrison I have had good luck with your last set of calibration numbers- thanks!

elf6c is offline  
post #1284 of 1365 Old 05-24-2008, 03:33 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by gregr View Post

The AccuPel calculator doesn't "want" magenta to be anywhere! The calculator watches no movies. The calculator tells you what color (x,y,Y) a magenta signal will produce given a specific set of primaries and a specific gray point (with standard color decoding or RGB signals). You must decide what color you want a magenta signal to produce. If it were me, I would want the magenta signal to produce either the SMPTE-C (called Rec. 601 to make it more familiar to most users) color gamut's magenta color, or the Rec. 709 color gamut's magenta color. Most of the time, I would prefer the SMPTE-C magenta because that is the color gamut used by most professional monitors when converting film to video. So I would select the SMPTE-C color gamut (click Rec. 601 on the calculator) and compute the dE values of the measured values relative to those values.

No, what I said is that each x,y,Y value produces a specific color. If you want the magenta of Rec. 709 then you need to produce the x,y,Y value of the Rec. 709 magenta color regardless of what the primary colors are.

No, you have things backwards. The calculator tells you what color you will get with a magenta signal. If you want a Rec 709 magenta from a magenta signal then you need to produce the Rec 709 x,y,Y values.

Then I stand corrected. Thank you.

I guess the "word on the street" as they say about what to use your calculator for is erroneus. Most of us can't get Rec709 and can't actually do anything to improve the gamut our TV comes with, as you know. "They" say to use the calculator to find out where to put your secondaries if you can't adjust your primaries.

So now, I'm totally in the dark of what the value of this calculator is. If the calculator simply tells you what the dE is, the information isn't really helpful if I can't do anything with it. And if you have a CMS and can move the primaries to Rec709 spots, you don't need to know dEs.

There is a "white" setting in the Samsungs we have, and that setting moves the white spot relative to the primary colors. Changing this white setting requires adjustment of the grayscale. Once done, the result is that the secondaries are in different spots for each different "white" setting, especially magenta, since this white setting moves the white spot mostly along the x axis.

I did calibrations with several of these "white" settings looking for the "best" one. I used the Accupel calculator to figure out which one gave the lowest dEs. The best setting (as defined by lowest dEs overall) results in magenta farther away from it's Rec709 spot than some of the others. Other white settings get magenta closer to Rec709, but the combination of worse xy locations for the other secondaries and worse Y values results in higher dE values per Accupel calculator.

Is this not the proper use of this tool?

Another thing I've been experimenting with using your calculator is choosing a different white spot to set up the grayscale around other than D65, trying to find a white spot that gets the secondaries closer to Rec709 spec. Does this make sense?
kjgarrison is offline  
post #1285 of 1365 Old 05-25-2008, 04:39 PM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by kjgarrison View Post

Then I stand corrected. Thank you.

I guess the "word on the street" as they say about what to use your calculator for is erroneus. Most of us can't get Rec709 and can't actually do anything to improve the gamut our TV comes with, as you know. "They" say to use the calculator to find out where to put your secondaries if you can't adjust your primaries.

It's useful if you wanted to set your TV up so that it would be in the most uniform state to accept something using a simple colorspace matrix conversion (some equipment/software uses that) and then you'd want to set it to the accupel secondaries so everything was uniform and the matrix conversion would apply optimally. It could be useful in working with colorspaces theory and seeing where secondaries and various other things go in a more abstract sense. Neither of those really applies to what's trying to be accomplished in this thread though, certainly not the second part (the first only in scattered cases).

The calculator did point out that the set more or less automatically adjusted the Y's for our primaries to match their altered positions.

One thing I'm not sure of is it better to adjust the Color knob so that it most closely matches the red (G,B)-Y that accupel says goes with our actual primaries, or is it better to try to line up the saturation curves as close as possible on the ColorHCFR chart or to split the difference? I need to check again, but i THINK, those two suggest different settings (which in an ideal world would probably suggest the same thing but the TV doesn't have enough controls) and I thought the former suggested something like 43-44 for color and the latter something like 40ish for Color, but I need to check again, maybe once everything is taken into account it will suggest more or less the same thing. I also seem to recall though that the samsung was not so even on its saturation curves, often having the 50% area either over or undersaturated comapred to the 25% or 100% level so that you need some compromise to begin with.
skibum5000 is offline  
post #1286 of 1365 Old 05-25-2008, 04:52 PM
AVS Forum Special Member
 
skibum5000's Avatar
 
Join Date: Jun 2006
Location: Boston, MA
Posts: 3,826
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 107 Post(s)
Liked: 72
Quote:
Originally Posted by kjgarrison View Post

Then I stand corrected. Thank you.

I guess the "word on the street" as they say about what to use your calculator for is erroneus. Most of us can't get Rec709 and can't actually do anything to improve the gamut our TV comes with, as you know. "They" say to use the calculator to find out where to put your secondaries if you can't adjust your primaries.

So now, I'm totally in the dark of what the value of this calculator is. If the calculator simply tells you what the dE is, the information isn't really helpful if I can't do anything with it. And if you have a CMS and can move the primaries to Rec709 spots, you don't need to know dEs.

There is a "white" setting in the Samsungs we have, and that setting moves the white spot relative to the primary colors. Changing this white setting requires adjustment of the grayscale. Once done, the result is that the secondaries are in different spots for each different "white" setting, especially magenta, since this white setting moves the white spot mostly along the x axis.

I did calibrations with several of these "white" settings looking for the "best" one. I used the Accupel calculator to figure out which one gave the lowest dEs. The best setting (as defined by lowest dEs overall) results in magenta farther away from it's Rec709 spot than some of the others. Other white settings get magenta closer to Rec709, but the combination of worse xy locations for the other secondaries and worse Y values results in higher dE values per Accupel calculator.

Is this not the proper use of this tool?

Another thing I've been experimenting with using your calculator is choosing a different white spot to set up the grayscale around other than D65, trying to find a white spot that gets the secondaries closer to Rec709 spec. Does this make sense?

my gut tells me to just keep D65 as is and adjust all secondaries as close to 709 (or 601) as you can. Sure that will make the triangle formed around D65 by connecting the primaries to secondaries larger than placing magenta in certain other spots (although I believe still much smaller than placing it out in the dervied secondary location), but this set's color engine isn't working in a way to care from what I can see.

If you look at the saturation plots it doesn't draw them from green straight across to magenta, but from green to D65 and then D65 to magenta, so if magenta is on, it's on and it's not really warping anything away from 709 in that whole sector, it seems like the error is all isolated to shades from D65 out towards green and we just have to live with those errors.

Best to keep everything else that we can place onto 709/601 on those spots to minimize more errors it seems to me based upon how the color engine seems to be working and that we are not trying to set it up for use with things will be applying colorspace conversion themselves and exptecting a more uniform space to map into.
skibum5000 is offline  
post #1287 of 1365 Old 06-01-2008, 03:14 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Not a lot different from others I have posted, but there's a twist.

I do not like the amount of color I'm getting from the Accupel settings. I do like how Accupel helps me use the My Color Control white setting to adjust where the secondaries go. And since the wide color gamut puts cyan closer to where Accupel says to put it, I use wide. (Anyway, I have run full calibrations with standard and wide and done A/B comparisons, and there isn't any visible difference as far as I could tell.)

I have settings for Movie and for Standard. For both CR was a little over 1300:1 and gamma was ~2.2 according to ColorHCFR. The settings were actually a product of CalMAN.

There are some differences between ColorHCFR and CalMAN and I'll try to resolve them before I leave you guys for my in-transit LN 46A750 that Samsung has offered to exchange for my TV.

Here are the settings:

Movie
c 90
b 44
s 50
c 46
t 48/52

warm1
wide

white balance
ro 16
go 15
bo 12
rg 19
gg 11
bg 30

my color control blue 30, white 16

For Standard:

same c, b, s, c, t as Movie

color temp normal
wide

white balance
ro 15
go 14
bo 13
rg 30
gg 13
bg 8

MCC white 15 and blue 30 (although on Standard, I don't think the blue slider does anything. In fact the fact that these sliders don't do anything is why Samsung is exchanging ... not TBE or stutter ... although I have TBE some even after fw2004)

For both, HDMI black is normal. None of the other enhancements are on (edge, dyn cont, black adj) other than I always leave noise reduction on Auto.

What's the twist I mentioned? Here it is. At least this is what I did. I turned the main Color control down to 37. To me colors look natural there. I ran a calibration measurement set at 37 and the dE values were way up. I don't care. The grayscale was still as good as I can get it. At 46 it is just too much color, at least to my eyes on my TV. Try for yourself and see where it looks right to you.

These aren't the best measured settings I have found, but they are the best looking settings (IMO).
kjgarrison is offline  
post #1288 of 1365 Old 06-01-2008, 03:53 PM
Newbie
 
lawd's Avatar
 
Join Date: Dec 2007
Posts: 14
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by kjgarrison View Post

(Anyway, I have run full calibrations with standard and wide and done A/B comparisons, and there isn't any visible difference as far as I could tell.)

Pause on a few examples of a bright blue sky and switch between wide and Auto ... I find the wide saturates the sky a lot more, sometimes bordering on unnatural.

The Narnia trailer on PS3 has some blue sky.
lawd is offline  
post #1289 of 1365 Old 06-02-2008, 03:03 PM
Advanced Member
 
Raptor007's Avatar
 
Join Date: Jun 2006
Location: Seattle, WA, USA
Posts: 647
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by kjgarrison View Post

Not a lot different from others I have posted, but there's a twist.

I do not like the amount of color I'm getting from the Accupel settings. I do like how Accupel helps me use the My Color Control white setting to adjust where the secondaries go. And since the wide color gamut puts cyan closer to where Accupel says to put it, I use wide. (Anyway, I have run full calibrations with standard and wide and done A/B comparisons, and there isn't any visible difference as far as I could tell.)

I have settings for Movie and for Standard. For both CR was a little over 1300:1 and gamma was ~2.2 according to ColorHCFR. The settings were actually a product of CalMAN.

There are some differences between ColorHCFR and CalMAN and I'll try to resolve them before I leave you guys for my in-transit LN 46A750 that Samsung has offered to exchange for my TV.

Here are the settings:

Movie
c 90
b 44
s 50
c 46
t 48/52

warm1
wide

white balance
ro 16
go 15
bo 12
rg 19
gg 11
bg 30

my color control blue 30, white 16

For Standard:

same c, b, s, c, t as Movie

color temp normal
wide

white balance
ro 15
go 14
bo 13
rg 30
gg 13
bg 8

MCC white 15 and blue 30 (although on Standard, I don't think the blue slider does anything. In fact the fact that these sliders don't do anything is why Samsung is exchanging ... not TBE or stutter ... although I have TBE some even after fw2004)

For both, HDMI black is normal. None of the other enhancements are on (edge, dyn cont, black adj) other than I always leave noise reduction on Auto.

What's the twist I mentioned? Here it is. At least this is what I did. I turned the main Color control down to 37. To me colors look natural there. I ran a calibration measurement set at 37 and the dE values were way up. I don't care. The grayscale was still as good as I can get it. At 46 it is just too much color, at least to my eyes on my TV. Try for yourself and see where it looks right to you.

These aren't the best measured settings I have found, but they are the best looking settings (IMO).

Thanks! I appreciate you doing a Standard mode setting with Normal color temperature, too. I'll try this out when I get home.

I'll probably end up leaving the color saturation at the calibrated value though; I like having lots of color. But I've noticed even good movie theaters tend to show a lot less saturation than what I'd prefer to see. Maybe you're trying to get yours closer to that look?

Samsung BD-UP5000 (1.3), Samsung 4671 (1013.1), Denon AVR-3808CI, HTD Level 3 5.1
Denon AVR-3300, Denon PMA-700V, Advent Loudspeakers
Raptor007 is offline  
post #1290 of 1365 Old 06-02-2008, 07:00 PM
AVS Forum Special Member
 
kjgarrison's Avatar
 
Join Date: Feb 2007
Location: Northwest Wisconsin
Posts: 2,237
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 12
Quote:
Originally Posted by Raptor007 View Post

Thanks! I appreciate you doing a Standard mode setting with Normal color temperature, too. I'll try this out when I get home.

I'll probably end up leaving the color saturation at the calibrated value though; I like having lots of color. But I've noticed even good movie theaters tend to show a lot less saturation than what I'd prefer to see. Maybe you're trying to get yours closer to that look?

I am not targeting any particular look. All I am saying is that if you take those calibrated settings and decrease color it looks better and more natural to me. When other people start talking about "eyeballing" something, I usually stop reading, but that's exactly what I'm saying here. Leave it at 46, or go to whatever number is believable to your eyes. To mine, and on my TV with my sources, 37 looks about right.
kjgarrison is offline  
Sponsored Links
Advertisement
 
Reply LCD Flat Panel Displays

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