Reading and interpreting calibration charts and data for dummies - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #1 of 181 Old 09-03-2005, 12:49 PM - Thread Starter
Super Moderator
 
Bob Sorel's Avatar
 
Join Date: Jan 2000
Posts: 8,463
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 116
In the last year or so, home calibration of front projectors has become feasible for the average enthusiast, as the prices of test gear have come down and the quality of the instruments has gone up. I feel that more and more we will see people investing in colorimeters, and much like they have taken on the task of performing some of the more simple calibration procedures, like setting brightness, contrast, sharpness, color decoding adjustments, and convergence, they will begin performing more extensive procedures like measuring and adjusting greyscale, gamma, contrast ratio, black level, and light output.

So I figured this would be a good time to discuss what this data means and how it should be applied to get the most performance possible from your projector. I am not writing this to provide a guide, as I am just beginning on this journey myself, but rather to pose questions that I have concerning charts and data in public forum rather than asking them for myself privately.

And by the way, all pictures of charts and data that I will be posting are actual measurements of my InFocus 7210/Greyhawk (edit - Firehawk, not Greyhawk), using the DVI connection from my Panasonic S-97, measured with a Gretag Macbeth EyeOne Pro spectroradiometer running UMR's i1 Pro DCS calibration software in emission mode.

Ok, let's start with the CIE 1931 Tristimulus chart, a chart which provides the user with an overall picture of the color balance:


The squares are for the primary and secondary colors, but of what? I mean, are the 6 squares representing ideal values for the 6 colors, and not actually the measurements from my projector? Well, that would be my assumption, and I would also assume that regardless of how many times I remeasured and what changes I made, those squares would appear at exactly the same coordinates as they are presently. Is this correct?

Now, I can see that there are 6 triangles, and my assumption is that those 6 triangles are the actual measured values from my projector. The red and the blue seem to be pretty much right where they are supposed to be, but as we can see, green is southeast of its ideal values, and is pulling toward yellow. So what do I need to do with green to make it greener and less yellow? And what do those x and y coordinates represent? It looks like I need to go up .04 and to the left .04 units, but what are those units?

Ok, now let's take a look at the secondary colors. They are all off a bit, though nowhere near as much as green is. But there are no secondary color adjustments in my software - only primary gains and biases. So again my assumption would be, if I can move green so that it is on target as well as red and blue, then the secondaries would be pulled into place as well. Is this sound logic? So how do I get green "on target"?

While I am thinking of it, I have the option to measure using two different color gamuts - HDTV or SMPTE-C. I have read "HDTV is recommended for 720p and
1080i tests and SMPTE-C is for 480p and 480i.". My Panny S-97 is an upconverting DVD player, so the source is 480i (DVD) and it is being upconverted to 720p, so which color gamut do I use?

I know that by posting all of these questions I am showing just how stupid I am, but at my age I really don't care any more... I could probably get most , if not all, of these answers offline, but I figured that there would be a lot more people who would read this and learn from it, and they might be more concerned about showing their thorough lack of knowledge than I am. If things go well and other "dummies" post more questions or at least post that they are interested in knowing more and continuing with this, AND some of the more knowledgeable members decide to help out, then I will continue on with other charts and graphs until virtually all of the dumb questions have been asked.

_______________________________

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Bob Sorel is offline  
Sponsored Links
Advertisement
 
post #2 of 181 Old 09-03-2005, 01:48 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
This is the standard CIE chart that most magazine reviews use.

Like a world map - it folds the color wavelengths and their mixtures so they are shown in 2D representation. It is one of many color spaces - but this is the one that video standards are based on.

RGB primaries cannot be changed on most projectors - though you can select the color space on the SP7210 - you cannot edit the colors as those are primarily determined by the color wheel. On the NEC HT1100 you can slide the primaries/secondaries along the triangle to the next color. More expensive projectors you can edit the location of the primaries but you cannot get deeper colors than the color wheel allows.

The RGB gain/offsets are used to edit the mixture of primaries that creates white - which is your target in the middle. From there you draw a line from a primary thru white - and that gets the opposing secondary. If they are off the target or off the line or off the triangle it is not perfect- this can be because your white point is wrong, or your primaries are wrong, or your color matrix is wrong, or your video decoder is wrong, or your tint is wrong. How do you know what is wrong you ask? Years of experience getting it wrong before you get it right....

If you use the 1976 CIE it is a perceptual space (what your brain perceives) - rather than the 1931 which is a physical space (what your eyes see)

You would need to research the DVD player forum to find out what your player does on upconversion with the colorspace. The SP7210 defaults to the HD color space (REC709) - you may need to change it to SD space (REC601) to match your player.

BTW that chart is wrong - the SP7210 has deeper reds than the standards - as does the Optoma H79. ColorFacts has an option called constrain to gamut - maybe similar option here.
krasmuzik is offline  
post #3 of 181 Old 09-03-2005, 03:53 PM - Thread Starter
Super Moderator
 
Bob Sorel's Avatar
 
Join Date: Jan 2000
Posts: 8,463
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 116
Hi Kevin,

Thanks for your help!

Quote:
RGB primaries cannot be changed on most projectors - though you can select the color space on the SP7210 - you cannot edit the colors as those are primarily determined by the color wheel.

That's kinda what I thought. So no matter what adjustments I make to the RGB gains/offsets, the primaries will not change. Is that correct?

Quote:
How do you know what is wrong you ask? Years of experience getting it wrong before you get it right....

Yeah, I'm beginning to figure that out...

Quote:
If you use the 1976 CIE it is a perceptual space (what your brain perceives) - rather than the 1931 which is a physical space (what your eyes see)

I don't think that I have the option, though I will have to check with Jeff on that.

Quote:
You would need to research the DVD player forum to find out what your player does on upconversion with the colorspace. The SP7210 defaults to the HD color space (REC709) - you may need to change it to SD space (REC601) to match your player.

The colors look right...If I needed to change to Rec601 wouldn't it be blatantly obvious that something was wrong?

Quote:
BTW that chart is wrong - the SP7210 has deeper reds than the standards - as does the Optoma H79. ColorFacts has an option called constrain to gamut - maybe similar option here.

What is wrong? I assume that you mean that the red triangle should be elsewhere on the chart, but where? And what exactly is a "deeper" red? Oversaturated?

My measuring technique is far from optimum yet, as I using the emission mode and reading directly off of the screen, so my readings below 30% are rpobably quite inaccurate. Once my new, taller tripod arrives, I plan on remeasuring using the ambient mode by putting a diffuser on the sensor and measuring directly from the projector, which should allow me to get more accurate readings below 30%, and hopefully a more accurate picture of the color balance.

_______________________________

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Bob Sorel is offline  
post #4 of 181 Old 09-03-2005, 03:57 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Look at my review for what I got on the CIE chart.

Usually the SD/HD colorspace issue is most noticeable in greens. I had an email into gregr as I had my accupel set wrong - he explained it pretty good.


Think of white balancing as a painter colors wheel - you have three colors Red, Green, Blue that are you mixing to prime the canvas before you paint the picture (though paint is subtractive color so you actually get Mud, light is additive color so you get White). Think of the video picture as water colors - any color in the canvas will show thru.
krasmuzik is offline  
post #5 of 181 Old 09-03-2005, 04:06 PM
AVS Special Member
 
odyssey's Avatar
 
Join Date: Jan 2001
Posts: 1,088
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Bob,

Is your Greyhawk the regular or the reference version? If it's the regular, it shifts towards blue by about 500 CCT (correlated color temperature) and you should continue to measure from the screen. In other words, if you calibrate the projector directly for D65, the screen will absorb some red and green and the reflected CCT will be about 7000.
odyssey is offline  
post #6 of 181 Old 09-03-2005, 04:15 PM - Thread Starter
Super Moderator
 
Bob Sorel's Avatar
 
Join Date: Jan 2000
Posts: 8,463
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 116
Quote:
Is your Greyhawk the regular or the reference version?

Oops, did I say Greyhawk again? It is a Firehawk, but I don't know if it is the RS or not. I bought just the material itself, and all it said on the packing slip was "bulk material" and never specified anything further.

_______________________________

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Bob Sorel is offline  
post #7 of 181 Old 09-03-2005, 04:18 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
This is what gregr said about choosing the wrong color space.

"I think you are seeing the effect of the wrong color decoder that you
selected in the projector. If you decode 601 encoded signals with a
709 decoder then the green amplitude is reduced by 15.5% so it looks
darker, (which you may be interpreting as more saturated - it isn't,
it's just darker) and the red amplitude is increased by about 9% and
about 10% green is added to the red, so it looks lighter and is more
orange (it really isn't less saturated, it's a different hue.)"
krasmuzik is offline  
post #8 of 181 Old 09-03-2005, 04:22 PM
AVS Club Gold
 
Alan Gouger's Avatar
 
Join Date: Jan 1999
Location: Florida
Posts: 18,726
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 16
Quote:
Reading and interpreting calibration charts and data for dummies

Im really sorry, I wanted to contribute something to this thread but have to exclude myself.
Alan Gouger is offline  
post #9 of 181 Old 09-03-2005, 04:41 PM
AVS Special Member
 
odyssey's Avatar
 
Join Date: Jan 2001
Posts: 1,088
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Bob Sorel View Post

Oops, did I say Greyhawk again? It is a Firehawk, but I don't know if it is the RS or not. I bought just the material itself, and all it said on the packing slip was "bulk material" and never specified anything further.

I think that there is only one version of the Firehawk. Unlike the Greyhawk, I have not measured it and I don't know if it's D65 neutral.
odyssey is offline  
post #10 of 181 Old 09-03-2005, 08:06 PM
umr
AVS Club Gold
 
umr's Avatar
 
Join Date: Oct 2002
Location: USA
Posts: 10,137
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 4 Post(s)
Liked: 29
Bob,

It looks to me like you have extraneous light contaminating your measurements or your light levels are too low. Note how green, yellow and red do not form a line on your chart. That should not be the case. You should be able to draw a line from the primaries and the secondary should be on the line. When that is not the case you need to examine your measurement technique.

It is best to use ambient mode for a front projector and place the instrument very close to the lens. The attached file shows how the color triangle should be shaped.
LL

W. Jeff Meier


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
umr is online now  
post #11 of 181 Old 09-04-2005, 05:42 AM - Thread Starter
Super Moderator
 
Bob Sorel's Avatar
 
Join Date: Jan 2000
Posts: 8,463
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 116
Quote:
Im really sorry, I wanted to contribute something to this thread but have to exclude myself.

Sure you can, Alan! I represent the "dummy" contingent, so you can contribute by adding to the knowledgeable posts...

Quote:
I think that there is only one version of the Firehawk. Unlike the Greyhawk, I have not measured it and I don't know if it's D65 neutral.

Actually I think there are 2 versions, but from what I have heard the shift toward blue on the original Firehawk is so small as to be neglible for these purposes.

Quote:
This is what gregr said about choosing the wrong color space.

Regarding the color space issue of the 7210, I found a menu with 4 settings:

1. Auto (default)
2. RGB
3. Rec709
4. Rec601

My setting is "1. Auto". When I tried unchecking the auto box, the projector was now set to RGB (that was the setting that "auto" had chosen for my S-79 [edit: S-97]). I then tried both Rec709 and Rec601, and either of those settings resulted in a very very very green picture, totally unwatchable in any way, so there was no mistake about it - the correct setting is "RGB". How does "RGB" relate to using either SMPTE-C or HDTV as measurement choices? It would seem to me that RGB probably equals SMPTE-C, right? If I am correct, then I should have measured using the SMPTE-C color gamut and not the HDTV.

Quote:
It looks to me like you have extraneous light contaminating your measurements or your light levels are too low. Note how green, yellow and red do not form a line on your chart. That should not be the case. You should be able to draw a line from the primaries and the secondary should be on the line. When that is not the case you need to examine your measurement technique.

Thanks, Jeff. I only posted one graph so far, and when I post the other graphs, I am sure that we will see problems in my mearsuring technique become more obvious. At this point my intent is just to discuss the graphs to better understand what we are seeing and what it means, good or bad, and you and KRAS have shown how your experience and trained eyes have spotted that something was wrong just from that one graph alone...

Quote:
It is best to use ambient mode for a front projector and place the instrument very close to the lens. The attached file shows how the color triangle should be shaped.

Yes, and that is my intention. I will remeasure in ambient mode as soon as my new 8.5' tripod arrives...

_______________________________

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Bob Sorel is offline  
post #12 of 181 Old 09-04-2005, 03:39 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Bob Sorel View Post

Regarding the color space issue of the 7210, I found a menu with 4 settings:

1. Auto (default)
2. RGB
3. Rec709
4. Rec601

My setting is "1. Auto". When I tried unchecking the auto box, the projector was now set to RGB (that was the setting that "auto" had chosen for my S-79). I then tried both Rec709 and Rec601, and either of those settings resulted in a very very very green piucture, totally unwatchable in any way, so there was no mistake about it - the correct setting is "RGB". How does "RGB" relate to using either SMPTE-C or HDTV as measurement choices? It would seem to me that RGB probably equals SMPTE-C, right? If I am correct, then I should have measured using the SMPTE-C color gamut and not the HDTV.

The four settings above are to match the color processing in the projector to the type of signals it is being fed. Apparently you are providing the projector with RGB signals. The Rec 709 and Rec 601 settings are for YCbCr (or YPbPr) signals, (Rec 709 is for HD and Rec 601 is for SD YCbCr to RGB color decoding), which is why they look completely wrong with your RGB signals. The Auto setting will default to RGB when you use DVI signals, but with HDMI signals it will "negotiate" with the source to automatically pick a format that is common to the source and the display, and that will usually be YCbCr (either Rec 601 or Rec 709 depending on the whether the signals are SD or HD).

Greg Rogers
AccuPel
Widescreen Review

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #13 of 181 Old 09-04-2005, 04:40 PM - Thread Starter
Super Moderator
 
Bob Sorel's Avatar
 
Join Date: Jan 2000
Posts: 8,463
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 116
Thanks for clearing that up, Greg! One more question - when I measure it (Panny S-97 720p DVI to IF 7210) with my EyeOne, should I choose the HDTV or SMPTE-C gamut?

_______________________________

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Bob Sorel is offline  
post #14 of 181 Old 09-04-2005, 06:01 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Bob if you are upconverting in the Panny such that it send RGB over DVI - that means it did the video decoding (YCbCr->RGB) So you need to know when the Panny upconverts to 720P does it assume SD or HD colorspace in the process? Or maybe it assumes HD colorspace but then does not send a full green so as to represent the max SD green which is less than HD? Or maybe it sends a full green thinking you have a SD display - yet full green on the HD display is different?


There seems to be no consistency on how that is done - but see if the DVD forum knows what your specific player is doing.

Yet another reason I am happy with gregr's Accupel black box - you know what you are getting and what to measure!

Digital interconnects have not made video connections bulletproof at all - it just added yet another layer of variables to deal with!
krasmuzik is offline  
post #15 of 181 Old 09-04-2005, 06:11 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Bob Sorel View Post

Thanks for clearing that up, Greg! One more question - when I measure it (Panny S-97 720p DVI to IF 7210) with my EyeOne, should I choose the HDTV or SMPTE-C gamut?

The gamut you select shouldn't affect the measurements (unless you have the measurements constrained to the gamut - forced to be wrong - I have no idea why that option is included). I don't use this software, but I suspect the gamut you chose just changes the reference points on your plots, not the measurements that you take.

But something is clearly wrong on the chart at the start of this thread. The yellow measurement is outside the line between the green and red measurements, and that is physically impossible unless the projector has special processing to adjust the primaries separately from the complementary colors, and the 7210 doesn't have that type of processing.

Greg Rogers
AccuPel
Widescreen Review

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #16 of 181 Old 09-04-2005, 06:15 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by krasmuzik View Post

Digital interconnects have not made video connections bulletproof at all - it just added yet another layer of variables to deal with!

Ain't that the truth!

Greg Rogers
AccuPel
Widescreen Review

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #17 of 181 Old 09-04-2005, 06:24 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
gregr

Any idea why ColorFacts defaulted to 'constrain to gamut' at some point - guitarman and I figured that one out after your H79 review - and never got a response from ColorFacts forum (which is now shut down for transferring to the parent company support forum)
krasmuzik is offline  
post #18 of 181 Old 09-04-2005, 06:29 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Bob

Actually I have the same perfect hue but oversaturated chromaticity for Yellow - that does not lie on the Red/Green line - see my .sig review CIE chart. It is barely noticeable in the CIE76 chart. There must be something in the spectrum that is confusing the sensors (Spyder2 and Eye One).

I think htprojectors.com has the spectral plots of the SP7205.....yep the lamp has a spike at yellow-green then another spike at yellow-orange - but a dip at yellow. I bet that confuses the instruments that maybe do not sample exactly at reference yellow wavelengths (but that is an Eye-One spectral plot). So maybe it is a matrix translation issue from spectral to CIE?
krasmuzik is offline  
post #19 of 181 Old 09-04-2005, 06:41 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by krasmuzik View Post

gregr

Any idea why ColorFacts defaulted to 'constrain to gamut' at some point - guitarman and I figured that one out after your H79 review - and never got a response from ColorFacts forum (which is now shut down for transferring to the parent company support forum)

No idea. I can't even think of a reason that ColorFacts has that option. It can only lead to wrong measurements.

Greg Rogers
AccuPel
Widescreen Review

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #20 of 181 Old 09-04-2005, 06:59 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Mark Hunter says here it is a greyscale vs. primary accuracy option....

http://archive.avsforum.com/avs-vb/s...05#post2821105

Not sure why that would be but that is what he said....
krasmuzik is offline  
post #21 of 181 Old 09-04-2005, 08:46 PM
Advanced Member
 
ghibliss's Avatar
 
Join Date: Aug 1999
Location: New York, NY
Posts: 740
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
The reason for providing "Constrain" primaries is due to the fact that when XYZ colorspace is converted to xyY colorspace an exact match for the coordinates is not always possible! You may in fact end up with a negative value (which is obviously impossible) which will mean that the primary lies outside the boundary of the selected colorspace gamut. When constraining the primaries you are mapping the displays measured RGB primaries and using them on the CIE graph. Since the actual measured values are being used the primaries will always lie directly at the points of the triangle.

The constrain function also allows the bar graph function to be more precise as the distance from the White target to the primaries changes with the colorspace selected. For example the distance from D65 is diffferent to Green for BT-709 then for BT-601 and to a much smaller degree for the other color spaces. This function would not provide accurate percentage values if you did not constrain the primaries.

Cliff Plavin
Progressive Labs
ghibliss is offline  
post #22 of 181 Old 09-04-2005, 09:08 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
I presume you mean when converting XYZ to RGB that you can be negative - assuming your actual primaries are out of the xy gamut selected. xy is just normalized XY so not sure how it can be negative. So I suspect Mark's comment about more accurate greyscale - really means more accurate % on the RGB barchart/histogram used to help tune the greyscale - but the actual xy target is still accurate either way.

I always make sure this option to constrain is off - the %RGB does not match any projectors RGB adjustment scales anyways so is of little help in being exact beyond math perfection.
krasmuzik is offline  
post #23 of 181 Old 09-04-2005, 11:56 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by ghibliss View Post

The reason for providing "Constrain" primaries is due to the fact that when XYZ colorspace is converted to xyY colorspace an exact match for the coordinates is not always possible! You may in fact end up with a negative value (which is obviously impossible) which will mean that the primary lies outside the boundary of the selected colorspace gamut.

Not true.

x = X / (X+Y+Z)
y= Y / (X+Y+Z)

therefore x and y are always positive because X,Y,Z are always positive.

The X,Y,Z tristimulus values are always positive because they are the integration of the SPD (spectral power distribution) (always positive for any real primaries) weighted by the x-bar, y-bar, z-bar CIE color matching functions (CMFs). The CMFs are entirely positive values. So there is no way to get a negative x,y value from physical primaries, which is what we are measuring.

Primaries that lie outside the standard color triangle gamuts do not generate negative x,y values (as you certainly know). You are confusing issues. Colors that lie outside the display's actual primary triangle would require negative amplitudes of the primary colors, which is why they can't be displayed. But the actual primaries, which is what we are trying to measure, can lie anywhere within the spectral locus without requiring negative RGB amplitudes.

Quote:
When constraining the primaries you are mapping the displays measured RGB primaries and using them on the CIE graph. Since the actual measured values are being used the primaries will always lie directly at the points of the triangle.

No, that's exactly the problem. Constraining the primaries means that the display's RGB primaries are not being represented correctly on the CIE x,y diagram. They are being arbitrarily constrained within some selectable color gamut (color triangle). This is obvious when you use the software since it produces totally incorrect CIE x,y values when the display primaries are outside arbitrary selected color gamut constraints.

Mark Hunter acknowledges that the "constrain color gamut" option should be turned off in order to measure the color primaries accurately in the link posted above. This is a well-known operator trap in the ColorFacts software.

Quote:
The constrain function also allows the bar graph function to be more precise as the distance from the White target to the primaries changes with the colorspace selected. For example the distance from D65 is diffferent to Green for BT-709 then for BT-601 and to a much smaller degree for the other color spaces. This function would not provide accurate percentage values if you did not constrain the primaries.

I'm don't remember exactly what the bar graph function does in ColorFacts. Does it display the relative RGB levels required to produce D65 in order to help the user adjust the RGB gain/bias to calibrate the grayscale? If so, then it should compute those levels relative to the measured values of the actual RGB primaries, not some arbitrarily constrained primary coordinates. There would be no point in calculating the relative RGB levels necessary to achieve D65 gray, using RGB primary chromaticities (or SPDs) that are different than what the display actually has to use. But perhaps the bar graph performs some other function.

Mark says the "constrain color gamut" option should be used to optimize grayscale calibration. I would trust that he knows how his product works.

Greg Rogers
AccuPel
Widescreen Review

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #24 of 181 Old 09-05-2005, 09:22 AM
AVS Special Member
 
therealgeno's Avatar
 
Join Date: Dec 2004
Location: Knoxville, TN
Posts: 1,434
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by gregr View Post

Not true.

x = X / (X+Y+Z)
y= Y / (X+Y+Z)

therefore x and y are always positive because X,Y,Z are always positive.

The X,Y,Z tristimulus values are always positive because they are the integration of the SPD (spectral power distribution) (always positive for any real primaries) weighted by the x-bar, y-bar, z-bar CIE color matching functions (CMFs). The CMFs are entirely positive values. So there is no way to get a negative x,y value from physical primaries, which is what we are measuring.Greg Rogers
AccuPel
Widescreen Review

I thought this was a thread for DUMMIES !

Anyway, I have a dummy question: Kras, I've read several times where you have pointed out that Infocus sort of "cheats" on its green so that it can produce higher lumens - is that one of the reasons why Bob's green is off or is that something totally different?

We contend that for a nation to tax itself into prosperity is like a man standing in a bucket and trying to lift himself up by the handle. (Winston Churchill)
therealgeno is offline  
post #25 of 181 Old 09-05-2005, 12:18 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
therealgeno

As long as a few dummies post - then the subject is still valid


Bob's green appears to be way off because he is using the original CIE chart. Infocus is aware of the 1976 chart in which the light colors Yellow, Green, Cyan are given less area (and thus errors are less important) than the dark colors Blue, Magenta, Red which are given more area (and thus errors are more important).

So Infocus does push green towards Yellow - as does Optoma who also seems to have recently 'discovered' you can make brighter projectors this way (as opposed to earlier ones that used an overstated Green and were thus dim - but were noticeably different color). Often marketers confuse 'different' with 'better'....

Now manufacturers try to make their projectors noticeably different (Infocus, Sim, Optoma) by pushing reds beyond the Orangy-Red specified in the standards - since pushing Green just makes it dim. Using the latest CIE chart you can see why this is not a good thing - even if it may be pleasing natural Reds it is much more inaccurate than compromising Green for brightness. Video is not capable of a natural Red - so it does look different when it does have a natural Red!

But using the original CIE chart - you would think Red is fine and Green is way off - but that is not what you perceive! See my SP7210 review for both charts to see the differences (while Bob figures out how he can redo his chart!)

That is why our resident expert gregr is using the latest CIE charts in WSR.

Anyways to add to your OCD - you will be pleased to know that your SP4805 has more accurate SD colors than the SP7210. It is not as bright and is thus higher contrast, the greens are more natural per the spec, and the reds are more orange per the spec. (the attached plot is for the HD spec)

Infocus by krasmuzik

Yes I just threw in a curve ball with new charts!
krasmuzik is offline  
post #26 of 181 Old 09-05-2005, 12:24 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
gregr

It was my understanding that was why Colorfacts has you measure primaries before greyscale - so that RGB% can be computed to the actual projector rather than the standard that is never met. If so the 'constrain to gamut' option is still rather unclear why it exists.
krasmuzik is offline  
post #27 of 181 Old 09-05-2005, 02:31 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by krasmuzik View Post

That is why ... gregr is using the latest CIE charts in WSR.

The newer (1976) CIE charts that krasmuzik is talking about are called CIE u'v' diagrams. The idea is to convert the x,y coordinates into perceptually more uniform u'v' coordinates, i.e. the ability to perceive differences between colors is more nearly equal to the distance between the colors over the entire u'v' diagram.

However, the u'v' diagram still doesn't express my subjective reaction to color errors. It does a better job of expressing one's ability to discern color errors, but it doesn't express how much I dislike the color error. For instance, I dislike a discernable red color error more than a discernable green color error, even though the errors may be the same distance on the u'v' chart. And I dislike green errors more than blue errors on the u'v' chart. Put another way, I least tolerate red color errors and most tolerate blue color errors. I believe this is a function of how colors typically occur in real images. I'm most sensitive to errors in skin tones, and red primary errors have the most effect on skin tones. Then I'm real sensitive to objects that I know should be red, but look either orange or crimson. After that, I'm particularly aware of green color errors in landscapes, grass, trees, etc. Of course there are exceptions when you are very familar with a particular object that may be rendered incorrectly. For instance, if your favorite sports team wears a bright blue uniform, you may be particularly upset if it is not rendered with exactly the correct hue and saturation. I think my subjective reactions to color errors are fairly typical of most people, but I've never seen a study on that subject.

So interestingly, despite the fact that the CIE u'v' diagram is perceptually more uniform for discerning color errors, the CIE x,y diagram is sometimes a better measure of my subjective dislike for color errors. (It would take another "uniform dislike" diagram to best capture my subjective reaction to color errors.) Nevertheless, I started using the CIE u'v' diagrams back in Issue #82 of WSR. We are now at issue #102. Over that time I've gone back and forth between the x,y, and u'v' diagrams, and sometimes show both in my reviews. Some readers have objected to the u'v' diagrams. They are not as familar with them and make their own measurements using x,y. So by showing both it helps to increase familiarity with the u'v' charts. Othertimes, I still use the x,y charts when I think they better represent my subjective reaction to the color errors.

Greg Rogers
AccuPel
Widescreen Review

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #28 of 181 Old 09-05-2005, 03:02 PM
AVS Special Member
 
therealgeno's Avatar
 
Join Date: Dec 2004
Location: Knoxville, TN
Posts: 1,434
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by krasmuzik View Post

therealgeno

Bob's green appears to be way off because he is using the original CIE chart. Infocus is aware of the 1976 chart in which the light colors Yellow, Green, Cyan are given less area (and thus errors are less important) than the dark colors Blue, Magenta, Red which are given more area (and thus errors are more important).

So if Bob was using the correct charts, then green would be closer, if not dead on EVEN THOUGH Infocus pushes green towards yellow?

BTW, gregr, thank you for your last explanation - I could actually follow that one and it makes very good sense.

Could someone post what the 1976 CIE chart looks like - is it the one Kras just linked? Cause I have no idea what I am looking at when those charts are in front of me .

And for another dummy question, what do the x,y coordinates mean? And there has been mention of z, although I don't see any z coordinate on the CIE 1931 chart - perhaps it is only on the 1976 chart.

Oh, thanks for starting this thread Bob! I love TRYING to understand this stuff - so thanks to all the vets for chiming in!

We contend that for a nation to tax itself into prosperity is like a man standing in a bucket and trying to lift himself up by the handle. (Winston Churchill)
therealgeno is offline  
post #29 of 181 Old 09-05-2005, 03:51 PM
AVS Addicted Member
 
krasmuzik's Avatar
 
Join Date: Oct 1999
Location: NewPort, VA
Posts: 11,270
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
gregr

The delC* and delH* charts is based on u*v* which is u'v' scaled by the lightness (L*) of the colors. This tends to temper the u'v' chart which tends to minimize Green error subjectively too much , but does not overstate Green error as much as x,y.

I think these new charts work because green is lighter than blue so it brings out green error a little more even though there are more blue hues it is darker and less objectionable errors. I've been using these charts a few months now- and they generally agree with my subjective impressions of the displays colors, greyscale and gamma.

I suspect the SP7210 having too vivid reds and pale yellow greens but perfect blues does not align with your likes - I need to check your review to see how different the H79 is - as I think the colors are somewhat similar (aside from factory greyscale and 480i decoding). Maybe slightly more red, more green, less blue?


therealgeno

Read Poynton's color science FAQs - he can explain it a lot better than we can! It is a very complicated science.

In a nutshell - XYZ matches your eyes response curves in the blue, green, red spectra - and xy are normalized versions of that (thus you don't need z). CIE u', v' charts are perceptually weighted versions of the CIE x,y color chart, while u*, v* scales u'v' by the lightness L*, which is a perceptual (gamma curve) version of luminance Y.

http://www.poynton.com/ColorFAQ.html

My SP7210 CIE charts are in my .sig link.... click on the review links marked CIE1931 and CIE1976.

The charts I just posted are new ones that I am developing.....
krasmuzik is offline  
post #30 of 181 Old 09-05-2005, 04:08 PM - Thread Starter
Super Moderator
 
Bob Sorel's Avatar
 
Join Date: Jan 2000
Posts: 8,463
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 66 Post(s)
Liked: 116
Yes, many thanks to Greg and Kevin for explaining these things to us!

I'm sorry that I haven't posted recently, as I've been plenty busy, but I have been reading and keeping up with the posts. I wonder if Jeff has been lurking as well and might consider adding CIE 1976 charts in a future release.

The next chart is from the same measurement session, and it will hopefully help to explain some of the readings in the CIE 1931 graph:

Here we can clearly see that the readings I took below 30% stimulus are inaccurate. I believe this was due to the fact that my projector was mounted normally in its ceiling mount and the readings were taaken in emission mode off of the screen. AAs I can see it, there would have been two better ways to take these measurements that would have resulted in more accuracy in the low percentage stimuli:

1. Move the projector closer to the screen so that the projected image was MUCH smaller (like around 32"X18"), resulting in much higher ftL coming off of the screen and thus moving the measurements into an easier range for the EyeOne to measure.
2. Leave the projector where it was, use a tripod, put a diffuser on the EyeOne sensor (included in the Beamer package), and then take the measurements in the ambient mode, reading directly from the projector lens from a distance of about 2 feet. This would have a very similar effect of moving the measurements to a higher range like in solution #1, except that any color shift happening in the screen would not be considered. Since I am using a Firehawk, my belief is that this color shift would be neglible, but it is still less accurate (IMHO) than taking the measurements off of the screen.

Please also note that the inaccurate measurements in the low percent stimuli also throws any meaningful CR calculations right out the window, though it appears as if no harm was done to the gamma curve measurement, as no numbers were included in its calculations. I also assume that the color coordinates are being affected, as well as the obvious percentage stimulus errors. My intention is to remeasure in ambient mode from about 2 feet once my new 8.5' tripod arrives here (my current tripod is too short for this type of measuring).

Comments? Further clarifications of what we are seeing here?

_______________________________

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Bob Sorel is offline  
Reply Display Calibration

User Tag List

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