AVS Forum banner
1 - 13 of 13 Posts

·
Registered
Joined
·
13,998 Posts
Discussion Starter · #1 ·
I had an interesting discussion and demonstration the other day about why one could dial in the triangle corner and mid points on a digital display to precisely match the 709 standard.... but still not be looking at properly reproduced color space on the display.


How? Because one might need to over or undershoot one corner, for example, to pull the colors that rest in the center of the triangle to the right spot. That is, one cannot expect all the points within the triangle to line up correctly with the standard just because the edges line up with the edges of the standard.


Or another way to say it: one doesn't only look at pictures with those extremes of color present, one also looks at pictures with a combination of colors, that fall at various points within the triangle. And if those points are "off" from where they should be (which can happen since it's not a "linear" consistent relationship on each display between the color points within the triangle and those at the edges of the triangle), then one could have hit the mark for the extreme cases, and missed it for some very common cases within the colorspace.


I may not be describing this well, and am surely oversimplifying the technology and physics involved, but I was partially convinced by a verbal explanation and then further convinced by a demonstration -- where missing the edges of the triangle actually resulted in more accurate color in the center of the colorspace range (ie, hitting closer to the standard).


This is of course display dependent. I understand the Sony lcd mastering monitors get the whole space correct. And that the legendary last generation of Kuros did, as well.


Am I smoking expensive dope? Both the logic and the demonstration were convincing.
 

·
Registered
Joined
·
12,649 Posts

Quote:
Originally Posted by nathan_h /forum/post/19649386


I had an interesting discussion and demonstration the other day about why one could dial in the triangle corner and mid points on a digital display to precisely match the 709 standard.... but still not be looking at properly reproduced color space on the display.


How? Because one might need to over or undershoot one corner, for example, to pull the colors that rest in the center of the triangle to the right spot. That is, one cannot expect all the points within the triangle to line up correctly with the standard just because the edges line up with the edges of the standard.


Or another way to say it: one doesn't only look at pictures with those extremes of color present, one also looks at pictures with a combination of colors, that fall at various points within the triangle. And if those points are "off" from where they should be (which can happen since it's not a "linear" consistent relationship on each display between the color points within the triangle and those at the edges of the triangle), then one could have hit the mark for the extreme cases, and missed it for some very common cases within the colorspace.


I may not be describing this well, and am surely oversimplifying the technology and physics involved, but I was partially convinced by a verbal explanation and then further convinced by a demonstration -- where missing the edges of the triangle actually resulted in more accurate color in the center of the colorspace range (ie, hitting closer to the standard).


This is of course display dependent. I understand the Sony lcd mastering monitors get the whole space correct. And that the legendary last generation of Kuros did, as well.


Am I smoking expensive dope? Both the logic and the demonstration were convincing.

It's possible to measure the primaries and secondaries at multiple saturation points, like 25%, 50%, 75% and the usual 100%. Also, measurements can be taken at either 75% or 100% stimulus/brightness. So, it's possible to map some points within the gamut itself and not just the boundaries of the gamut at the outer six points.
 

·
Registered
Joined
·
2,914 Posts
Checking 25/50/75 saturations is a must for calibration imo - with a lot of sets it's possible to line up 100sat points, have a "perfect triangle", yet the lower saturations can be wildly off.
 

·
Registered
Joined
·
13,998 Posts
Discussion Starter · #5 ·
So I guess that means that is *is* possible to measure points *other than* the corners and midpoints along the edges of the triangle. Or, in other words, it's possible to measure points within the triangle, and see if the are in the right location for a given display (and settings) versus the spec.


It follows, I suppose, that some of those measurements might be spot on, when the corners are wrong, or vice versa, depending on the qualities of the display itself.
 

·
Registered
Joined
·
214 Posts
You can also hit the Rec709 color points but have luminance errors, resulting in poor color performance. The CIE chart doesn't show this dimension of color.
 

·
Registered
Joined
·
13,998 Posts
Discussion Starter · #7 ·
Interesting. So one might have poor color performance, and a perfect CIE chart, or very good color performance, and an apparently inaccurate CIE chart (if the only points one measures are the extremes).


Then, with displays that are imperfect (which is most) it really does become not just a measurement task, but also a set of intelligent compromises, to achieve the best performance for the most critical pieces of the most common source materials to be watched on a particular display, in a particular setting.
 

·
Registered
Joined
·
12,649 Posts

Quote:
Originally Posted by Lee Gallagher /forum/post/19650859


You can also hit the Rec709 color points but have luminance errors, resulting in poor color performance. The CIE chart doesn't show this dimension of color.

To add to this, color luminance can also shift depending on the whether the saturation is 25%, 50%, 75%, or 100%. So, getting the most accurate luminance across various saturations and minimal xy shifts (hue) across saturations (ideally a straight line from the center to the edge) is the only way to make sure color is accurate always. However, such an approach is very tedious and time-consuming to follow, which might make it impossible for professional calibrators who don't have enough time to dedicate to the process. After all, just calibrating CMS at 100% saturation alone can be quite time intensive (along with maintaining accurate grayscale/gamma in the process).
 

·
Registered
Joined
·
12,649 Posts
Quote:
Originally Posted by engelba
All of this is very true... I had to make the same kind of choices... do I want 100% saturation to be right and 75%,50%,25% to be wrong or the other way around? I of course chose to have my 100% wrong and the rest right (or closer to right anyways).


You can see this on my CIE diagram quite elegantly....
http://www.avsforum.com/avs-vb/showthread.php?t=1295323
what does the luminance shifts graph look like?
 
1 - 13 of 13 Posts
Top