Calibrators and DVE users—cnet high and low light readings?

I’m trying to understand the cnet hdtv tests.
What does cnet mean here where they explain their test procedures for color temperature:


“”Before color temp
Example result: 6,057/6,784K
This is the out-of-the-box color temperature in Kelvin, measured using the best-available user-menu presets, usually the Warm color temperature and the Movie picture preset. We measure both the high and low light levels from Digital Video Essentials: HD Basics Blu-ray disc (Advanced Video Test Patterns, Chapters 24-41). The exact levels we use depend on the type of TV being reviewed, but they're generally 20% and 80%. The rating is based on how close the measurements come to 6500K. “”

Then in the Geek Box itself it says



“”Geek box
TEST.......................................RESULT.......... SCORE
Before color temp (20/80) .... 6666/6956.......... Average
“”


MY QUESTIONS:

• What high and low light levels are they talking about? Light levels of what?

• “20% and 80% “ of what?

• Does the 20% refer to high light and the 80% refer to low light ?

(notice the order when they say -- “we measure both the high and low light levels…” and then refer to “generally 20% and 80%...” later on and in the geek box it’s “ 20/80 “ .


I’ve e-mailed the cnet reviewer with no response. I’m just needing clarification on this reporting, at this time. Thanks.

References:
How we test tvs: http://reviews.cnet.com/Labs/4520-6603_7-5109683-3.html
Geek box: at end of each review, e.g. at end of page 2 of this xbr8 review;http://reviews.cnet.com/flat-panel-tvs/sony-bravia-kdl-55xbr8/4505-6482_7-33060615.html?tag=box4505

THANKS for the help
TVbc

100% is white signal
0% is black signal

20% is grey with 20% signal in referenz to the white signal.
80% is grey with 80% signal in referenz to the white signal.

Exactly as mentioned above. They are displaying greyscale test patterns which consist of varied levels of white output. As mentioned above 100% is full brightness and white, as it decreases it goes from towarda grey and then black.

The two readings are for low(20%) and high(80%). The ideal reading is 6500K.

Daniel

D65 on a CIE chart is the ideal reading. You can be at 6500k and not be correct. i.e too much/too little green

They're measuring greyscale at two points, 20% and 80% stimulus. Black is 0%, white is 100%, half-gray is 50%. These are referenced to signal, not light %, and this is normal.

thank you mlaun, dachness, lee gallagher and chriswigg for your answers
to my immediate request about what the 20% and 80% refer to in the cnet
results of color temperature testing.

they have helped me start to get an understanding on this -- enough to do some
more research -- and enough to ask some more questions, based on what seem
like different ideas on this from different sources encountered during this long
weekend.

about what the 20 and 80 refer to:
* cnet says it is % of light
* you here are saying it is % "signal", as others in my research
* others have indicated it is luminance and a % of IRE values

is this just a matter of semantics? distinctions without a difference? --or perhaps
cnet dumbing it down to make it more understandable to the noobs like me? if technically
it is % signal, is % light fundamentally wrong, or is that acceptable on some level?
and is reference to IRE an even more specific, better, way?

what is the "signal", exactly. from a signal generator? i've learned that what is
displayed is a series of small 'windows' of different shades of whites/greys as
mentioned here from 0 black to 100 white.

does this mean that the image generated is the same and only changed by changing
the *amount* of signal? [i got the impression from one explanation that the
different grays from black to white are only the result of the difference in
luminance,(brightness) ]. i'm clearly lost here trying to understand the concept,
not helped by different designations i've come across trying to learn about this.

Finally, i've been talking here about the Color Temperature tests and
results. but, this looks the same as the greyscale test, which cnet reports
seperately. am i correct that the the 20% and 80% are just points near each
end of the greyscale tests? if so, why are these apparently commonly reported
and discussed seperately?

i need to pin down each little thing as i go along, or i won't get the next thing,
so i appreciate your help along the way.


thanks for the help (i am getting there)

TVbc

thank you mlaun, dachness, lee gallagher and chriswigg for your answers
to my immediate request about what the 20% and 80% refer to in the cnet
results of color temperature testing.

they have helped me start to get an understanding on this -- enough to do some
more research -- and enough to ask some more questions, based on what seem
like different ideas on this from different sources encountered during this long
weekend.

about what the 20 and 80 refer to:
* cnet says it is % of light
* you here are saying it is % "signal", as others in my research
* others have indicated it is luminance and a % of IRE values


CNET is being lazy in its terminology so as not to confuse laypersons.

% signal is basically the % 'brightness' we percieve the display to be, with 50% being half gray. This is also usually (but not always the same) as IRE. IRE is a horrible unit that should die and go away forever. Read the link in my signature for more about that.

The thing is that we see light nonlinearly, so the % we mean here is subjective, which is represented in the same way by the signal, thus when we say 50%, we all understand this to be 50% lightness, or halfway between black and white: gray. However, in light intensity, this is actually 18%. This makes things more confusing than necessary.


is this just a matter of semantics? distinctions without a difference? --or perhaps
cnet dumbing it down to make it more understandable to the noobs like me? if technically
it is % signal, is % light fundamentally wrong, or is that acceptable on some level?

For a layperson, you probably don't know or care that we percieve light nonlinearly, that this is why gamma exists, or that when you see something that appears halfway between white and black it's actually not 50% light intensity, but 18% again because we see nonlinearly. 50% lightness is 18% luminance. Confused yet? ;)

and is reference to IRE an even more specific, better, way?

No. IRE is a horrible unit that should never be used in my opinion.

what is the "signal", exactly. from a signal generator? i've learned that what is
displayed is a series of small 'windows' of different shades of whites/greys as
mentioned here from 0 black to 100 white.

The signal is used to represent light, but it is inverse-compensated (gamma corrected) for both the response curve of a CRT and our eyes, which happen to be almost the same. Because of this, it makes it easy to be lazy because the % of the signal basically matches what we percieve as % of brightness of an image, or

does this mean that the image generated is the same and only changed by changing
the *amount* of signal? [i got the impression from one explanation that the
different grays from black to white are only the result of the difference in
luminance,(brightness) ]. i'm clearly lost here trying to understand the concept,
not helped by different designations i've come across trying to learn about this.

I'm not sure what you're asking.

http://www.poynton.com/PDFs/GammaFAQ.pdf