Calibrating Quantum Dot lit displays? - AVS Forum
Forum Jump: 
Reply
Thread Tools
post #1 of 482 Old 08-30-2014, 01:44 PM - Thread Starter
AVS Special Member
 
nuke's Avatar
 
Join Date: Jul 2001
Location: CA, USA
Posts: 1,159
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 165 Post(s)
Liked: 84
Calibrating Quantum Dot lit displays?

I've got a Sony XB65X900A display. One of the unique features of this model is it has a quantum dot back light. (Sony branded it "Triluminous" like they did their RGB LED lit models a few years ago).

Quantum dots are different than white LED phosphors and emit in much narrower bandwidths. This can allow for a wider color gamut when the emission peaks are well-aligned with the RGB filters employed in the LCD.

In just observing the light through a simple diffraction grating, I can see the RGB peaks are pretty narrow compared to the white LED and CCFL back lit displays in my house. So, it isn't all marketing BS.

So I just got a I1-Display Pro 3 and have been experimenting with it and the software it comes with along with HFCR software. I notice there are correction matrices for various types of display and LCD backlighting. I'm still experimenting, but I'm finding the "OLED" setting for the sensor to better than some of the others.

Anyone characterized this type of display with a spectrophotometer, or even better, have a correction matrix for the I-1DP3?

nuke
nuke is offline  
Sponsored Links
Advertisement
 
post #2 of 482 Old 08-30-2014, 07:26 PM
AVS Special Member
 
WiFi-Spy's Avatar
 
Join Date: Feb 2004
Location: Seattle, WA
Posts: 2,320
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 30 Post(s)
Liked: 37
I think the RGB LED table would be the closest.

Tyler Pruitt - Technical Liaison at SpectraCal
WiFi-Spy is offline  
post #3 of 482 Old 09-02-2014, 11:47 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
I would also be interested in a qualified opinion on the matter. I researched it for quite some time finding nothing but guesses from all parts of the community here. With the consensus approaching the White LED preset.

Again, it is startling to me that there is a new kind of display technology out there and no one actually bothered to look at it with equipment suitable for the matter and then share the outcome.

Also meter vendors dont care to update their correction matrizes.
-

As far as factory calibration goes, so far I've looked at three different W905A devices with that being the spread:

0,-2,-3;+1,0,0 (+1 almost could have been 0 as well)
0,-1,-2;+1,0,0 (+1 almost could have been 0 as well)
0,-4,-9;0,0,-1 (this one was a "on display model" and to this day I'm not sure if the access of blue was caused by color shifting or it being in store mode (should have checked at the time)


another user reported
0,-1,-1;0,0,0

All of these four measurements were taken on 55W905As with i1d3s on the White LED correction matrix on warm2. Just use them to get an idea on how "such a calibration" strays from the factory default settings on these sets.
-

Also this is the first time that I hear that RGB LED could be closest, so unless anyone is willing to argue the point, or better provide some kind of proof - I would not take it as a valid one.
--

Point being. This community has even managed to ignore an entirely new display technology, without coming up with a distinct common consensus on the matter. No one cared to look at it.

Its a "believes" thing. As long as there are a hand full of plasma displays that behave as expected, no one in here really cares. That might be a bit pronounced - but the fact is that this community hasnt even managed to agree on the mean deviation of LED backlight LCD devices and if it is 0,4 dE or more like 4 dE as some of the recalibration vendors are suggesting.

And you can be sure that no one with a half decent spectroradiometer will pitch in, because if they loose their *hush, hush* attitude - they would destroy the business of others in recalibrating the i1d3, or selling them as rebranded C6's.

So it basically comes to this. Don't expect to get an informed answer here.

Also I would VERY much like to be proven wrong.

Last edited by harlekin; 09-02-2014 at 12:41 PM.
harlekin is online now  
post #4 of 482 Old 09-02-2014, 12:50 PM
Advanced Member
 
CalWldLif's Avatar
 
Join Date: Feb 2011
Location: South Bay'Los Angeles California.
Posts: 518
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 29
Quote:
Originally Posted by harlekin View Post
I would also be interested in a qualified opinion on the matter. I researched it for quite some time finding nothing but guesses from all parts of the community here. With the consensus approaching the White LED preset.

Again, it is startling to me that there is a new kind of display technology out there and no one actually bothered to look at it with equipment suitable for the matter and then share the outcome.

Also meter vendors dont care to update their correction matrizes.
-

As far as factory calibration goes, so far I've looked at three different W905A devices with that being the spread:

0,-2,-3;+1,0,0 (+1 almost could have been 0 as well)
0,-1,-2;+1,0,0 (+1 almost could have been 0 as well)
0,-4,-9;0,0,-1 (this one was a "on display model" and to this day I'm not sure if the access of blue was caused by color shifting or it being in store mode (should have checked at the time)


another user reported
0,-1,-1;0,0,0

All of these four measurements were taken on 55W905As with i1d3s on the White LED correction matrix on warm2. Just use them to get an idea on how "such a calibration" strays from the factory default settings on these sets.
-

Also this is the first time that I hear that RGB LED could be closest, so unless anyone is willing to argue the point, or better provide some kind of proof - I would not take it as a valid one.
--

Point being. This community has even managed to ignore an entirely new display technology, without coming up with a distinct common consensus on the matter. No one cared to look at it.

Its a "believes" thing. As long as there are a hand full of plasma displays that behave as expected, no one in here really cares. That might be a bit pronounced - but the fact is that this community hasnt even managed to agree on the mean deviation of LED backlight LCD devices and if it is 0,4 dE or more like 4 dE as some of the recalibration vendors are suggesting.

And you can be sure that no one with a half decent spectroradiometer will pitch in, because if they loose their *hush, hush* attitude - they would destroy the business of others in recalibrating the i1d3, or selling them as rebranded C6's.

So it basically comes to this. Don't expect to get an informed answer here.

Also I would VERY much like to be proven wrong.
whoa there bucharoo.
You asked and it was answered.
I have never seen one let alone put a meter on one.
if you are in SoCal, I would be happy to spectro it.

As far as plasma theme, guess that is because they are hard to
calibrate well.
Generally LCD based TVs are easy because of no ABL and other
peculiarities reserved for plasmas.

I have read enough problem LCD threads to know people help.

Loving D65
CalWldLif is offline  
post #5 of 482 Old 09-02-2014, 01:10 PM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
I'm sitting in Vienna, Austria I'm afraid - but thank you very much for the offer.

Also - as always with my pronounced accusations - please no one take them personally, they are meant to first and foremost put a spotlight onto the little cracks and problems within this community to hopefully wake it up from their slumber (success ).

The point that everyone in here basically cares about their plasmas and then drops any ambition when it comes to other display technologies is another red herring, but again, not that much distant from reality, as we now for example see regarding the black correction factor of bt1886.

My point being, that after three (?) generations of this technology being on the market no one with a spectro in here even felt compelled to look at a quantum dot display and share his/her findings. Meter vendors don't feel compelled.

And if you don't bring it up, somone will utter his/her best guess - problem solved. Bliss, because most of us are watching on plasmas has at least a grain of truth to it.
harlekin is online now  
post #6 of 482 Old 09-02-2014, 01:10 PM
AVS Special Member
 
chunon's Avatar
 
Join Date: Dec 2007
Location: Wilmington Ohio
Posts: 5,390
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 343 Post(s)
Liked: 565
There is plenty of help offered here, but if you have a hostile attitude you will get nowhere.

65VT60(Calibrated by Chad B)
55ST60(Calibrated by Chunon)
Harmony 650
Darbee DVP5000
Sony BDV-F7 3dbluray/soundbar
Calman Enthusiast/I1Pro/I1D3
chunon is offline  
post #7 of 482 Old 09-02-2014, 01:19 PM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
There is, except I didn't ask for help, but rather for an informed opinion - maybe even some data to construct some sort of proof out of.

You all are free to decide - if your ego forces you to ignore me, then by all means do so - but as with so many other cases I have made in this forum so far - by doing so, you are not only harming the entire audience at large, but also the believability of calibration as something other than the pseudo science I myself believe it to be at this point in time.

Look at the bt1886 (black correction factor of the curve) and the "what happened between CIE76 and CIE2000" threads to get at what I'm pointing.

If I say "desolate" you say "we won't change for you"?

By all means you are free to do so...

But please, don't just think of me, also think of what the OP must feel. Or other people reading this topic.

edit: I will not kindle the argument any further as I believe it might harm the chances of the OP to get a meaningful answer (from which I also could benefit).

Last edited by harlekin; 09-02-2014 at 01:28 PM.
harlekin is online now  
post #8 of 482 Old 09-02-2014, 01:40 PM
Advanced Member
 
CalWldLif's Avatar
 
Join Date: Feb 2011
Location: South Bay'Los Angeles California.
Posts: 518
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 29
a quick google search and
http://www.madamenire.com/forum/166-...hread-365.html
owners thread complete with guys who calibrated them.
advertising says it is a LCD/LED 120hz.
UltraHD


A guy from Spectra cal already said RGB LED is what you should set the meter to.

Loving D65
CalWldLif is offline  
post #9 of 482 Old 09-02-2014, 03:10 PM - Thread Starter
AVS Special Member
 
nuke's Avatar
 
Join Date: Jul 2001
Location: CA, USA
Posts: 1,159
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 165 Post(s)
Liked: 84
I've been all through the X900A owner's thread - as far as I know, no one has profiled one with a spectrophotometer yet.


I have though, pulled up a bunch of white papers on quantum dot lighting systems, as employed in the X900A 4K models. It is similar to, but not exactly the same as RGB LED. We also don't know exactly the R and G quantum dots Sony picked, which white point they shot for.

(I have my suspicions - Sony being Sony, I'm guessing it ~9K and peaked for highest brightness through the LCD RGB filters, rather than best gamut and accurate white point.)

QD lit LCD has the potential to perform very well, nearly as well as the potential for OLED.

I've compared the spectra using a simple visual diffraction grating and it is definitely different than a white LED display.

I don't think a filter based colorimeter is going to nail it without a matching profile. I don't know how close RGB LED, or OLED are to the QD backlight.

I'm in Silicon Valley if anyone nearby has a spectrophotometer to profile it.

nuke
nuke is offline  
post #10 of 482 Old 09-02-2014, 03:26 PM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
(edit: @CalWldLif )

Sorry, but now Im getting angry.

I was asking for at least a source for the RGB LED correction matrix statement.

And I get -

- A link to total fluff. With people high fiving themselves in it.

- A half serious lecture on that the TV is a LCD/LED 120hz (native), which was never challenged, asked for or disputed - to divert attention

Topped of by a - statement:

A guy from Spectracal already said so.
-


Is that so? And may I ask where he did so. And on what it was based. Or at least for the source?

Because when I googled for the better part of two days the consensus on this forum came out to be "use the White LED matrix". Also the other person I'm quoting with the one calibration result also came to the same conclusion independent of myself.

I understand that at this point we both are quoting "some folks" but rest assured, that I have a vested interest in getting the correct answer because I own such a TV, and also I am not suggesting that I have given the correct answer - effectively stating that all of the suggestions I saw were anecdotal at best. While at the same time you are stating a "truth" without feeling the need to channel anything other than -

"Someone already said so".
--

Sorry, but as we are playing the "he said she said" game here and most of you at this point and also to my fault, because I have challenged aggressively, have an interest in keeping your reputation -

diverting the topic off conversation and then leisurely dropping a name - doesnt cut it.
-

Also and just so you dont get hoodwinked easily - the quantum dots technology as used by Sony uses the same blue LEDs that are used in White LED panels, but without the coating, directly fireing at those quantum dots layer where they are activating the reaction that then produces the colors green and red. If that produces a color spectrum similar to individual RGB LEDs I would like to see who says so, and what he/she has done to measure it.

My best guess at that point is that as Sony at one point called both of these technologies Triluminus, someone easily could have been misinformed, or - as usual, just said anything to fill the silence. And then back to their plasmas and bliss.

Last edited by harlekin; 09-02-2014 at 03:42 PM.
harlekin is online now  
post #11 of 482 Old 09-02-2014, 03:31 PM
Advanced Member
 
CalWldLif's Avatar
 
Join Date: Feb 2011
Location: South Bay'Los Angeles California.
Posts: 518
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 29
I have gone through the doubts about colorimeters as most DIYers have.
that is the bottom line.
it is simple and direct. no need for embellished posts to try and
force someone to spectro a TV we don't have.

I solve my need for accuracy by getting a spectro.

Loving D65
CalWldLif is offline  
post #12 of 482 Old 09-02-2014, 03:48 PM
Advanced Member
 
CalWldLif's Avatar
 
Join Date: Feb 2011
Location: South Bay'Los Angeles California.
Posts: 518
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 29
Quote:
Originally Posted by WiFi-Spy View Post
I think the RGB LED table would be the closest.
here is the spectracal guy's post.

funny how you are angry because you can't doit on your own.
When I had doubts I did not whine and blame and cuss out a forum because
it was frustrating.
No I manned up and did what I had to do.

If you would have put out a general request to forum members about
needing/wanting a profile matrix then it would have been clear and readily answered.

But no, you talk and bitch and whine.

to fricken bad no one owns one and has a spectro.

Loving D65
CalWldLif is offline  
post #13 of 482 Old 09-02-2014, 04:50 PM
AVS Special Member
 
Chad B's Avatar
 
Join Date: Oct 2002
Location: Piqua, OH
Posts: 2,158
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 127 Post(s)
Liked: 539
I've calibrated a couple of 900s and profiled my i1D3 off my Jeti 1211 on them. I am posting screenshots of the profiles. I am also posting 2 Samsung 4K LED LCD profiles, one 8550 and one 9000, for comparison.
Note: I make sure the Jeti is on axis and the room is dark, but I do not always match FOV or screen position because I usually do the 1 pass profile option in CalMAN. However, moth meters are reading near screen center. I use the i1D3 in contact mode.
You can see by the title of the profile which display it is.
Attached Images
File Type: bmp New Picture (1).bmp (279.9 KB, 142 views)
File Type: bmp New Picture (2).bmp (269.5 KB, 69 views)
File Type: bmp New Picture.bmp (260.4 KB, 65 views)
File Type: bmp New Picture (3).bmp (271.0 KB, 63 views)

ISF/THX calibrator with Jeti 1211 reference spectro
Latest reviews:Samsung UN65HU8550UN65HU9000, UN85HU8550, LG 55EC9300, LG 55EA8800 (including 3D)

Why I don't publish settings, LG OLED calibration tips, Review index and rankings, Florida tour Jan '15, Texas tour March '15

Last edited by Chad B; 09-02-2014 at 04:53 PM.
Chad B is offline  
post #14 of 482 Old 09-02-2014, 05:39 PM
AVS Special Member
 
Chad B's Avatar
 
Join Date: Oct 2002
Location: Piqua, OH
Posts: 2,158
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 127 Post(s)
Liked: 539
To make it more visual, I took a quick run of my laptop screen with the generic LED LCD profile and then each of the above 4 Jeti based profiles without moving the meter. The results are for relative comparison only; they do not indicate the true accuracy of any of the profiles.
The 8-6 Sony 900 profile is quite different from any of the others... Could be differences in FOV and uniformity. Keep in mind what I said in my previous post; this isn't a scientific comparison by any means.
Attached Thumbnails
Click image for larger version

Name:	generic led lcd.jpg
Views:	106
Size:	22.0 KB
ID:	242882   Click image for larger version

Name:	8-15 sony 900.jpg
Views:	101
Size:	21.7 KB
ID:	242890   Click image for larger version

Name:	8-6 sony 900.jpg
Views:	96
Size:	22.7 KB
ID:	242898   Click image for larger version

Name:	6-26 sam hu8550.jpg
Views:	84
Size:	22.1 KB
ID:	242906   Click image for larger version

Name:	8-6 sam hu9000.jpg
Views:	86
Size:	21.5 KB
ID:	242914  


ISF/THX calibrator with Jeti 1211 reference spectro
Latest reviews:Samsung UN65HU8550UN65HU9000, UN85HU8550, LG 55EC9300, LG 55EA8800 (including 3D)

Why I don't publish settings, LG OLED calibration tips, Review index and rankings, Florida tour Jan '15, Texas tour March '15
Chad B is offline  
post #15 of 482 Old 09-03-2014, 01:04 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Thank you very much for the profiles - I will run them against my W905A and at least report the differences in terms of bias/gain - from the White LED matrix as well as from the RGB and OLED matrizes that come with the i1d3, also to provide a relative measure people could look at to wrap their heads around it.

Just to make sure that I get the points made exactly -

You've profiled two different sets of W900s (same model?, probably and understandably not marked down the panel spec?) with the Jeti 1211 > for use with the i1d3 and you've got two results that were "quite different".
-

Also, sorry that I didn't realize that user WiFi-Spy was from Spectracal - but the phrasing "think that would be" didn't exactly fill me with confidence.

Thanks to ChadB we now have two (exceedingly?) different profiles of the W900 done with a Spectroradiometer which we can put to the test (critical viewing at least).
harlekin is online now  
post #16 of 482 Old 09-03-2014, 02:59 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Measurements done. As this is the first time I've worked with correction matrizes that werent i1d3 presets I have some fault potential here which I hope you could comment on.

- The matrix does alter the output of the chosen preset. Meaning, it depends on which preset (LCD LED, Plasma, OLED, ...) is used and would alter the output again if I'd type in the matrix at another preset.

- I've used the matrizes in combination with the White LED preset in HCFR (= LCD (LED) preset shown in Chad B's screenshots) - so they are altering the White LED preset. Could I get a verification that this has been the right approach?

- I've checked once before that the HCFR and Calman presets for the i1d3 are "about the same", but I havent double checked it today (will do so in the future).
--

Results.

The RGB-LED preset would indeed be the closest if we go by the matrizes Chad B provided. Especially - when emphasizing on his 8-15-14 example. That said, the RGB-LED preset still would be off by quite a bit.
-

Graphs.



-

Comments.

This correlates with my own personal preference over the last three weeks, where I've reset the W905A to default greyscale and watched that way. Default greyscale turns out to be closest to Chad Bs matrizes - IF I have applied them correctly (used in conjunction with the i1d3 white LED (= LCD (LED)) preset).

Last edited by harlekin; 09-03-2014 at 03:15 AM.
harlekin is online now  
post #17 of 482 Old 09-03-2014, 07:12 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Doublechecked Chad B's matrizes with Calman v4 - results are very similar. I therefore assume that my integration of the matrizes in both HCFR and Calman was correct and you can use the graphs above as guidelines.

As promised these are the changes in the whitebalance settings:

factory default:
0,0,0;0,0,0

optimized for LCD (white LED):
0,-2,-3;+1,0,0

optimized for Chad B matrix 8-15-14:
-2,-2,0;0,0,0

optimized for Chad B matrix 8-6-14:
-3,0,0;0,0,0

optimized for OLED (same as RGB LED):
0,-1,-3;0,0,0

optimized for RGB LED (same as OLED):
0,-1,-3;0,0,0

I havent got the time for critical viewing today, but I will do so in the next few days.

Last edited by harlekin; 09-03-2014 at 07:55 AM.
harlekin is online now  
post #18 of 482 Old 09-03-2014, 08:37 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
@Chad B - After some research, you have to go into a bit more detail I'm afraid.

Do you have any additional information on from which 900 series you've pulled the matrizes?

According to hdtvtest:

W905A (last years model) - uses Quantum dots
edit: X900A - also should use Quantum dots
X950B - doesent use Qantum dots anymore but rather an inhouse tech
X9005B - also doesnt use Quantum dots anymore but rather an inhouse tech
W955B - uses an normal (?) IPS display (White Leds or inhouse tech?)

Please specify for which one you provided the matrizes for.

Last edited by harlekin; 09-03-2014 at 11:21 AM.
harlekin is online now  
post #19 of 482 Old 09-03-2014, 10:54 AM
AVS Special Member
 
Chad B's Avatar
 
Join Date: Oct 2002
Location: Piqua, OH
Posts: 2,158
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 127 Post(s)
Liked: 539
The one on 8-15 was a 65X900B, and the one on 8-6 was a 65X900A.

ISF/THX calibrator with Jeti 1211 reference spectro
Latest reviews:Samsung UN65HU8550UN65HU9000, UN85HU8550, LG 55EC9300, LG 55EA8800 (including 3D)

Why I don't publish settings, LG OLED calibration tips, Review index and rankings, Florida tour Jan '15, Texas tour March '15
Chad B is offline  
post #20 of 482 Old 09-03-2014, 12:24 PM - Thread Starter
AVS Special Member
 
nuke's Avatar
 
Join Date: Jul 2001
Location: CA, USA
Posts: 1,159
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 165 Post(s)
Liked: 84
Quote:
Originally Posted by Chad B View Post
The one on 8-15 was a 65X900B, and the one on 8-6 was a 65X900A.
Thanks Chad. The 900A and 900B have different backlighting technology. The 900A is QD and Sony switched to a white LED (two phosphor LED?) for the 900B series. They branded both as "Triluminous"

I'll try the matrix for the 900A on mine and see what I get.

I have a feeling the only way to really get the 900A right is with a spectroradiometer - and I'm not sure how good of one it will take.

nuke
nuke is offline  
post #21 of 482 Old 09-03-2014, 12:37 PM
AVS Special Member
 
Chad B's Avatar
 
Join Date: Oct 2002
Location: Piqua, OH
Posts: 2,158
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 127 Post(s)
Liked: 539
The Jeti 1211 is as good as they come. Problem is, these profiles were taken with the sole intent of being used during that session only.

While it may help somewhat with i1D3s in general, a more scientific and rigid approach should be used for your intended purpose. In other words, match and standardize meter field of view and position, etc.

ISF/THX calibrator with Jeti 1211 reference spectro
Latest reviews:Samsung UN65HU8550UN65HU9000, UN85HU8550, LG 55EC9300, LG 55EA8800 (including 3D)

Why I don't publish settings, LG OLED calibration tips, Review index and rankings, Florida tour Jan '15, Texas tour March '15
Chad B is offline  
post #22 of 482 Old 09-04-2014, 04:21 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Thanks to ChadB for providing the actual model numbers - also thank you to nuke for providing a few more details about the "Sony inhouse tech" they currently use on their Trilluminus marketed TVs (I found the same "white" LEDs with new coating (to become "white", initially blue)" when trying to research them myself).

Which brings us to the following problem in the end.
-

As far as the "chain of trust" with colorimeters goes, it seems to be fully broken for LCDs at this point.

If we just look at the matrizes Chad B provided you'll see significant, non linear, deviations from the white LED, RGB LED and OLED presets that come with the i1d3.

Furthermore I acquired a recalibration report from one of the i1d3 vendors who also replace the X-Rite presets with more of their own, TV brand specific ones, and they only induce minor changes compared to the original X-Rite ones.

That said - the matrizes Chad B just provided, compared against the plain preset only reads of a X-Rite i1d3 show deviations in the range of +/-8 dE (sum of 16 dE between two readings possible) in greyscale alone.

While the TV vendors at this point do not even communicate openly which backlight technology they are using. And there are significant changes to be seen even within two generations of the same backlight technology.

(Sonys "inhouse tech" is a deviation of the white LEDs backlight with other (phosphor?) coatings.)

About the same goes for the Samsung models ChadB provided matrizes to. They perform significantly different from presets a Colorimeter (i1d3) would expect.
-

Which leaves us with the following conclusions.

Colorimeters for now are unusable to calibrate LCDs, because the inter model deviations in current LCD backlight technologies are simply all over the place and without any preset database to compare against - updated at least once a year for each manufacturers backlight generations - using a Colorimeter alone to calibrate is simply not possible.

This problem could be fixed ENTIRELY if the whole scene would adopt a "share deviation matrizes" approach - at which point i1d3s would be sufficient for the entire industry - because of their low inter model deviations and them not degrading over time.

But who at that point would buy the 10.000 USD equipment?

It would only make sense for testers, TV manufacturers and whole sale resellers of calibration equipment - because of the one side simply already having the ability to look at many TVs throughout the year and the other side would be interested to make it a for profit service. But, at the current point in time the market is skewed against this, because its more profitable to sell 10k equipment, which is serviceable every 6 months, to whales and disregard the fact that even most test sites probably are using Colorimeter presets not applicable to what they are testing.

Also "recalibrating" or "preloading" certain presets on a i1d3 for money at this point basically can be concluded to be fraudulent behavior, because inter model variation is well within the realm of introducing significant color errors, even within the product range of one TV vendor a preset would be based on.
--

Then why is the community here so happy and the problems at hand dont get any form of recognition?

Because, again, we have a bias - where most in here are happy as long as two or three TV models every five years get profiled sufficiently (correction matrizes are shared) to also being able to be calibrated with a colorimeter.

Also, because this seems to be a problem that is just starting to get recognition and that will get traction in the future, because almost noone in here cared about LCDs and their backlight technologies at all - which now should be changing as plasmas are dying.

The problem being, that it might take five years before a critical mass of the readers in here become sensitive to whats going on, because of personal dependencies/buying cycles.
--

@Chad B : I have two things to ask from you.

1. As you just shared four correction matrizes for four different LCD models for the i1d3 - which show highly significant deviations from the i1d3 presets AND also from each other - do you have any "sense" of how big the actual deviations between the LCD (LED) (= white LED) presets on the i1d3 are in your daily work > uncorrected vs corrected (with a spectro). How "big" is the difference in general? (dE (uv or 2000) or actual Gain/Bias).

2. As I myself also own a Quantum Dot display - and right now have almost no idea on what part of the spectrum my Colorimeter readings would fall, but right now I have a strong indication that with that much uncertainty being introduced into the process by backlight alone, I can't use it at all - would you also have and be willing to share a correction matrix of a (55)W900 (W905a) (also a quantum dot display), for a i1d3, because I wasn't tuned in enough to buy one of those 3-4 TVs where those are publicly available. Also I am a student, living in the heart of Europe, so I would, probably for some time, not be able to afford the airfare for bringing you around for a calibration.

As it stands right now we only have one correction matrix for one Quantum dot display publicly available - and that is not very much. At all. If possible even publicly so that we could dissect it for any kind of "trend" within at least two quantum dot displays.
-

I will wait a few days for reactions from the forum and then write an introduction to this problem and pin it on top of the calibration tutorial I have written - because with deviations as huge as you can find within the released four matrizes in this thread - I have to warn others not to think about or buy calibration equipment below the price point of 10k USD - when calibrating LCDs.

Last edited by harlekin; 09-04-2014 at 05:35 AM.
harlekin is online now  
post #23 of 482 Old 09-04-2014, 04:49 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Also - if Chad B matrix 8-6-14 turns out to be "representative" (again, we are worlds away from scientific right now) for Quantum dots displays - the statement that the "RGB LED" preset is "closest" - still is more problematic than anything (the white LED preset even more so).

Look at the Greyscale (Gain/Bias) settings of the W905A "optimized" for each of them individually.

optimized for Chad B matrix 8-6-14:
-3,0,0;0,0,0

optimized for RGB LED (same as OLED):
0,-1,-3;0,0,0

Last edited by harlekin; 09-04-2014 at 05:02 AM.
harlekin is online now  
post #24 of 482 Old 09-04-2014, 07:00 AM
AVS Special Member
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,103
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 496 Post(s)
Liked: 466
Chad,

If you measure one of these again it would be very useful to provide a spectral sample set (ccss). This would remove the uncertainty with shared matrix/inter-probe variability and be directly useable by other D3 owners on these displays. You can create a ccss from your profiling measurements using HCFR, ArgyllCMS, or dispcalGUI.
zoyd is online now  
post #25 of 482 Old 09-04-2014, 07:16 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
@zoyd - problem being that you can see _significant_ variation from the usual (Xrite) correction profiles also in the 900B series Chad B profiled - which is not a quantum dot display, but a "white" LED one, with "different coating".

So you are very fast closing in on a "need to look at each background lighting tech each year" approach that would be needed.

Thats why I was that interested in the "amount of variation" Chad B sees in calibrating LCD displays vs. the normal (Xrite) LCD (LED) preset.

Also, with Sony now not licensing the Quantum dots technology any more - the chance of Chad B coming by such a display will be getting slimmer over time, BUT at the same time the problem is not becoming smaller - as other LCD display technologies seem to show variations that are all over the place.

Or to put it differently - you could change the thread title from "quantum dots displays" to "all LCD displays nowadays". At least that is my presumption right now.
harlekin is online now  
post #26 of 482 Old 09-04-2014, 08:35 AM
AVS Special Member
 
steve1971's Avatar
 
Join Date: Jul 2008
Location: Saint Paul, Minnesota.
Posts: 2,074
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 232 Post(s)
Liked: 306
The only Sony HDTV's that used the Quantum Dot tech with Triluminous was the 4K X900A and the W900A. The new tv's that followed didnt use Quantum Dot because Sony decided to drop it for their own in house Tech which they paired up with Triluminous. Their reasoning for it was to save money. My W900A is Pro Calibrated and the PQ is beautiful. Just my 2 cents
steve1971 is offline  
post #27 of 482 Old 09-04-2014, 10:31 AM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
They did not "pair it up" with Triluminous - they just marketed it under this brand. Triluminous at this point stays for nothing except probably for "wide color gamut" but that said - even the W65* series also had a wide color gamut option (with normal white LEDs?) but wasnt labled Triluminous.

Also please do not pitch in with "pro calibration picture quality is beautiful" when giving no indication if a Spectroradiometer was used, and also not posting the white balance settings. Not because we want to copy them, but in this thread alone I've already posted the greyscale settings for four different W905As, calibrated with a i1d3, according to the LCD (LED) preset - and three of them dont show much variation - so a fourth one at this point might at least "tell a story". If you could deliver those two factoids it would be much appreciated. By me and probably the OP.
---

Also I have to add an observation regarding Chad B's 8-6-14 matrix.

When I calibrate my W905A according to it - greyscale, and prim/sec colors show dE's below 1.1/2.3 - BUT

delta xy (without Luminence) of green is off by 0,02 which is QUITE a bit. The color triangle at that point is obviously smooched. To the extend where CIE76(uv) shows a dE of 7,6 on the color green and only the fudge factor of CIE94/2000 allows the error to creep under dE 3.

This is the FIRST time I have seen such a color triangle error associated with the W905A in all reviews I have read.
--

Which leads to the following possible outcomes.

1. Color Matrix is not applicable
a. because Chad B doesnt know how to produce such a matrix (but uses it in calibration every day) - probably unlikely (?)
b. because of inter device variation (unlikely to that extend)
c. because the W905A is a slightly different model than the X900A - which would point to even more "inter technology" variation in backlight technology alone (both devices use quantum dots, licensed and implemented by the same manufacturer at about the same time.)
d. My personal i1d3 measures something totally different than the i1d3 the matrix was created for (unlikely (see low inter device variance in the W905A measurements above (two different persons with two different i1d3s) and also the already confirmed low inter device variance between i1d3s in general) ).

2. Each and every review site out there doesnt have the equipment to correctly measure LCD TVs. (Also manufacturers dont, when deciding which panels go into their top of the line models.)
--
@nuke : At this point - please post a full calibration report of your X900A, calibrated against probably both matrizes Chad B provided. (edit: And against the default (white) LED LCD preset.) If we both see the same "dip" at green (100% saturation) with Chad B's 8-6-14 matrix, something here is VERY wrong.
-

If other members would want to pitch in with "probability assessments", this would be the time..

Last edited by harlekin; 09-04-2014 at 11:32 AM.
harlekin is online now  
post #28 of 482 Old 09-04-2014, 02:18 PM - Thread Starter
AVS Special Member
 
nuke's Avatar
 
Join Date: Jul 2001
Location: CA, USA
Posts: 1,159
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 165 Post(s)
Liked: 84
I have a:

Sony XBR65X900A, 4k, UHD, quantum dot back lit LCD.

Brand new I1D3 colorimeter, which seems to be working properly.

The latest build of Color HCFR.

The AVS Forum test disk burned to blue-ray.

PS3 as the blue ray player. (set to recommended settings, but experimented with full and limited HDMI, different color spaces, extended Y, etc to no unexpected effect)

So I experimented with the correction matrix from Chad for both Sony 900's and I can say that neither one is conclusively doing anything positive with my I1D3. By visual comparison with both a calibrated LCD display and a photographic lightbox with a grey target and D65 lamps on, using any of the corrections results in clearly shifted grey scales.

In fact, the best results I have gotten so far have been with an identity matrix (no correction at all) and the "non-refresh display" selection in HCFR software. In other words, just pretty much using the sensor data with no spectral corrections at all.

Two curious things in my experimenting so far:

1. 100% saturated red and blue extend to the corners (and red a bit beyond) of the 709 gamut triangle, 100% green comes up far short.

2. The HCFR/I1D3 says red is way too high, but dialing in what it says is D65 results in a really funky green/blue color that is clearly not right.

3. There is an uncorrectable "bump" at 50-60% gray steps where the bias and gain controls on the Sony white balance have little effect that the meter says is way off (dE of like 6) but is visually pretty close to the known-good targets I'm comparing to.

4. The I1D3/HCFR are more than happy calibrating displays of the types "it knows about". I set about experimenting with several other people's TV's and it happily dialed in good white balance on CCFL and white LED lit displays. I did this to prove it wasn't the I1D3, the software or me doing something completely boneheaded.

5. Picking any gray field from 0-100% and running the backlight up and down results in nearly zero color balance shift.

My conclusion thus far:

The I1D3 is likely being fooled by the QD backlighting and is reporting less green and more red than is really present.

Sony's setup for Cinema mode and warm2 are reasonably good out of the box. Best overall picture comes from slight changes to the user-available white balance controls to visually approach a known D65 reference and to strive for uniformity of greyscale tracking.

FWIW, the 65X900A is definitely capable of more color performance than your average LCD display.

I would be real curious how it performs with a 4k UHD test pattern. I have the FMPX10 and the display enters some other, factory pre-set mode when the FMPX10 is playing content.

Not sure how to construct a set of data that the FMPX10 would recognize, display and cause the TV to enter whatever secret mode it does go to.

Bottom line: I don't think it can be calibrated at all with the I1D3, at least not without a proper spectral profile specific to this display.

nuke
nuke is offline  
post #29 of 482 Old 09-04-2014, 03:07 PM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Your statement is hard to dissect. Especially as you arent clearly stating for which preset (+ matrix?) your observations are valid. (Also, how exact are those D65 lamps.. )

That makes it unclear f.e. if the prim/sec colors are looking like this because of a certain matrix/preset chosen (runaway value), or because of screen characteristics.

You are mentioning that 100% green comes "far short" - is this a statement with Chad Bs correction matrix (one of the two and which one), without Chad Bs corrections matrix, both...?
-

Also, your conclusion that the i1d3 is "fooled by the technology" is a bit strange, when at the same time you are stating that "the best results are reached with no correction tables at all". Even if that should be the case, you should only count it as "a deviation from existing presets" and not something that is "special, or magical". Apply the correct correction table and a Colorimeter should deliver correct results. If the correct correction turns out to be "no correction at all" it makes me more worrisome than anything else..
-

I would ask you to give me something I can work with so we can compare displays (both quantum dots, same manufacturer, almost same manufacturing date).

Please provide "baselines" by -

calibrating your greyscale for each of the presets I've listed above (optimized for LCD (white LED), optimized for Chad B matrix 8-15-14, optimized for Chad B matrix 8-6-14, optimized for OLED (same as RGB LED), optimized for RGB LED (same as OLED))and then write down the individual greyscale settings -

I will do the comparison writeup afterwards (looking at relative movements).

edit: Just noticed that you use a PS3 to switch the patterns - I now understand that this would take you much more time...
--

Also with the 100% green coming short statement - is this true for all presets/correction matrizes, or just for some - and do you notice any significant change dependent on the preset/correction matrix used.
--

Also, notice that with the correction matrizes Chad B provided - there is no such thing as a "normal white LED lit LCD" - you can see very significant deviations in all models shown. I don't want this talking point to be buried.

Last edited by harlekin; 09-04-2014 at 03:55 PM.
harlekin is online now  
post #30 of 482 Old 09-04-2014, 03:28 PM
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 280
Mentioned: 50 Post(s)
Tagged: 0 Thread(s)
Quoted: 92 Post(s)
Liked: 20
Just had to make a major edit in my previous post.

@nuke : If you choose to, please provide Gain/Bias "settings" optimized according to the different presets or presets/matrix combinations, as the factory default of 0,0,0;0,0,0 is also a "marker" I would want to look at (probably to no avail - but its a loose end I also would want to ponder about).
-

edit:

Also, the use of Ultra HD calibration material is useless.

I've used the 720p version of color patterns several times, just because it was the smallest package to download - and there are NO measurable differences to the same material in 1080p. (I even could leave all "smooth color gradation" and "reality creation" TV settings gizsmos on (I dont.) and it doesnt change a thing).

Also Ultra HD calibration material is especially useless - as there is no "universally decided upon" Ultra HD color standard yet. AND the color space is very likely to change beyond the rec709 one with the standards creation in the following years. At which point most if not all current 4K TVs will be incompatible with the then set Ultra HD color standard.

Last edited by harlekin; 09-04-2014 at 03:48 PM.
harlekin is online now  
Reply Display Calibration
Gear in this thread



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