Hi.

I made my first calibration using Lightspace and it didn't go very well. Well, it was the first with the new software.

The main problem, in my opinion, is the gamma (graphic below), the average gamma is 1.795 and the objective 2.3. The display is a VT50 plasma and the ABL is very agressive, also:



- The black level is the same before and after activation of the LUT in eeColor. I used PLUGE from Spears&Munsil, HDBasics and Ted's calibration Disc (in this, it depends the pattern used because the ABL).

- The content should be very washed but the black looks correct and the detail is good. The only content looking washed is SD and Daredevil (in some scenes). This makes me thing the real gamma is different that the post-calibration report.

- Probably it there was a mess with max luminance, I read the value in options before the calibration and it was around 110 nits after LUT is 92.1. In this TV is always in this way, the max luminance post-calibration is less than before-calibration.

In order to minimize this problem, I created a CSV with the 101 grayscale points (0-255) in a spreadsheet and tested it with display characterization and virtual probe and it looks correct.

The next step is create the parametric gamma to try to correct the gamma, I read the guide on the web but I want to have it so clear as possible:

- In the bcs file, I will have to write the differences between the objective gamma and the real, is it correct?

- The input points of the gamma they have to be normalized in 0-1? The output on each point is the difference?

- The differences to each point are among the values (by example: 50.2 - 50) or the factor of gamma (ex: 2.35 - 2.3).

- The each point is: objective value - measured value with sign, no?

- When I will make the cube profile again, do I have to load the 101 grayscale points?

My intention is to make a program using the gamma function, measured values, max and min creates the point for the bcs file.

Thanks in advance.

Hi fonsocm
The first comment is the Gamma Graph shows clipping in the whites.
Did you check using the CalImages as per the User Guides?
And what patch size did you use for the display?
Plasmas need small patch sizes to avoid ABL kicking in when profiling.

Before getting into Parametric Gamma you need to verify the basics are correct.
With Plasmas there is actually no benefit using Parametric Gamma, as ABL will always defeat it)

Hope that helps.

Steve

Well, I remember used the BarsandBlack and it was correct. I don't remember if I used the contrast.

The patch size was medium from a Lumagen Mini.

Also, the pregamma didn't show problems, the reference was Rec.709:



And the RGB separation was very bad:



If the problem is clipping in whites, the only solution is going to down the contrast, no? I say because the problem in this case could be a dimmed image.

Thanks.

'Medium' patch size is likely to be too large for a Plasma...
So that is possibly the issue, as the RGB Separation graphs also suggests.

Steve

Well, I go again.

Thanks for the answers and the fast reply.

I made new tests using the small windows to characterize the display and the results are much, much better, principally visually. I used a cube of 17 and tested it with medium windows (to be able to compare with the other results) and rec709:





The main problem/doubt is: I think there is not more space to get better results. I made tests using a 10 cube (it takes 1h 15m, 17 takes 3h 30m) and the results are similar:





It looks the ABL acts as limiting the possible values in some IREs, IMHO. In this context, I don't know if the use of parametric gamma is really usable, although the problem would serve me to practice Scala.

The next week, I should receive my i1 Pro certificate and I could get more reliable results.

Good progress!

Hi.

I received my i1 Pro and I profiled the i1 Display. One thing I drew my attention is the recommendation to use the generic profile for the spectro and the colorimeter to make the probe matching, I thought is was better to use the profile for the display always, in order to get the better measures for that kind of display (LCD, Plasma, OLED, ...).

I made a new characterization using the probe matching with a 17 cube, also I modify the brightness and contrast, in this display when you modify one of them, usually, you have to change the other to fit the patterns. I turned down the contrast five points and increased the brightness one point. Also, I wanted to test if I turned down the contrast I could limit the ABL effect, fulfilling the patterns.

The results are with medium window, rec709 and 2.2:





They are very similar to previous results. It is a new unit profiled with i1 Pro, the unit must be "good".

I made another test, I profiled the display using the before LUT and small windows:





The result was a surprise for me, should not the result very better than medium? The influence of the ABL should be less and the characterization was made with this size of window. Perhaps, the new brightness and contrast although visually similar, they are not so good like the old values.

Also, I am using LUTs for Bt1886 and rec709 2.3. Rec709 2.2 looks very clear for me, I used a calibration to Bt1886 and I got used to a darkness image. The results are similar, the differences are in the DifGama and RGB Balance, they are not so good like Rec709 2.2.

No, when using probe matching you should use no pre-set matrix, as per the instructions.


You should pre-set the Brightness and Contrast as per the user guides - using the CalImages.
Setting the contrast lower ill have very little (if any) effect on ABL, unless you set it very low.


The profiles show ABL issues, and possibly issues with display image retention, or a possible sync issue with the patch display path. Are you using Extra Delay Time? You may need to. You may also need a smaller patch size.



Without seeing the Gamut of the original display it is not possible to comment on the difference in the RGB Separation.
If the original display gamut has issues it is entirely possible for the 'calibrated' display to show a poorer 'RGB Separation'.
See: http://www.lightillusion.com/profiling_manual.html#rgb_separation

Steve

Well, I am more used to other pluge pattern and in my tv the CalImages are not very clear. I explain it: In Spears&Munsil using the pluge to fix the brightness, I can a value to see the pattern exactly as instructed; however, with the brightness pattern of CalImages, I couldn't get the same. I'll try again, it is probably get used.


I am using Lumagen Mini with small windows, I can try to use Resolve, it lets more sizes for the window. And I using an extra delay time of 0.25 to minimize the retention and an integration time of 1.5.


If I don't remember bad, the gamut is good with red and green and little hue error with blue.

Thanks for the answers and the patience.

Pluge only works with TV Legal displays.
Many display today use used full data range.
And Pluge only works for black, not white levels.


That probably means the display is just inherently unstable...
You may want to run stability test to verify this (manual measurements of the same patch over time, and compare the readings).


Again, without seeing the Gamut graphs we cannot tell anything...

Steve

Hi.

I was yesterday reviewing the brightness and the contrast using the CalImages and the correct values (trusting in my vision, I should have left half retina) are different that using calibration discs: contrast = 40 and brightness = 0; measured: min = 0.013 (using other calibration discs around 0.025), max = 120.35 and CR = 8964.9:1 (amazing value comparing with calibration discs around 4000-3500).

In the first I used the BarsandBlack to make sure the data levels were correct, also I was using my laptop: Samsung with Intel card video. No problem with Xnview, until I used the Ted's Calibration Disk BarsandBlack mkv using Windows player, the image shows without the gray border in the white square and black square disappeared. Well, I go to the card video panel control and change limited range to full range and the video shows well, it looks like the image in first case in data levels guide (http://www.lightillusion.com/img_li/barsandblack_cal.png), only the black square distinguish better the background than the web. The curious thing is when using the image is look the same independently of this option.

In my humble opinion, when the limited option is selected the card sends all in legal range and the TV shows in legal range and when the full option is selected the card sends all in full range and TV shows all in full range.

After, I used the brightness and contrast images to set them; first, I fix the brightness and then contrast. I made a recheck and I had to readjust the brightness, no being necessary more modifications.

I went to Lightspace and I used the patterns for make a Primary&Secondary profile, almost sure there is ABL but I don't think is critical, the gamut:





Also, in case my interpretation was incorrect I ran the same profile with limited option:





Moreover, using the profile generated with the full option, I created a ITU-R BT.1886 LUT and I ran a Primary&Secondary profile using this LUT and the results look very, very, very better. I have to recheck the results.

Also, IMHO, the best option is to use the full option, the problem is when I use the Lumagen Mini like pattern generator is sending all in legal range. I trying to use Resolve, I am able to use it like pattern generator but I am not able to show the pattern in the TV.

PD: I am sorry for too much text but I would want to be so clear I could.

I am sorry, but I really cannot follow what you are saying that well...

But, it seems you are not sure if you are using a TV Legal, or a Data Range workflow.
You must understand what is correct from your image workflow before you can do anything with calibration.
For example, you should really calibrate using the same image path as you will use for viewing final images/video after calibration.

Separately, looking at your Gamut Graphs, the display will never calibrate that accurately to Rec709/BT1886 as the gaumt is already too low, especially in Blue (and Red to a lesser extent), and has a white point that is far too Yellow.

This will greatly restrict you ability to calibrate the display accurately.

Steve

I think I understand, if I go to use Bluray I should use all the time tv legal. I thought I could calibrate using data range when measuring the patches and after I use tv legal without problem to see all type of contents.

I know one more thing. I use an Oppo Bluray and NVIDIA Shield TV, I'll use tv legal.


I'll try with normal mode, it lets to change gamma and two point rgb balance. But I am thinking the tv does not allow more.

Comment, I was watching the tv using the LUT generated yesterday using one of the profiles and it could be fortune, random or placebo effect. But the image look stunning, much better than using Lumagen mini and the other calibration program: the whites look more whites, very big detail and more luminosity.

Thanks for all the help.

PD: I do not want to be disrespectful, I apologize in advance. But do you think about impart courses?

How were you using the LUT generated?

And what do you mean by 'impart courses'?
If you mean training, we offer that on as a service at customer locations.
We do not run general training sessions.

Steve