AVS Forum banner

1061 - 1080 of 1251 Posts

·
Registered
Joined
·
18,309 Posts
As one of the Klein owners who very insistently asked Steve to add some sort of intelligent integration feature to the K-10A in LS / CS, I was really excited to test the latest beta of CS, as it brought the new Klein API and the highly anticipated intelligent integration.

After some hour of testing, I felt like I finally have a Klein! Fast and furiously accurate! I won't annoy you with what I did and how. I will jump directly to what are my settings for accuracy and speed and what I think is the best compromise between them.

Accuracy: Max integration 5 seconds, Intelligent integration 1 NIT, Average low light OFF. It needs 9' 36" to read 252 GS points.

Speed: Max integration 0.125 seconds, Intelligent integration OFF, Average low light OFF. It needs 1' 23" to read 252 GS points.

The Genius settings (which is as accurate as the "Accuracy" one but a lot faster): Max integration 1.250 seconds, Intelligent integration OFF, Average low light OFF. It needs 2' 35" to read 252 GS points.

HERE you can see the difference on 252 GS points between the "Accuracy" and the "Genius" settings.

So thanks @Light Illusion (and Klein) for anticipating Christmas ;)
252 GS points?!?!?!
 

·
Registered
Joined
·
1,012 Posts
Oh yes DeWayne! That's the new frontier of GS calibration and a secret also :p
 
  • Like
Reactions: jrref

·
Registered
Joined
·
7,111 Posts
Every time you say “at a high level” I owe you a slap on the wrist. You are retired! Let that jargon go.
LOL, You can't take the corporate mentality out of me.
 

·
Registered
Joined
·
7,111 Posts

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,067 (Edited)
Miki, Thanks for the testing. At a high level can you describe how intelligent integration works? Is your Klein now a lot faster than your i1D3? :LOL:
Using the 'Intelligent Integration' feature, you set a nits value below which the meter will perform a second, third, fourth, etc. reading, increasing integration times up to the maximum integration time of 6 seconds for i1Display PRO. For Klein K-10A 4 seconds is the default, the max is a lot of higher.

The readings will stop as soon as a stable and valid reading is achieved.

When no valid and stable reading is achieved, a 'zero' value will be returned, enabling the software a replacement value when generating a calibration LUT intelligently.

That function was available from September 2017 in LightSpace for i1Display PRO users only.

Now it's available for Klein users also.

You can say that Klein had to update their API to make this possible, so its a teamwork with ColourSpace Team with Klein.

It was also a request from advanced ColourSpace users like Miki, Leon, Enrico, Fabio etc. ...to improve even further the calibrations and reduce the total profiling time.

So you can say that at the same time, you can measure twice the number of patches, or reduce to half the 5000 regular patchsets for profiling.

High-Level language required when a color-scientist talk with another color-scientist, or a developer to another developer.

There no need to talk/ask for 'high-level' info every day, as the only information you need is some additional info but simplified for all users to understand.

High-Level info is something like that:


While I posted, the average number of users hasn't understood that post...., so it was pointless for the average users to have such 'high-level' info.

About ColourSpace/LightSpace, any meter function is explained to each meter page, so all pages are updated to cover any change:


 

·
Registered
Joined
·
7,111 Posts
Using the 'Intelligent Integration' feature, you set a nits value below which the meter will perform a second, third, fourth, etc. reading, increasing integration times up to the maximum integration time of 6 seconds for i1Display PRO or 4 seconds for Klein K-10A.

The readings will stop as soon as a stable and valid reading is achieved.

When no valid and stable reading is achieved, a 'zero' value will be returned, enabling the software a replacement value when generating a calibration LUT intelligently.

That function was available from September 2017 in LightSpace for i1Display PRO users only.

Now it's available for Klein users also.

You can say that Klein had to update their API to make this possible, so its a teamwork with ColourSpace Team with Klein.

It was also a request from advanced ColourSpace users like Miki, Leon, Enrico, Fabio etc. ...to improve even further the calibrations and reduce the total profiling time.

So you can say that at the same time, you can measure twice the number of patches, or reduce to half the 5000 regular patchsets for profiling.

High-Level language required when a color-scientist talk with another color-scientist, or a developer to another developer.

There no need to talk/ask for 'high-level' info every day, as the only information you need is some additional info but simplified for all users to understand.

High-Level info is something like that:


While I posted, the average number of users hasn't understood that post...., so it was pointless for the average users to have such 'high-level' info.

About ColourSpace/LightSpace, any meter function is explained to each meter page, so all pages are updated to cover any change:


Ted thanks for the Detailed explanation. The reason why i said High-level is it's nice to know all the details but i just wanted to understand how the feature worked so i know how to use it. Faster for me is better so i'm glad this was implemented. I will help me a lot. Good work!
 

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,069
Thanks Steve & Ted!

It's interesting that you were able to save the file with colons in it, for me it did not work. It doesn't work trying to rename an existing file into one that has a colon in its name either, Windows Explorer doesn't let the typed colon be added to the new name.

It's saved, but after closing ColourSpace and restarting it, the measurement files with invalid characters were no longer available.

They were not available to the 'Spaces' folder, and I had some entries with 0 bytes.

After some hours from the time you've reported the problem, the problem was resolved with the new Beta!

I managed to save a 17-sized cube after profiling the C8 this time, but I have a question.

I used a drift setting of 15, and I got this drift graph, with an irregularity near 0.4:

View attachment 3044796

Is there a way to know which measured points were affected, in order to retake those measurements?

Would you advise creating a LUT with "Drift Compensation" enabled for it in the "LUT Tools" window, or not?
Drift 15 is not required for any reason to use, and it will not provide any help with any display, its very low value.

Ideally, use 50 drift as the lowest.

The drift plots chart shows the plotting of chromaticity changes (RGB lines) and luminance in nits (+-3 nits axis) during the profiling run.

If the profile took 100 minutes to complete, then the 0.4 to the chart will show the 40th-minute drift frame.

When you see such a drop, you had a bad read, so the pattern generator you used hasn't generated the patch on time, so meter read something else.

Which pattern generator have you used?

For PGenerator users, we recommend always to use the Direct LAN connectivity method to avoid any network delay, which may add a similar issue like that.


However, you can fix it manually.

Export the profile file and open it with a notepad of windows:

The profile name of windows can have a different 'profile name' when you open import and view it in ColourSpace Space Manager.

For example, the file in windows is named: ''CX FILMMAKER MODE 072920''.



If you change that name to 'NEW' and import it to your ColourSpace, it will say that this profile is already available as ColourSpace will see the name from:

<builder_color_space name="CX FILMMAKER MODE 072920" version="2">

So when you want to edit manually a file, change the name of the ''<builder_color_space name='' to something else, like:

<builder_color_space name="CX FILMMAKER MODE 072920 Drift Fixed" version="2">

At the end of the file, you will see collected all the drift plots frames and the number of the total frames.



You can look the Y (luminance) while you are scrolling down these frames until to locate the problematic frame.

So when you find that frame, copy the XYZ from the previous frame and paste them to the problematic frame)

Then save and import it to your ColourSpace.
 

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,070
Ted thanks for the Detailed explanation. The reason why i said High-level is it's nice to know all the details but i just wanted to understand how the feature worked so i know how to use it. Faster for me is better so i'm glad this was implemented. I will help me a lot. Good work!
For Klein, the 'factory default' is 4 seconds.

The i1Display PRO has no factory default, but the ''Max'' is 6 seconds.

The Klein ''Max'' is what you set via the Max slider.

The i1Display PRO has no ''Max'' slider, so this is unique to Klein at the moment.

Max Integration Time, sets the maximum time up to which the Klein K10-A/K-80 will integrate all readings over.

Intelligent Integration sets a threshold nits value, below which the Klein probes will start with the set Max Integration Time, and then increase the integration time in repetitive readings until a stable value is read, up to the absolute maximum integration time.
 

·
Registered
Joined
·
600 Posts
I had the chance to test the latest ColourSpace beta version because, as a K10-A owner, I was curious about their implementation of the new Klein API. The team at @Light Illusion and Klein did a great job and the settings @Anker.Mikiprovided working perfectly.

The improvement in speed and accuracy is outstanding. Compared to the old Klein API, I see significant improvement in Gamma tracking, such as more accuracy at low luma points. Also I created a new optimized patchset which contains 5400 patches and 50 grayscale points. The results of my profile are ridiculous low the 3 points over 1 have an max dE value of 1.16 credits to this awesome excel sheet goes to @Anker.miki




My optimized handmade patchset looks like this

 

·
Registered
Joined
·
440 Posts
I wanted to come back and say thanks to everyone who is willing to help me. I still don't think I'm getting perfection, but I feel like I'm getting a more optimized picture, slowly. I forgot about having to finesse the controls on the TV. Do Not make hard cuts in RGB. I don't feel like it's perfect still, yet! but I feel like I actually got something this time that is pleasing from my testing so far. And that is progress. Thanks for replying to my questions again
 

·
Registered
Joined
·
414 Posts
So with this new klein api implementation,the settings for lg oleds are:

1. Stabilisation 0.350 sec
2. Intelligent int ”unticked”
3. Average low light ”unticked”
4. Integration 1.250 sec
5. Extra delay time 0 sec
 

·
Registered
Joined
·
440 Posts
I finally did it. I was able to get great results on custom 1 and 2. I know it's silly, but I am pretty happy right now. It has fixed trouble spots on certain testing scenes that I could never fully get corrected.
 

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,075
The improvement in speed and accuracy is outstanding. Compared to the old Klein API, I see significant improvement in Gamma tracking, such as more accuracy at low luma points. Also I created a new optimized patchset which contains 5400 patches and 50 grayscale points. The results of my profile are ridiculous low the 3 points over 1 have an max dE value of 1.16 credits to this awesome excel sheet goes to @Anker.miki

My optimized handmade patchset looks like this
Hi Fabio,

How much time did it take with the new API of Klein to finish the profiling measurements?
 

·
Registered
Joined
·
600 Posts
Hi Fabio,

How much time did it take with the new API of Klein to finish the profiling measurements?
It just took 2h 4 minutes, including 30 minutes of pre-roll, to stabilize the panel so much faster then before.

The 1000p Verification Set needed 16 minutes.
 
  • Like
Reactions: ConnecTEDDD

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,077
So with this new klein api implementation,the settings for lg oleds are:

1. Stabilisation 0.350 sec
2. Intelligent int ”unticked”
3. Average low light ”unticked”
4. Integration 1.250 sec
5. Extra delay time 0 sec
Hello,

Using ''extra delay 0.5 sec'', the settings you have posted will be OK with the current beta.

I believe we will have another optimization with newer beta.
 

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,078
I finally did it. I was able to get great results on custom 1 and 2. I know it's silly, but I am pretty happy right now. It has fixed trouble spots on certain testing scenes that I could never fully get corrected.
Great.

Your results are now better because you have improved your knowledge about adjusting your particular display controls.

You have improved your knowledge about manual calibration by understanding all that useful info you get from multiple ColourSpace charts about how TV controls/settings affect panel performance.

So now you know more stuff about calibration generally ;)
 

·
Registered
Joined
·
898 Posts
Silly question probably but now that there is a Probe Matching Active box in Graph Options I just want to make sure-- In order to use Manual Measure only is probe matching "On" for the Connected Probe when you have selected the Active Probe and Reference Probe in Probe Matching while using the Connected Probe (presumably after the Measure All steps have been made for each probe)?
 

·
Registered
Joined
·
9,575 Posts
Discussion Starter #1,080
Silly question probably but now that there is a Probe Matching Active box in Graph Options I just want to make sure-- In order to use Manual Measure only is probe matching "On" for the Connected Probe when you have selected the Active Probe and Reference Probe in Probe Matching while using the Connected Probe (presumably after the Measure All steps have been made for each probe)?
After selecting the meter profile data files @ Active and Reference Probe for your colorimeter, tick the Probe matching.

 
1061 - 1080 of 1251 Posts
Top