AVS Forum banner

221 - 240 of 3759 Posts

·
Registered
Joined
·
4,291 Posts
Hi,



sorry for asking newbie questions continuously. I now managed to run a full 1D Dolby Vision Autocal on my C8. It seems, pattern insertion interferes with prolonged measurements my i1Display performs in the near black range. I turned pattern insertion off and voilá, Autocal did its work. I know, the panel drifts due to excessive heating so I‘d better find a way with pattern insertion turned on, but did not succeed so far.

Another question is about the correct settings for the LG HDR workflow. As far as I understand, one would calibrate DV against Gamma 2.2 and then let the TV do the HDR work based on the file CalMan generates and uploads, which itself represents the specific capabilities of the set under calibration. So, are my Workflow settings correct, see attached screenshot? What about the target colorspace? What would change, if I would do HDR10?

BTW, I had to replace each , with . inside the text file before uploading to the TV in order to get valid reading, otherwhise the set would turn pink in Dolby Vision mode. Maybe thats because I‘m resident in Germany/Europe.



Thanks, Robert


The workflow is configured correctly if you change any of the calibration targets you will cause major issues and not get an accurate calibration.

We are working with LG on the pattern insertion flickering problem in Dolby Vision mode.

Calibrating HDR10 on the 2018 LGs Calibrates both HDR10 and HLG at the same time. This is because we are calibrating the panel Gamma and the only difference between those formats is the tone map.
 

·
Registered
Joined
·
98 Posts
Thanks Tyler,

very helpful post.

Robert
 

·
Registered
Joined
·
588 Posts
I measure 9000+ patches all the time. Yes it takes multiple hours (6 in my last test last night with 11,000 patches) compared to he 30 min quickie but the end result is far superior and worthy of the name “3DLUT”.... if done right. I pointed out from the very beginning that you should not expect better than human results with a 3DLUT without doing the time. It’s not possible.





Measure your C8 post lightning LUT with all possible patches available with the colorchecker and see what your max dE is. No matter what I try, I can’t get the max to go below 4.5. In the world of 3DLUTs that is completely unacceptable. Every TV I have here attached to a LUT with 9,000+ patch measurments have a max dE of 1.6 or less...depending on the TV for the same patches I’m asking you to measure. That result is far superior to human results with these specific TVs.





How many if, then else statements do you think it is going to take to get that done?



That is still being done but Portrait Displays is not Verizon :)


11,000 Patches with CalMAN or Lightspace?
 

·
Registered
Joined
·
13,713 Posts
I measure 9000+ patches all the time. Yes it takes multiple hours (6 in my last test last night with 11,000 patches) compared to he 30 min quickie but the end result is far superior and worthy of the name “3DLUT”.... if done right. I pointed out from the very beginning that you should not expect better than human results with a 3DLUT without doing the time. It’s not possible.
I understand why 6 hour calibration sessions are a non-starter for professional calibrators, but I'm not understanding why this is considered impractical for enthusiast DIY calibrations.

1/ Set everything up and initiate Autocal.

2/ Go to bed.

3/ Wake up the next mosting to find your calibration complete.

I understand that if the SW crashes before calbration is complete, this will not work, but that 's because of faulty SW, not impracticality.

A side benefit of overnight calibration sessions is that inter-measurement blanking intervals can be almost as long as you want them, so it seemscas though panel heating issues could be completely avoided.

So what am I missibg here? Now that we've moved from hands-on calibration to Autocal, why wouldn't an enthusiast want to perform the longest, most-precise calibration possible (while he/she is off doing something else):confused:
 

·
Registered
Joined
·
9,142 Posts
^^ I agree. If I'm leaving it overnight or while I'm out at work, I don't mind if it's 11,000 or 22,000 patches. Give me the most-precise calibration possible! :)
 

·
Registered
Joined
·
13,713 Posts
^^^Right so i understand you have to manage the heat but reading 10,000 patches and managing heat is going to take a long time to run :)
This issue of 'managing heat' seems to be a big deal with getting an accurate calibration of these WOLEDs, but I have a question about that:

Don't you want the panel calibrated at the estimated heat it be at when displaying content? If you calibate the panel 'cool' won't that mean it is farther off than it needs to be when displaying content (which will inevitably heat it up)?

I understand that each piece of content will be different and will result in a different heat profile, but wouldn't it be better to calibrate to whatever the 'average' heat profile for SDR and (seperately) HDR would be rather than an artificial 'cool' profile which is guaranteed to be as far as possible from the heat profile of any actual content?

I thought at one point there was some talk of using actual content for inter-measurement blanking rather than a black screen - wouldn't that be better? It seems to me that sequentially cycling through R, G, B, W (grey) fields at some predetermined luminance levels between measurements would be a better way to assure that all measurements are taken at temperatures as close to use temperatures as possible.

You could even balance the inter-measurement fields with the field used to take a measurement, so if r, g, b was measured with a field containing some RGB value (translated to an RGBW value by LG's algorithm), the R, G, B, W fields inserted between reading could be adjusted down by the intensity presented during measurement. So if measuring 10% saturated red at 50% luminance, for example, the inter-measrement red field would be presented well below 100% of it's usual intensity (but above 50% since the vompensation is based on heat which does not followa gamma curve).

I understand that until LG builds-in thermal compensation into the panel algorithms, there is always going to be drift between luminance values measured during calibration and luminance presented when viewing actual content but until then, it seems like taking measurements with the panel as close to estimated actiual-content temperature as possible should minimize the amount of that drift far better than calibrating 'cool'...
 

·
Registered
Joined
·
13,713 Posts
^^ I agree. If I'm leaving it overnight or while I'm out at work, I don't mind if it's 11,000 or 22,000 patches. Give me the most-precise calibration possible! :)
Yeah, since time is 'free' for us DIYers (with Autocal), we should be able to get more accurate calibrations than pros can deliver because a longer, more previse calibration did not 'cost' any more.

As far as pro calibrations, this would alao motivate the pros to push for a remote access solution.

If an automatic calibration can be set up and initiated remotely, even the calibrator can do something else (including sleep ;)) while the calibration is performed.

Even without that, if a pro calibrator has 5-10 calibrations lined up in one city/locale, they can spend one day setting up and launching all 5-10 calibrations and a second day checking the results, printing reports, and collecting their gear. Of course this would mean 5-10 meters, 5-10 computers, 5-10 pattern generators, etc... so the remote calibration solution is less costly (for customers with their own meters and computers).

Autocal should mean it is no longer the calibrators time that is the most valuable (and costly) resource, it is now the use of the calibration equipment that becomes the bottleneck.

Professional calibrators have always had a very different perspective on equipment/license cost than DIYers (for good reason), but Autocal could/should change that...
 

·
Registered
Joined
·
13,713 Posts
Hi,

Correct, OLED's maintain color saturation to low luminance levels also while alternative display technologies, including traditional CRT and LCD displays, as well as projectors, have a de-saturation effect as the luma levels drop below about 20% peak luma, depending on the actual display.

For more info about this, its here, to the 'OLED Displays': https://www.lightillusion.com/advanced_lightspace.html#oled_displays

Traditional Display Shadows Gamut



OLED Shadows Gamut




But, that is not the issue with the calibration, the issue is the use of the White sub-pixel, more detail there: https://www.lightillusion.com/forums/index.php?action=vthread&forum=8&topic=485#msg2817

For displays with inherent underlying non-linear inaccuracy[/b], as shown above when a display is profiled to itself (and not directly associated with thermal instability, which requires the use of 'Stabilisation' patches to overcome), a very large profiling sequence will be required, in combination with a very large 3D calibration LUT to overcome the non-linear issues.

WRGB OLED technology will inherently suffer such issues, due to the inclusion of the 'white' pixel, as this will distort the standard RGB color channel relationship.

If you look at this image...



...you can see the way the ‘vectors’ change direction, and how multiple ‘input’ points all point towards the same actual color point.

(The colored dot is the measured color point, color coded for dE accuracy, while the ‘vector line’ shows where the point should actually be...



All these data coming from SDR profiling data, stuff are lot of worse in HDR mode.


I understand that use of a white subpixel will introduce 'distortion' into a pure RGB mapping, but I'm not understanding why that distortion is inherently non-linear.

If we assume the RGB composition of each subpixel (R, G, B, W) is known and if we assume the color filters are pure (like QDCF) so R contains 100% red and 0% Blue and Green, etc... and W contains aR, bG, cB, then wouldn't the resulting mapping of any r,g,b triplet to RGBW be linear if the mapoing was properly implemented?

Is the 'inherent nonlinearity' you are referring to because the red filters pass a small amout of blue and green (and same for green and blue filters) or because the filters pass a 'bell-curve' of color spectra rather than a 'pure' spike, or because the white subpixel passes a complex curve of varibg intensities all WOLED spectra which is impossible to model (and shifts over age)? Or all three?

Said another way, with 'pue' color fiters for RGB (like QDCF) and proper modeling/translation algorithms could WRGB WOLED be linear or does the complex spectral output of the WOLED itself mean that even in that case linearity is impossible?

And if is is impossible even in that case, what about if the white subpixel istelf also has RGB QDCF (all 3, like for QDEF film used on current QLEDs)?
 

·
Registered
Joined
·
6,577 Posts
Discussion Starter #229
^^^
I agree, if you are a home enthusiast then you can set a longer blanking interval and go to bed but as far as calibrating with the panel having a specific load, that is a complex problem. We calibrate with an 18% window or full screen because, from what i understand, the average content has approximately 18% white. I don't want to get into and argument over this percentage but for arguments sake lets assume this is valid.

That said, if you were to use some sort of content in the blanking interval vs 10% gray as specrtacal recommends for SDR for exampe, you would need to use the exact same content every time or you wouldn't have a reference and your calibrations would vary based on panel loading.

So in my opinion, 10% gray should be ok since it's a known value. Also when you do a ful scan, at the end calman displays a full screen 10% gray patch so you probably want to use that as the blanking interval to keep the panel as constant as possible.

When i first started calibrating, this problem came to mind immediately but my guess is everyone has their own methods when calibrating that works for them.:)
 
  • Like
Reactions: L30Z3N

·
Registered
Joined
·
2,243 Posts
I’m a diy calibrator and where I think being able to do a large number lut overnight sounds like fun, the last time I tried doing something similar with an eecolor box my computer actually froze during the process and left a pattern up for an extended period of time. Thankfully I was only out running errands and it hadn’t been sitting for too too long...think that was with my f series oled.


Sent from my iPhone using Tapatalk
 

·
Registered
Joined
·
17,806 Posts
I understand why 6 hour calibration sessions are a non-starter for professional calibrators, but I'm not understanding why this is considered impractical for enthusiast DIY calibrations.

1/ Set everything up and initiate Autocal.

2/ Go to bed.

3/ Wake up the next mosting to find your calibration complete.

I understand that if the SW crashes before calbration is complete, this will not work, but that 's because of faulty SW, not impracticality.

A side benefit of overnight calibration sessions is that inter-measurement blanking intervals can be almost as long as you want them, so it seemscas though panel heating issues could be completely avoided.

So what am I missibg here? Now that we've moved from hands-on calibration to Autocal, why wouldn't an enthusiast want to perform the longest, most-precise calibration possible (while he/she is off doing something else):confused:
My post is not and never was directed at enthusiast.
 

·
Registered
Joined
·
17,806 Posts
This issue of 'managing heat' seems to be a big deal with getting an accurate calibration of these WOLEDs, but I have a question about that:

Don't you want the panel calibrated at the estimated heat it be at when displaying content? If you calibate the panel 'cool' won't that mean it is farther off than it needs to be when displaying content (which will inevitably heat it up)?
I think you are missing the point. No one said eliminate the heat during calibration. You manage it. You essentially need a patch set and insertion that mimics what regular Joe Schmo watches on a daily basis.

I thought at one point there was some talk of using actual content for inter-measurement blanking rather than a black screen - wouldn't that be better?[/quote]Been there, done that. Doesn’t work very well.
 

·
Registered
Joined
·
8,893 Posts
I understand that use of a white subpixel will introduce 'distortion' into a pure RGB mapping, but I'm not understanding why that distortion is inherently non-linear.

If we assume the RGB composition of each subpixel (R, G, B, W) is known and if we assume the color filters are pure (like QDCF) so R contains 100% red and 0% Blue and Green, etc... and W contains aR, bG, cB, then wouldn't the resulting mapping of any r,g,b triplet to RGBW be linear if the mapoing was properly implemented?

Is the 'inherent nonlinearity' you are referring to because the red filters pass a small amout of blue and green (and same for green and blue filters) or because the filters pass a 'bell-curve' of color spectra rather than a 'pure' spike, or because the white subpixel passes a complex curve of varibg intensities all WOLED spectra which is impossible to model (and shifts over age)? Or all three?

Said another way, with 'pue' color fiters for RGB (like QDCF) and proper modeling/translation algorithms could WRGB WOLED be linear or does the complex spectral output of the WOLED itself mean that even in that case linearity is impossible?

And if is is impossible even in that case, what about if the white subpixel istelf also has RGB QDCF (all 3, like for QDEF film used on current QLEDs)?
The output of the display is not linear and predictable when the input signal is changed in a linear and predictable way.

The causes display volumetric non-linearity.

In its simplest form, a displayed color is simply the sum of its components C= R+G+B.

If it is ‘linear’, then the combination is predictable in that changes in the input red signal has no effect on the light coming from the blue or green channel.

This is mostly true for old school CRT's, and for the most part LCD's and even LED's.

With a WOLED display, it has 4 emitters.

So, when the red signal is adjust, it will change the emission from the red LED, and may be also the white LED depending on the overall RGB signal that it is receiving.

This breaks the simple C= R+G+B rule.

The device now has a complex signal to light behavior.

This does not mean it is non-linear in its response.

If you have a non-linear system, you need a full 3d model to make it do what you want it to. The greater the non-linearity, the bigger the size of the 3D LUT required to manage it.

(Actually, if you had direct access to RGB and W, then you would be best generating a 4 channel cube! This is often done internally in printers that have more than 3 inks; same basic idea.)

Additionally, it seems that the in the WOLED display, the that the W channel is sometimes “off” completely and that there is a step change to minimum “on” luminance (again defendant on the RGB signal received). This step discontinuity may well be very small, but it compounds the problem, especially in the blacks/shadows.
 

·
Registered
Joined
·
6,577 Posts
Discussion Starter #236
You turn Dynamic Tone Mapping Off when you calibrate.
 

·
Registered
Joined
·
13,713 Posts
The output of the display is not linear and predictable when the input signal is changed in a linear and predictable way.

The causes display volumetric non-linearity.

In its simplest form, a displayed color is simply the sum of its components C= R+G+B.

If it is ‘linear’, then the combination is predictable in that changes in the input red signal has no effect on the light coming from the blue or green channel.

This is mostly true for old school CRT's, and for the most part LCD's and even LED's.

With a WOLED display, it has 4 emitters.

So, when the red signal is adjust, it will change the emission from the red LED, and may be also the white LED depending on the overall RGB signal that it is receiving.

This breaks the simple C= R+G+B rule.

The device now has a complex signal to light behavior.

This does not mean it is non-linear in its response.

If you have a non-linear system, you need a full 3d model to make it do what you want it to. The greater the non-linearity, the bigger the size of the 3D LUT required to manage it.

(Actually, if you had direct access to RGB and W, then you would be best generating a 4 channel cube! This is often done internally in printers that have more than 3 inks; same basic idea.)

Additionally, it seems that the in the WOLED display, the that the W channel is sometimes “off” completely and that there is a step change to minimum “on” luminance (again defendant on the RGB signal received). This step discontinuity may well be very small, but it compounds the problem, especially in the blacks/shadows.
I think the non-linearity arises from the fact that the spectrum of the white subpixel does not equal any linear combination of the spectral outputs of the red, green, and blue subpixels.

Would love to see plots of red, green, blue, red+green+blue, and white if anyone has the data...
 

·
Registered
Joined
·
79 Posts
I've spent two days with european LG C8, and tested 3D LUT calibration for SDR. My goal was to compare classic calibration ("by hand") to 26 pt 1D LUT + 3300iRP LUT / Lightning LUT / Matrix LUT. The results are documented in pdf files i'm attaching.

I belive this method has really great potential especially for post-production studios, but i'm not sure if it will be possible to use it instead of classic OSD-based calibration for home users. Why?

Because 26pt + lightning lut or matrix lut, which takes reasonable time (about 20 minutes per picture mode) turns out to be significantly worse than classic calibration. dE's are higher with matrix LUT like you can see in measurement report. It has visibly worse tracking of dark shades of skin tones. In movie scenes such as this one in "Casino Royale"



Mads Mikkelsen's face gets pinkish tint. Colors in this area rotate towards magenta. They twist even more in matrix/lightning mode than in classicaly calibrated. Lightning mode has additional problems with posterization near black. I suspected that since i had seen the gamma curve after calibration. It's not severe, but visible.

So in this case classic calibration > matrix/lightning. What about irp with medium density (3300 points)? This one fixes skin tone twist very well. Colors look richer and more natural. I remember the same kind of improvements when i tested EZ1000 3D-LUT calibrated and compared it against PVM-A250.



The issue is that it creates other problems. You can see in the report worse color tracking near red primary and cyan oversaturation These issues are visible in movies. Gradation is just a little bit worse. It doesn't posterize but has some gamma peaks visible in the gradient picture. Therefore i can't say that 3300 points irp lut is superior to classic calibration. Probably denser lookup table would do the job better, but the time it consumes makes it impractical in the field.

Maybe the results would be different if my unit had poor linearity. I calibrated hundreds of OLEDs last year and the linearity varied significantly. The one i received is very linear so it's hard to make it better with Calman 3D LUT.

This is why, for now, i would love to see DDC communication which allows to adjust OSD settings, same way like in Panasonic ISF-certified displays, or Samsung with Ex-Link. In most calibration cases it will do the best job i believe. Not to mention that without ability to backup both 1D and 3D LUT this method is not an option for commercial pro calibrations at all.
 

Attachments

·
aka jfinnie
Joined
·
4,220 Posts
I belive this method has really great potential especially for post-production studios, but i'm not sure if it will be possible to use it instead of classic OSD-based calibration for home users. Why?
Interesting results, thanks for sharing. Can you also share the report for this set with hand calibration?

What's the reason you suggest it is good for post production but not for home use? Seems that if the reason not to use it is accuracy then there is even less reason to use it for post production facilities?

Did you see this from Steve @Light Illusion about the poor stability / drifting of WRGB OLED? Were you doing anything to try to counteract the panel drift / stability?
https://www.lightillusion.com/forums/index.php?action=vthread&forum=8&topic=485#msg2829

The drift chart is pretty shocking. It looks like LG must have some compensation built in which has "steps" causing the sawtooth drift chart behaviour. Pretty ugly.
 

·
Registered
Joined
·
588 Posts
I've spent two days with european LG C8, and tested 3D LUT calibration for SDR. My goal was to compare classic calibration ("by hand") to 26 pt 1D LUT + 3300iRP LUT / Lightning LUT / Matrix LUT. The results are documented in pdf files i'm attaching.



I belive this method has really great potential especially for post-production studios, but i'm not sure if it will be possible to use it instead of classic OSD-based calibration for home users. Why?



Because 26pt + lightning lut or matrix lut, which takes reasonable time (about 20 minutes per picture mode) turns out to be significantly worse than classic calibration. dE's are higher with matrix LUT like you can see in measurement report. It has visibly worse tracking of dark shades of skin tones. In movie scenes such as this one in "Casino Royale"







Mads Mikkelsen's face gets pinkish tint. Colors in this area rotate towards magenta. They twist even more in matrix/lightning mode than in classicaly calibrated. Lightning mode has additional problems with posterization near black. I suspected that since i had seen the gamma curve after calibration. It's not severe, but visible.



So in this case classic calibration > matrix/lightning. What about irp with medium density (3300 points)? This one fixes skin tone twist very well. Colors look richer and more natural. I remember the same kind of improvements when i tested EZ1000 3D-LUT calibrated and compared it against PVM-A250.







The issue is that it creates other problems. You can see in the report worse color tracking near red primary and cyan oversaturation These issues are visible in movies. Gradation is just a little bit worse. It doesn't posterize but has some gamma peaks visible in the gradient picture. Therefore i can't say that 3300 points irp lut is superior to classic calibration. Probably denser lookup table would do the job better, but the time it consumes makes it impractical in the field.



Maybe the results would be different if my unit had poor linearity. I calibrated hundreds of OLEDs last year and the linearity varied significantly. The one i received is very linear so it's hard to make it better with Calman 3D LUT.



This is why, for now, i would love to see DDC communication which allows to adjust OSD settings, same way like in Panasonic ISF-certified displays, or Samsung with Ex-Link. In most calibration cases it will do the best job i believe. Not to mention that without ability to backup both 1D and 3D LUT this method is not an option for commercial pro calibrations at all.


First of all thx for sharing your thoughts and results with us. Regarding to your reports and visual tests these aren't good news at all.

Did you use the pattern Inseration mode? And why didn't you run a 42p Grayscale run to improve the low IRE?

Furthermore did you also tested a iRP LUT with less points
 
  • Like
Reactions: jrref
221 - 240 of 3759 Posts
Top