AVS Forum banner

141 - 160 of 237 Posts

·
Registered
Joined
·
3,885 Posts
My laptop's bluetooth chipset is Marvell, which I have not always had the best luck with over the years (they also make USB chipsets). No issues with other devices I own but it does not play nice with the Pi. Not a huge deal for me so far as an Ethernet connection to my router is working.

But, to your point, I have an Ethernet to USB-C adapter on order just to rule out any potential issues.



It was answered sufficiently by others above, but instead of AutoCal using the internal pattern generator of the LG, you just use the Pi as an external (and superior) pattern generator. Ted's disc has patterns to use for validation pre and post calibration. You can just add this to the Pi instead of running it via a Blu Ray. Quality of life improvement.

I was implying that if you are not technically inclined at all, you may just want to continue using the internal pattern generator on the LG and AutoCal and call it a day. Depends on if all the time setting this stuff up is worth it to you. Some (aka my wife) would argue that all of this is the very definition of splitting hairs. Your mileage may vary.
Thanksm

I have tried multiple runs with the ITPG with not good results. Could be my meter too but this sounds like a fun project.

Just to be clear Calman autocal integrates the Pi instead of the internal generator but all other CM settings still apply?

bob
 

·
Registered
Joined
·
9,498 Posts
Thanksm

I have tried multiple runs with the ITPG with not good results. Could be my meter too but this sounds like a fun project.

Just to be clear Calman autocal integrates the Pi instead of the internal generator but all other CM settings still apply?

bob
Hi Bob,

Which TV models you want to calibrate?
 

·
Registered
Joined
·
9,498 Posts

OK, using PGenerator, you will replace the SDR iTPG of your TV.

The procedure you know will be the same, but you will connect PGenerator to your source instead of iTPG.

Before connecting with TV, display a patch from PGenerator (instead of using USB and iTPG).

If you get an HD Fury device with HDR10/AVI injection, you can use it to replace the HDR iTPG of your TV and verify the HDR10 post-calibration also.
 

·
Registered
Joined
·
1,248 Posts
It seems an unstated premise is that displays will process 8-bit RGB Limited with the same accuracy that they process 8-bit YCbCr 4:2:2 (SDR) and 10-bit YCbCr 4:2:2 (HDR). I'm curious, has someone validated that across a wide variety of makes and models of displays? Or is there some other reason to believe that it's a near-universal truth? FWIW, here are back-to-back HDR ColorChecker measurements I made on my 75" Samsung Q90R, using a C6 HDR2000 and a VideoForge Pro (v1.13) that make me think it's no so true for my display.

3040079
 

·
Registered
Joined
·
9,498 Posts
It seems an unstated premise is that displays will process 8-bit RGB Limited with the same accuracy that they process 8-bit YCbCr 4:2:2 (SDR) and 10-bit YCbCr 4:2:2 (HDR). I'm curious, has someone validated that across a wide variety of makes and models of displays? Or is there some other reason to believe that it's a near-universal truth? FWIW, here are back-to-back HDR ColorChecker measurements I made on my 75" Samsung Q90R, using a C6 HDR2000 and a VideoForge Pro (v1.13) that make me think it's no so true for my display.

View attachment 3040079
Hi,

Can you please post some more detail about your testing?

Are you comparing VideoForge PRO, RGB-Limited REC2020 vs. YCbCr 4:2:2 REC.2020 to these pictures?

How stable is the display? If you repeat the same measurement (for example, 2-3 times measure withy YCbCr), what kind of repeatability you have?

After many testing, we have found that LG's are calibrated better for SDR 3D LUT with RGB 16-255 patch generation.

Displays will have a difference if you send RGB vs. YCbCt signaling.

Just with LG's, the results are better when you will use RGB 16-255 for profiling and then verify with YCbCr instead of using YCbCr for profiling and YCbCr.

PGen is bit-perfect with RGB only, so users have to verify the results with RGB, or use my calibration disk to verify with YCbCr (manually).
 

·
Registered
Joined
·
1,089 Posts
Discussion Starter #147 (Edited)
Can I just clarify, when you say 8-bit RGB Limited what you mean by that? I mean, did you set the actual generator at RGB 16-235, or did you set the generator at RGB Full 0-255, and use video levels 16-235 within the software?

Of course, it is a known fact that not all displays will process RGB and YCbCr in the same way. In the same way that not all devices will output RGB and YCbCr in the same way. But that's not the point of the PGenerator.

Disregarding RGB vs YCC and how various displays handle it for the moment, the point of any generator is to reproduce via its HDMI output the exact RGB triplet that the software requests. Not all devices can do it, just look at The Cheap Test Pattern Generator Thread for evidence of that.

Let's say you use a streaming device to generate patterns in YCbCr, but that streaming device introduces extreme errors into the patch displayed on screen compared to the one that was requested (as many of them do). That's great, your calibration will be valid for that one particular input using that one particular device. And if that is the only device you are ever going to use, then it is a perfectly suitable solution. But what if it isn't the only device you use? What if you use several devices as well as internal apps?

In this case the PGenerator, when set to RGB Full with the levels controlled from software, will reproduce bit-accurate patterns that exactly match those that are requested from software. as verified with digital pixel analysers. No digital errors at all. Pixel perfect, as requested. Many notebook HDMI outputs can do this too, but unless you have a way of verifying it with said pixel analysers, you will never know for certain, and there is always the possibility this could change with an update.

Very few devices output accurate YCbCr patterns outside of expensive dedicated test pattern generators, and even some of those were only fixed to be perfectly accurate in YCbCr many years after they were initially released. For those that have money to spend on such devices, they certainly won't be looking here at a DIY kit to build their own. For those on a budget, it is a solution to provide a known quantity. That being, bit accurate RGB patterns.

Also, HDR is probably not the best mode to do such a comparison to be honest. Despite being a QLED which will be much more stable than an OLED, it still won't be fully stable so measurements can fluctuate. Maybe try the the comparison in SDR. The display will still process the YCbCr/RGB signal the same whether it is in HDR or SDR, but at least with SDR you rule out the possibility of fluctuating measurements.
 
  • Like
Reactions: mrtickleuk

·
Registered
Joined
·
9,498 Posts
Just to add some detail to the discussion.

Murideo and VideoForge PRO are produced from the same company.

While units released to the market from 2014/2015, Murideo became bit-perfect to 4:2:2 output in November 2019.

 

·
Registered
Joined
·
1,248 Posts
Are you comparing VideoForge PRO, RGB-Limited REC2020 vs. YCbCr 4:2:2 REC.2020 to these pictures?

How stable is the display? If you repeat the same measurement (for example, 2-3 times measure withy YCbCr), what kind of repeatability you have?
Yes, HDR10 BT.2020 2160p24 in both cases selected in the CalMAN Source settings. I ran the measurements twice each, they seemed reasonably stable.
 

·
Registered
Joined
·
9,498 Posts
Yes, HDR10 BT.2020 2160p24 in both cases selected in the CalMAN Source settings I ran the measurements twice each, they seemed reasonably stable.
Have you checked that the black level to the HDMI input was correct for RGB signaling because TVs can't detect that you are sending video levels?

Can you post the CIE chart also or a full screenshot?

Using dE, you can't see the error; you need additional charts when you perform comparisons.

dE 1.0 can have an error to blue, while dE 1.0 can have a towards red.

Both will look as dE 1.0 to the chart for the same color.
 

·
Registered
Joined
·
1,248 Posts
Have you checked that the black level to the HDMI input was correct for RGB signaling because TVs can't detect that you are sending video levels?

Can you post the CIE chart also or a full screenshot?
Yes, HDMI Black Level set to Low on the Q90R, which is correct for Samsung for RGB Limited. Screenshots attached. (Sorry I had mismatched scale on dE chart, now corrected.)

HDR 10-bit 422.png HDR 8-bit RGB.png
 

·
Registered
Joined
·
1,248 Posts
Can I just clarify, when you say 8-bit RGB Limited what you mean by that?
Video (16-235) Luminance Levels selected in CalMAN options, RGB Limited 8-bit Color Format selected in CalMAN Source settings for VideoForge Pro.
Disregarding RGB vs YCC and how various displays handle it for the moment, the point of any generator is to reproduce via its HDMI output the exact RGB triplet that the software requests.
Sure, and I'm not trying to knock the solutions in this thread, I think they're fantastic. Just pointing out that, without someone testing, you don't know if bit-perfect RGB patterns will produce the desired results for YCbCr content.
The display will still process the YCbCr/RGB signal the same whether it is in HDR or SDR
On that point I disagree, the colorspace conversions are different for Rec 709 vs 2020. I certainly would not assume the processing is the same.
 

·
Registered
Joined
·
9,498 Posts
Yes, HDMI Black Level set to Low on the Q90R, which is correct for Samsung for RGB Limited. Screenshots attached. (Sorry I had mismatched scale on dE chart, now corrected.)

View attachment 3040133 View attachment 3040126
I see that VPF FW you are using released in September 2019. The 'correct' Murideo FW for 4:2:2 released in November 2019.

I think its better to open a new thread to compare what you see to the Samsung with VFP, and then some others may comment with Accupel or Murideo.

Post there some measurements with SDR and HDR by using 4:4:4 also and probably other people may compare with Murideo (which has verified FW, with no errors with YCbCr output)
 

·
Registered
Joined
·
9,498 Posts
Sure, and I'm not trying to knock the solutions in this thread, I think they're fantastic. Just pointing out that, without someone testing, you don't know if bit-perfect RGB patterns will produce the desired results for YCbCr content.
The problem is that we don't know if your VFP is generating accurate 4:2:2 as it was available in the market for year, but it was not not-bit-perfect to all modes.

I can't commend to something I don't have access to personally test, while I have here the capability to examine the digital levels of any pixel up to 2160p SDR/HDR10 input signal.

If someone, not affiliated with a company, can examine the digital levels, we can see its accuracy.
 

·
Registered
Joined
·
9,498 Posts
Video (16-235) Luminance Levels selected in CalMAN options, RGB Limited 8-bit Color Format selected in CalMAN Source settings for VideoForge Pro.
When you compare HDR patches, use the same bit-depth per each colorspace.
 

·
Registered
Joined
·
1,248 Posts
I see that VPF FW you are using released in September 2019. The 'correct' Murideo FW for 4:2:2 released in November 2019.
FW 1.13 may have a date-stamp of September, but AFAIK it was not made public for download until November.
I think its better to open a new thread to compare what you see to the Samsung with VFP, and then some others may comment with Accupel or Murideo.
I don't feel compelled to take this further down the rabbit hole.
The problem is that we don't know if your VFP is generating accurate 4:2:2 as it was available in the market for year, but it was not not-bit-perfect to all modes.
I'm personally willing to believe Portrait Displays when they say the bug was fixed.
When you compare HDR patches, use the same bit-depth per each colorspace.
That would defeat the whole point, which is to compare 8-bit RGB (as would be output by RPI PGenerator + HD Fury) with 10-bit YCbCr 4:2:2 (as would be found in most HDR content).
 

·
Registered
Joined
·
1,089 Posts
Discussion Starter #157 (Edited)
Sure, and I'm not trying to knock the solutions in this thread, I think they're fantastic. Just pointing out that, without someone testing, you don't know if bit-perfect RGB patterns will produce the desired results for YCbCr content.
I think it's fair to say that anyone with any knowledge at all will know that different displays can indeed handle RGB and YCbCr differently. It's also fair to say that a Panasonic BD-Player outputting YCC422 and and ATV4K outputting YCC422 can also output drastically differently, perhaps even one or both of them different to the VFPro outputting YCC422.

But whilst such comparisons are a valid discussion, this is not really a comparison thread. It is solely intended to be a consolidation of information regarding setup and use of the PGenerator.

From the first post:
The purpose of this thread is to make an attempt to consolidate as much information as possible, giving users a ‘One Stop Shop’ of links and guides for set-up and operation of PGenerator only.
It is clearly stated many times in many posts that it is accurate when outputting an RGB Full signal only. This is no different to when others recommend generating patterns from a notebook/desktop HDMI output with the exception that when using the PGenerator in this state, it will indeed be outputting accurately now and always, and will not be altered in the future by any potential Operating System or Firmware Update.

Perhaps Ted's suggestion of creating another thread for such discussions would be a better idea. But for what it's worth, I also have a DVDO AVLab 4K TPG which is verified accurate in RGB and YCbCr, and I personally use my PGenerator for everything ;)
 

·
Registered
Joined
·
9,498 Posts
I'm personally willing to believe Portrait Displays when they say the bug was fixed.
Sorry but I don't believe anything regarding any testing when it's coming from SpectraCAL.

They have found anything they have tested as bit-perfect:

LG iTPG:


But see:


Murideo:


ATV:


FireStick:



..so everything SpectaCAL has tested is bit-perfect.

But we found that all these products were not bit-perfect.

So I have stopped believing anything tested from any company; I perform my tests or trust testing coming from verified contacts only.

When you calibrate for HDR with PGenetor and Fury, you performing RGB balance calibration to the grayscale.

When you measure just grayscale, you will not see a difference if you compare 8-bit RGB vs. YCbCr 4:2:2 REC.2020.

HDR calibration is not providing full calibration to current displays available to the market.

When you measure pure primary colors, with 100% saturation, you will not have any difference between RGB 8-bit vs. 10-bit YCbCr 2020.

The colorspace differences will appear when you will measure non 100% Saturation colors, so to the verification of ColorChecker, for example, but since you can't calibrate the displays to these colors, no worries.

When you measure x-Point Luminance with 100% Saturation of a Primary color, you will not see any difference between different colorspace during patch generation.
 

·
Registered
Joined
·
9,593 Posts
Just to add some detail to the discussion.

Murideo and VideoForge PRO are produced from the same company.

While units released to the market from 2014/2015, Murideo became bit-perfect to 4:2:2 output in November 2019.

Did they give customers compensation for the 4-5 years that their product was faulty? I can't imagine how many total calibrations would have been affected in all that time, adding up just user, let along all the users.
 

·
Registered
Joined
·
2,767 Posts
Not to sidetrack the conversation, but want to make sure I'm understanding correctly.

OK so I calibrated my Sony A9G using Autocal feature with Calman from Home....using the bit-perfect (RGB Full) PGenerator from RPi, and my settings with Calman and TV were the 16-235/Limited per their instructions.

Since my TV watching is done via blu-ray player (UB9000) and it outputs YCbCr 4:4:4....I would need Ted's disc in order to confirm the conversion is still accurate within the player and TV, right?
 
141 - 160 of 237 Posts
Top