HCFR - Open source projector and display calibration software - Page 361 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 1408Likes
Reply
 
Thread Tools
post #10801 of 11804 Old 06-07-2018, 09:11 AM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by webdove View Post
To allow precision measurements I set measures/parameters/grayscale_levels to 100 (I would have preferred 200 for more precision, but HCFR won’t let you type in a number more than 100. Zoyd, allowing that to be 200 would be helpful).
One of the outstanding feature requests is to have HCFR use code value steps (up to 256 steps) rather than % steps for input levels, to avoid rounding errors.

Last edited by Dominic Chan; 06-07-2018 at 11:44 AM.
Dominic Chan is online now  
Sponsored Links
Advertisement
 
post #10802 of 11804 Old 06-07-2018, 02:44 PM
Newbie
 
Join Date: Jun 2018
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 0
After all your teaching please a final check to my plan B Configuration:

Potplayer Full
MadVR Full
GPU Limited (YCbCr 4:2:2 10 Bit)
TV/VPR Limited

HCFR set to 0-255 to avoid double scaling

Thank you again
Silver12 is offline  
post #10803 of 11804 Old 06-07-2018, 04:01 PM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by Silver12 View Post
After all your teaching please a final check to my plan B Configuration:

Potplayer Full
MadVR Full
GPU Limited (YCbCr 4:2:2 10 Bit)
TV/VPR Limited

HCFR set to 0-255 to avoid double scaling

Thank you again
What happened to Plan A, which is recommended?
Dominic Chan is online now  
Sponsored Links
Advertisement
 
post #10804 of 11804 Old 06-08-2018, 03:04 AM
Newbie
 
Join Date: Jun 2018
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 0
Quote:
Originally Posted by Dominic Chan View Post
What happened to Plan A, which is recommended?
Plan A:
Potplayer Limited
MadVR Limited
GPU Full (RGB, 8 Bit only because I have HDMI 1.4 for 2160p, not HDMI 2.0)
TV/VPR Limited

Since my Plan A is just 8 Bit (RGB Full) vs Plan B 10 Bit (YCC Limited), I am considering both possibilities.
I have understood that A is the choice with RGB 10 Bit, no doubt for me there, but B could be as good since I only have RGB 8 Bit.
I guess I think that 10 Bit is WOW vs 8 Bit, marketing influence problably, 10 Bit is UHD, is the future…….
This is why I ask you, if you confirm that A is better even with the 8 Bit limitation, then I will go for it.
Silver12 is offline  
post #10805 of 11804 Old 06-08-2018, 05:14 AM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by Silver12 View Post
Plan A:
Potplayer Limited
MadVR Limited
GPU Full (RGB, 8 Bit only because I have HDMI 1.4 for 2160p, not HDMI 2.0)
TV/VPR Limited

Since my Plan A is just 8 Bit (RGB Full) vs Plan B 10 Bit (YCC Limited), I am considering both possibilities.
I have understood that A is the choice with RGB 10 Bit, no doubt for me there, but B could be as good since I only have RGB 8 Bit.
I guess I think that 10 Bit is WOW vs 8 Bit, marketing influence problably, 10 Bit is UHD, is the future…….
This is why I ask you, if you confirm that A is better even with the 8 Bit limitation, then I will go for it.
Didn’t zoyd already confirm that in post 10800 above?
Dominic Chan is online now  
post #10806 of 11804 Old 06-08-2018, 05:53 AM
Newbie
 
Join Date: Jun 2018
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 0
My fault, I thought A was better but even B was worth a try
Now everything is clear and forgive my misunderstanding


Yesterday I received my new X-500 (used) and I will go with plan A and do my first attempt for calibration, plus putting some velvet on the ceiling!!



thank you all
Silver12 is offline  
post #10807 of 11804 Old 06-08-2018, 07:39 AM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by Silver12 View Post
My fault, I thought A was better but even B was worth a try
Now everything is clear and forgive my misunderstanding


Yesterday I received my new X-500 (used) and I will go with plan A and do my first attempt for calibration, plus putting some velvet on the ceiling!!



thank you all
A is better but you need to "force" the GPU to full range. You need to measure to confirm; it may not sufficient to set it in the GPU control panel
Personally, I always start with Plan B (which is much easier to set up), and then switch to Plan A and confirm that the levels are still correct.
Dominic Chan is online now  
post #10808 of 11804 Old 06-08-2018, 08:13 AM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by Dominic Chan View Post
One of the outstanding feature requests is to have HCFR use code value steps (up to 256 steps) rather than % steps for input levels, to avoid rounding errors.
That would be splendid!

Do you know if my items #5 and #6 made it into the list of potential changes for HCFR?

“5. Zoyd, It would be helpful if HCFR reported the actual quantized floating point brightness in the first row of the data table as is reported in the tooltips when you hover over a data point. It would also be helpful if the 8bit, 10bit or 12bit code values (depending on what bit-depth the graphic card is set to) were reported on the left screen data column to make quantization things clearer.

6. Zoyd, It would be super duper helpful if one of the scaling choices of RGB levels with gamma for both HDR and SDR was "perceptual" using the Barten perception function mentioned in ITU BT.2390. Essentially the vertical scale at each horizontal level of brightness would then be scaled to the minimum detectable difference at that brightness rather than the percentage difference regardless of brightness as it is now. Currently the vertical scale is much too sensitive at the left and too insensitive at the right. Something like full scale up or down being 10 minimally detectable error units would be great.“
webdove is offline  
post #10809 of 11804 Old 06-08-2018, 08:30 AM
Newbie
 
Join Date: Jun 2018
Posts: 12
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 0
Quote:
Originally Posted by Dominic Chan View Post
A is better but you need to "force" the GPU to full range. You need to measure to confirm; it may not sufficient to set it in the GPU control panel
Personally, I always start with Plan B (which is much easier to set up), and then switch to Plan A and confirm that the levels are still correct.

So, I should do this way?



Set Plan B
Calibrate HCFR (0-255 to avoid double scaling)


Set Plan A
Calibrate HCFR


Check the 2 calibrations, if everything is ok with both then I stay on A settings
Silver12 is offline  
post #10810 of 11804 Old 06-08-2018, 08:49 AM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by Silver12 View Post
So, I should do this way?

Set Plan B
Calibrate HCFR (0-255 to avoid double scaling)

Set Plan A
Calibrate HCFR

Check the 2 calibrations, if everything is ok with both then I stay on A settings
You only need to measure greyscale using Plan B (which is relatively foolproof).

Then switch to Plan A (which is more accurate but potentially harder to set up) and repeat the greyscale. If that matches the results from B (within a few percent) then stay with A and do all the calibrations. If not, troubleshoot and fix the problem until they match.
Dominic Chan is online now  
post #10811 of 11804 Old 06-08-2018, 08:57 AM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by webdove View Post
Do you know if my items #5 and #6 made it into the list of potential changes for HCFR?
I don't have any inside information
I only know about the other one as I brought it up previously.
Dominic Chan is online now  
post #10812 of 11804 Old 06-08-2018, 08:27 PM
Senior Member
 
Kamikaze_Ice's Avatar
 
Join Date: Jan 2007
Posts: 366
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 211 Post(s)
Liked: 115
Quote:
Originally Posted by webdove View Post
Here is a fun fact.
I initially setup my Vertex to “follow” the colorspace and Deep Color of the input.
I set my Nvidia GTX 1050 for RGB Full 8bit.
However, the vertex reports an input signal that is RGB Limited 8bit. Which it transmits unchanged as RGB Limited 8bit to the TV.
If I do a sweep, the first several codes are all 0 luminance as you would expect by sending full 8bit codes to a TV has been told to expect limited 8bit. Thus both the Vertex and the TV are behaving as though the PC is generating limited 8bit when I set the display to full 8bit.
(BTW, it is not calibrated at the moment. I just did a reset. I have HCFR set to generate 65 levels in RGB full starting with codes 3, 8, 12, 16, 20, 24, 28. They should be every 4, but HCFR is probably truncating to integer percent brightness then truncating back to RGB code).
If I set my Nvidia to RGB Limited 8 bit, the Vertex still reports RGB Limited 8bits input and output, but now a sweep looks normal for an uncalibrated condition.
Thus the evidence indicates that the Nvidia when switched to RGB Full is indeed generating RGB full data, but is incorrectly reporting with metadata that the data is RGB limited according to both the Vertex and the LG E6.
Are you using the "PC Icon" mode AND have the latest firmware on your E6?
If so, it is likely screwed up. I know HDR with that combination forces a rec.709 gamut for whatever reason.
I had to downgrade back to 04.30.95 to have the proper gamut. I didn't check SDR, but I wouldn't trust that firmware version. I was to lazy to check other Icon modes since I don't use them anyways.

I don't have a Vertex but with this older firmware I've not seen any unexpected levels ("Auto" with SDR picks the correct levels for what I set my video card or PS3 to).
E6 + EVGA GTX 1080 (FTW), with latest driver and firmware ( http://www.nvidia.com/object/nv-uefi-update-x64.html ).
It's not listed there but this firmware apparently allows one to "enable HDR color accuracy" according to guru3d, which I'm assuming is only for DisplayPort connections (which I don't use). I figured why not stay up-to-date just in case.
Regardless I didn't notice any changes after applying this firmware update.


---


Quote:
Originally Posted by zoyd View Post
*SNIP*
...
I'd still choose 1. The higher bit depth for the limited conversion doesn't guarantee it's any better. You can always check the magnitude of the differences and then if they are small, use whichever is more convenient.
As far as I know MadVR should be set to 8-bit depth to avoid banding/dithering problems with driver/os set to 10/12-bit depth. Simply put GPU/OS dithering is crap which is always used for 10+ bit depth content outside of Exclusive Full Screen modes (or if Windows HDR is enabled, but has it's own banding issues).
You shouldn't see a difference between 8 and 10 bit depth with Exclusive Full Screen, which seems to be the general consensus in the MadVR thread on doom9.


---


Quote:
Originally Posted by Dominic Chan View Post
One of the outstanding feature requests is to have HCFR use code value steps (up to 256 steps) rather than % steps for input levels, to avoid rounding errors.
I'd like to see values for 10-bit measurements (like HDR), and potentially a way to use them via custom color checker list.

... and I wouldn't mind having more than 100 measure points either, lol.
Basically so I can pick which point best aligns with my display's 20-point controls which changes depending on brightness/contrast setting (especially for LG's unusual HDR points).
That's me being OCD though,

Still, I think would be nice to have more points for higher bit depths at some point.
Kamikaze_Ice is offline  
post #10813 of 11804 Old 06-09-2018, 12:36 PM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by zoyd View Post
I'll have to fire mine up this weekend and check your screenshots. DV has to be done in full range RGB end-to-end but I don't remember if you can actually change the range on the TV side, if so it needs to be full range. Also, the program will automatically set the Vertex to the right input/output scheme and override however you have it set.
I have tested an Nvidia GTX 1050, AMD HD5450, Playstation 3 and XBOne. In all cases when set to either RGB8 Full or RGB8 Limited, the Vertex reports them as RGB8 Limited.

If you then set the vertex scaler to output RGB8 Full rather than follow then the vertex translates the incoming 16-235 to outgoing 0-255.

I still have yet to get a visual image from my LG on the extended second video output in RGB Full mode on the Nvidia when I check DV under vertex in references.
I see the Dolby Vision icon, but the screen is black.
webdove is offline  
post #10814 of 11804 Old 06-09-2018, 01:02 PM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by Kamikaze_Ice View Post
Are you using the "PC Icon" mode AND have the latest firmware on your E6?
If so, it is likely screwed up. I know HDR with that combination forces a rec.709 gamut for whatever reason.
I had to downgrade back to 04.30.95 to have the proper gamut. I didn't check SDR, but I wouldn't trust that firmware version. I was too lazy to check other Icon modes since I don't use them anyways.
At some point I want to see if I can replicate what you are describing, but I know that I have my inputs set to the HDMI icon and not PC.

Quote:
Originally Posted by Kamikaze_Ice View Post
I don't have a Vertex but with this older firmware I've not seen any unexpected levels ("Auto" with SDR picks the correct levels for what I set my video card or PS3 to).
E6 + EVGA GTX 1080 (FTW), with latest driver and firmware ( http://www.nvidia.com/object/nv-uefi-update-x64.html ).
It's not listed there but this firmware apparently allows one to "enable HDR color accuracy" according to guru3d, which I'm assuming is only for DisplayPort connections (which I don't use). I figured why not stay up-to-date just in case.
Regardless I didn't notice any changes after applying this firmware update.
The two PC cards, PS3 and XBOne all behave the same. When set to RGB Full the Vertex input reports RGB Limited. I don't know whether there is no metadata for code range in the standard and the Vertex just assumes Limited or if all four of those video devices are failing to report the code range properly (which seems unlikely).

BTW, I could not find any way to get the LG E6 to report what it believes is the input coding system (YCrCB or RGB, bits and range). This is the first TV I have had that lacks that function.

Quote:
Originally Posted by Kamikaze_Ice View Post
As far as I know MadVR should be set to 8-bit depth to avoid banding/dithering problems with driver/os set to 10/12-bit depth. Simply put GPU/OS dithering is crap which is always used for 10+ bit depth content outside of Exclusive Full Screen modes (or if Windows HDR is enabled, but has it's own banding issues).
You shouldn't see a difference between 8 and 10 bit depth with Exclusive Full Screen, which seems to be the general consensus in the MadVR thread on doom9.
I was trying madvr yesterday. It appeared to be full screen. I am not sure how I verify exclusivity. It also does not seem to report what bit depth and coding it thinks are in use.

Quote:
Originally Posted by Kamikaze_Ice View Post
I'd like to see values for 10-bit measurements (like HDR), and potentially a way to use them via custom color checker list.

... and I wouldn't mind having more than 100 measure points either, lol.
Basically so I can pick which point best aligns with my display's 20-point controls which changes depending on brightness/contrast setting (especially for LG's unusual HDR points).
That's me being OCD though,

Still, I think would be nice to have more points for higher bit depths at some point.
Agreed. Perhaps having every code value from 0 to 32 then more sparse up to max.
webdove is offline  
post #10815 of 11804 Old 06-09-2018, 01:41 PM
Newbie
 
Join Date: Sep 2017
Posts: 11
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 10 Post(s)
Liked: 0
Black frame insertion

@zoyd

Hi zoyd,

Thank you for your work on HCFR and this thread . I have a request for the "black frame inserstion" feature:

Could you please allow a frequency setting of 1, so that after every pattern a black frame is shown, and add a setting to define the number of ms this black frame will be shown? For which time it is shown right?

ConnecTEDDD did some test with LG 8 and showed that inserting a black frame after every pattern can reduce panel drift: https://www.avsforum.com/forum/139-di...l#post56281456
maxOLED is offline  
post #10816 of 11804 Old 06-09-2018, 03:16 PM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by webdove View Post
Perhaps having every code value from 0 to 32 then more sparse up to max.
It may be more logical to implement the “high resolution” input range as part of Near Black Grey Scale.

Last edited by Dominic Chan; 06-09-2018 at 03:31 PM.
Dominic Chan is online now  
post #10817 of 11804 Old 06-09-2018, 08:14 PM
Senior Member
 
Kamikaze_Ice's Avatar
 
Join Date: Jan 2007
Posts: 366
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 211 Post(s)
Liked: 115
Quote:
Originally Posted by webdove View Post
At some point I want to see if I can replicate what you are describing, but I know that I have my inputs set to the HDMI icon and not PC.
PC Icon mode is the only way to get 4:4:4 chroma on the 2016 OLEDs, with a bonus of low input latency very close to Game modes without loosing access to calibration+CMS controls.
I don't have the option of using anything else as I use my E6 as my computer monitor.

If/when you check, simply measure primaries and compare (bt.2020 vs rec.709 with Windows HDR enabled). I didn't compare SDR on PC Icon at all as I immediately downgraded at that point.

Since the issue I poorly described happened from a recent firmware update, and someone with a 2017 model described this same behavior in the MadVR thread on doom9, I think this may effect any LG display with that version of WebOS (3.5? Whatever version came with the firmware that replaces the dog/flower/mushroom/mountain no signal screens).

I don't think there are enough complaints to get LG to fix this. Look how long and loud we had to get for HDR Game mode to be added, and then look at all the complaints about the darker tone mapping that replaced the original one for Game mode (darker to not clip) and all LG had to say is "enable Dynamic Contrast and set to HIGH for correct picture".
You know something is wrong when your phone/online support reps are telling them it's wrong lol.


Quote:
Originally Posted by webdove View Post
The two PC cards, PS3 and XBOne all behave the same. When set to RGB Full the Vertex input reports RGB Limited. I don't know whether there is no metadata for code range in the standard and the Vertex just assumes Limited or if all four of those video devices are failing to report the code range properly (which seems unlikely).
All I know about the Vertex is it can be used to force custom metadata flags, so if results from consoles and video card match this leads me to believe the issue lies with the Vertex.
Ignoring what it reports, do the levels LOOK like they match what you set them to (ex: 0-255/full from GPU and display at HIGH)?
If so, perhaps the Vertex wants to be like LG with backwards names (IMO "high" black should be 16, makes more logical sense to me).

I don't know if it will be helpful, but there are two different EDIDs on the E6, one pair for each input port. One is used when the "HDMI ULTRA HD Deep Color" option enabled and another is used when it's disabled.
Service Manual has timings and raw data for EDID blocks : https://lg.encompass.com/shop/model_...OLED55E6PU.pdf

There may be a few bad timings in the EDID, as I've had a few of them that resulted in a black screen with audio when I was playing around trying to see if I could force 4:2:0 or 4:2:2 at 10-bit depth for making custom resolutions for MadVR. None of them worked, but I think there is something going on as the non-deep color EDID would more frequently show as supporting 10-bit, but none of them worked (black screen + audio), even trying the ones that my PS4 reported as 10-bit which did work as expected.


Quote:
Originally Posted by webdove View Post
BTW, I could not find any way to get the LG E6 to report what it believes is the input coding system (YCrCB or RGB, bits and range). This is the first TV I have had that lacks that function.
I really wish this information was shown, it's not even in the service menu this time.
You can see SOME flags by clicking on the Input label in the top left. This will show flags for stuff like HDR, bt.2020, audio quality and crap like that (I mean it's pretty easy to know if HDR is on...)


Quote:
Originally Posted by webdove View Post
I was trying madvr yesterday. It appeared to be full screen. I am not sure how I verify exclusivity. It also does not seem to report what bit depth and coding it thinks are in use.
Press CTRL+J to show the OSD, it has everything you need to know.
You can also see which API is being used for HDR.

Another way to check for exclusive full screen is to enable "delay switch to exclusive mode by 3 seconds" as the screen will flash as it switches to FSE. (MadVR options: Rendering > exclusive mode).


---


Quote:
Originally Posted by maxOLED View Post
@zoyd

Hi zoyd,

Thank you for your work on HCFR and this thread . I have a request for the "black frame inserstion" feature:

Could you please allow a frequency setting of 1, so that after every pattern a black frame is shown, and add a setting to define the number of ms this black frame will be shown? For which time it is shown right?

ConnecTEDDD did some test with LG 8 and showed that inserting a black frame after every pattern can reduce panel drift: https://www.avsforum.com/forum/139-di...l#post56281456
While the black frame insertion helps, I don't think it should be the only option used IMO. In my case it caused more harm than good (more detail below).
This was something I tried doing thinking it would help issues caused by the behavior that Ted was referring to in one of his comments on my first HDR calibration way back in the 2016 LG OLED calibration thread. At some point I basically gave up following typical calibration procedures due to all the unexpected behavior of my E6 that I kept finding--How controls effected the picture, how a single control behaves and how they interact with each other, auto-dimming triggers and behavior, etc.

My E6 has a red push coming out of black, and the higher it goes over 100 nits the less time it takes for image retention to effect measurements and the magnitude of difference ala drift.
I've found using a low energy grey, like 35-35-35 rgb, helps stabilize the display for more accurate measurements between ~10-45% for HDR and -30% for SDR.
I saw minimal impact outside of these ranges, and a slight improvement to the red tint of HDR near blacks 0-4% (can't fix with adjustments via HDR control point 127 being way to damn strong).
This is what I mean by causing more harm than good.

Ted made a damn nice post HERE in the 2018 LG OLED claibration thread. This is basically the behavior I was talking about and never could understand why everyone kept trying to do sweeps for measurements.

Bobof (Fabio?) did a great job combating this issue even without having an oled of his own to test with when making his patch sets. HERE Anger.miki shows his differences between black frame insertion and one of Fabio's patch sets (same 2018 OLED thread as above).

I've been "planning" to make my own patch set for my E6 but for HDR (100+ nits) this for HDR, but there's so many variables that need to be considered that I've been way to afraid to start. I know how bad my OCD is and it scares me knowing how deep this particular rabbit hole will go.

Here is a brief outline of some of my thoughts on this based on observed behavior of my own E6:


First, what's the issue? Stability over 100+ nits. The general consensus is that this makes calibrating HDR impossible (see link to Ted's post above). I don't entirely agree or disagree on this.
Calibration patterns are "synthetic" content so how the display behaves WILL be different.
What I'm getting at is a signal sent to a pixel is always going to be changing in real content far differently than it does to show a calibration pattern.
To illustrate what I mean, imagine pixels as group of people who got special instructions that all say "you are brighter than the others around you" so a big fight starts as everyone tries to follow their instructions to be brighter than everyone else.

I think there are three critical "energy" points to consider.

Before: As mentioned above, my E6 has a red push coming out of "black". It decays over ~5 seconds depending on what energy is already stored at the pixel level, which depends on the average energy of everything shown in the last ~15-20 seconds (Linear in luminance increase over time but not rgb balance due to white subpixel).

During: Because of image retention behavior, the current state of retention needs to be factored into the current measurement. Too high and results will "drift". Too low and response will be inconsistent (unstable, like my red push).
I've found best results by using a dark grey full screen in between measurements. I tried to find something neutral so all pixels would be "on". By not using black I can raise or lower a small amount of image retention to get pixel behavior that behaved closer how retention in real content would effect it. This seems to be and agrees with what some 3D LUT users were doing by pre-emptively running a patch set for a few minutes before profiling.

I've recently started playing with the Idea of using different levels of grey for different ranges I'm measuring. but only showing the brighter ones for X seconds before switching back to the dark grey then immediately taking a measurement. This seems to help stabilize consistent measurements in 200-350 nit range, and the time can be changed to simulate more or less retention.
I've been going with 5 seconds which gives green a notable boost with no real impact on blue or red.

After: You need to know how the last measurement will effect the one, all thanks to retention again.
I've been ignoring this entirely by doing manual measurements, but this is crucial to automate any of this even for 50-100 nits.

I've been trying to think of a way to visualize all of this, but Ted and BlackJoker's posts in that 2018 thread have done better than I could think of (I don't have lightspace).
But a static picture doesn't really show the impact of this behavior, IMO, as it all depends on TIME which isn't easily represented by something without time :P
Kamikaze_Ice is offline  
post #10818 of 11804 Old 06-10-2018, 04:41 AM
AVS Forum Special Member
 
ConnecTEDDD's Avatar
 
Join Date: Sep 2010
Location: Athens, Greece
Posts: 7,930
Mentioned: 187 Post(s)
Tagged: 1 Thread(s)
Quoted: 3002 Post(s)
Liked: 3698
Quote:
Originally Posted by Kamikaze_Ice View Post
PC Icon mode is the only way to get 4:4:4 chroma on the 2016 OLEDs, with a bonus of low input latency very close to Game modes without loosing access to calibration+CMS controls.
I don't have the option of using anything else as I use my E6 as my computer monitor.
Hi, while this is correct, it works only if you send 60p signal only, if you send 24p/30p/50p, you will never have full chroma, so it makes no difference having PC icon enabled/disable with that frame rates.

You can see that if you have some custom RGB balance settings using 60p signal (with PC icon active), when you will send 24p signal the custom settings will go back to default ones. When you go back to 60p signal, your custom adjustments will be there again.

Ted's LightSpace CMS Calibration Disk Free Version for Free Calibration Software: LightSpace DPS / CalMAN ColorChecker / HCFR
S/W: LightSpace CMS, SpaceMan ICC, SpaceMatch DCM, CalMAN 5, CalMAN RGB, ChromaPure, ControlCAL
V/P: eeColor 3D LUT Box - P/G: DVDO AVLab TPG
Meters: JETI Specbos 1211, Klein K-10A, i1PRO2, i1PRO, SpectraCAL C6, i1D3, C5
ConnecTEDDD is online now  
post #10819 of 11804 Old 06-10-2018, 04:50 AM
AVS Forum Special Member
 
ConnecTEDDD's Avatar
 
Join Date: Sep 2010
Location: Athens, Greece
Posts: 7,930
Mentioned: 187 Post(s)
Tagged: 1 Thread(s)
Quoted: 3002 Post(s)
Liked: 3698
Quote:
Originally Posted by webdove View Post
The two PC cards, PS3 and XBOne all behave the same. When set to RGB Full the Vertex input reports RGB Limited. I don't know whether there is no metadata for code range in the standard and the Vertex just assumes Limited or if all four of those video devices are failing to report the code range properly (which seems unlikely).
Hi, I have seen the same with Linker or Integral, from 2-3 different consumer stand-alone players I have checked, when you send RGB signal throu HD Fury, its mixing the video/pc levels (it can randomly has different issue). Even if you send RGB Video, and display works in Limited, the levels via patterns can be wrong, while to the same setup if you just remove the HD Fury, the levels can be ok. It's something you have to contact HD Fury to give you a workaround. I have seen this when HD Fury products released, long time ago, but I haven't reported since it was not affecting my setup as an issue, since the signal is always YCbCr there.

Loading a custom EDID from HD Fury it will probably fix the issue, but need some testing.

Ted's LightSpace CMS Calibration Disk Free Version for Free Calibration Software: LightSpace DPS / CalMAN ColorChecker / HCFR
S/W: LightSpace CMS, SpaceMan ICC, SpaceMatch DCM, CalMAN 5, CalMAN RGB, ChromaPure, ControlCAL
V/P: eeColor 3D LUT Box - P/G: DVDO AVLab TPG
Meters: JETI Specbos 1211, Klein K-10A, i1PRO2, i1PRO, SpectraCAL C6, i1D3, C5
ConnecTEDDD is online now  
post #10820 of 11804 Old 06-10-2018, 01:24 PM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by Dominic Chan View Post
It may be more logical to implement the “high resolution” input range as part of Near Black Grey Scale.
The problem that I have with the near-black window is that I lose the ability to view "RGB-levels" "w/gamma" which is my preferred adjustment view.
Therefore I prefer to stay in the gray scale mode with small steps and adjust my RGB-levels window scale to look at near black.
Also this saves me having to mode switch between grayscale and near-black while adjusting.

My strongest desire is to see a vertical scale option in RGB levels w/gamma that weights each horizontal position with the Barten visual sensitivity curve rather than just uniform percentage from 0 to 100% brightness. If that were done then the errors across the window would all be equally perceptually weighted where as at the moment the low brightness errors are depicted much too large and the high brightness errors much too small.
webdove is offline  
post #10821 of 11804 Old 06-10-2018, 01:31 PM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by webdove View Post
The problem that I have with the near-black window is that I lose the ability to view "RGB-levels" "w/gamma" which is my preferred adjustment view.
I don’t use the Near Black view to adjust gamma as my projector doesn’t not have any adjustments below 5%.
Dominic Chan is online now  
post #10822 of 11804 Old 06-10-2018, 04:37 PM
AVS Forum Special Member
 
Stuntman_Mike's Avatar
 
Join Date: Feb 2009
Location: NYC
Posts: 1,639
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 690 Post(s)
Liked: 526
I'm going to do an HDR10 calibration on my OLED E6P, using HCFR and the Mehanik HDR10 test patterns. The LG OLEDs use codes instead of percentages for window patterns in their 20pt controls. The Mehanik patterns have 2016 OLED patterns that are for HCFR they say.

My question is, does HCFR step through using the codes when doing a grayscale run or the normal percentages? What do I need to choose to get HCFR do use the codes instead, if that's possible?

Stuntman_Mike is online now  
post #10823 of 11804 Old 06-10-2018, 05:22 PM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 148 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Quote:
Originally Posted by stama View Post
@zoyd

I have a feature request and a bug report, which would help with HDR calibration:

- feature request: enter HDR mode when starting the measurements, and exit HDR mode when finishing the measurements.

At the moment, the HDR mode is exited not when the measurements stop, but when closing HCFR. The HCFR user interface is burned out and reading what's written on it is almost impossible while HDR mode is on. And if you have an OLED, then you also get image retention in the rectangle where the HCFR window is.
I will look into adding this option for people using a single monitor to HCFR and calibrate. Note that DV mode requires 2 screens (1 for HCFR and 1 connected as external monitor to a PC).


Quote:
Originally Posted by maxOLED View Post
@zoyd

Hi zoyd,

Thank you for your work on HCFR and this thread . I have a request for the "black frame inserstion" feature:

Could you please allow a frequency setting of 1, so that after every pattern a black frame is shown, and add a setting to define the number of ms this black frame will be shown? For which time it is shown right?
yes, this is easy to add.

Quote:
Originally Posted by webdove View Post
I have tested an Nvidia GTX 1050, AMD HD5450, Playstation 3 and XBOne. In all cases when set to either RGB8 Full or RGB8 Limited, the Vertex reports them as RGB8 Limited.

If you then set the vertex scaler to output RGB8 Full rather than follow then the vertex translates the incoming 16-235 to outgoing 0-255.

I still have yet to get a visual image from my LG on the extended second video output in RGB Full mode on the Nvidia when I check DV under vertex in references.
I see the Dolby Vision icon, but the screen is black.
I have found an issue with getting the pattern levels to display correctly in DV mode. I'll fix this and send out a new test version for you.

Quote:
Originally Posted by Stuntman_Mike View Post
I'm going to do an HDR10 calibration on my OLED E6P, using HCFR and the Mehanik HDR10 test patterns. The LG OLEDs use codes instead of percentages for window patterns in their 20pt controls. The Mehanik patterns have 2016 OLED patterns that are for HCFR they say.

My question is, does HCFR step through using the codes when doing a grayscale run or the normal percentages? What do I need to choose to get HCFR do use the codes instead, if that's possible?
The 2016 and 2017 LG OLED levels are available as selections in the color checker dropdown, greyscale page is currently limited to up to 100 evenly space levels. Looks like I need to add some options to that for HDR calibration.
Stuntman_Mike likes this.
zoyd is offline  
post #10824 of 11804 Old 06-10-2018, 07:20 PM
AVS Forum Special Member
 
Stuntman_Mike's Avatar
 
Join Date: Feb 2009
Location: NYC
Posts: 1,639
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 690 Post(s)
Liked: 526
Quote:
Originally Posted by zoyd View Post
The 2016 and 2017 LG OLED levels are available as selections in the color checker dropdown, greyscale page is currently limited to up to 100 evenly space levels. Looks like I need to add some options to that for HDR calibration.
Very, very much appreciated.

Stuntman_Mike is online now  
post #10825 of 11804 Old 06-11-2018, 08:04 AM - Thread Starter
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 148 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
3.5.1.7 is available. Hopefully corrects issue with LG DV patterns, it's been tested with video card resolution set to 1080p so that there is no scaling done on the Vertex.

I also added duration and level settings to the black frame insertion and allow for insertion after every pattern (frequence = 1).
zoyd is offline  
post #10826 of 11804 Old 06-11-2018, 12:52 PM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by zoyd View Post
3.5.1.7 is available. Hopefully corrects issue with LG DV patterns, it's been tested with video card resolution set to 1080p so that there is no scaling done on the Vertex.

I also added duration and level settings to the black frame insertion and allow for insertion after every pattern (frequence = 1).
Thanks much. I will test. BTW, were you able to find out why 3.5.1.6 generated instrument errors when you opened a second window? It was as though in that version it was attempting to reopen the device, rather than using an existing already opened device descriptor for the second or subsequent window. I find that having multiple windows to the same device is helpful when doing adjustment comparisons.
webdove is offline  
post #10827 of 11804 Old 06-11-2018, 01:09 PM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by ConnecTEDDD View Post
Hi, I have seen the same with Linker or Integral, from 2-3 different consumer stand-alone players I have checked, when you send RGB signal throu HD Fury, its mixing the video/pc levels (it can randomly has different issue). Even if you send RGB Video, and display works in Limited, the levels via patterns can be wrong, while to the same setup if you just remove the HD Fury, the levels can be ok. It's something you have to contact HD Fury to give you a workaround. I have seen this when HD Fury products released, long time ago, but I haven't reported since it was not affecting my setup as an issue, since the signal is always YCbCr there.

Loading a custom EDID from HD Fury it will probably fix the issue, but need some testing.
I have been emailing with Ken Taurus at HDFury about this today and reading up on the CTA-861-F signalling spec.

In the display to source EDID Video Capability Data Block the display can specify whether either or both RGB and YCbCr have selectable range.
In the source to display HDMI InfoFrame bits Q0 and Q1 can be set to three codes: default, limited-range, full-range.
For CE Video Format timings (nearly all the formats I use except 640x480) sending "default" means the display should use limited-range.

He said that the PC, XBox and PS3 are all sending "default" regardless of what output format setting I am choosing.
This is a mistake for the XBox and PS3 since when they are set to full-range that guarantees that the display will misinterpret the data and crush blacks.

The right design is less obvious on the PC, since I may set the driver to Full-range and then HCFR or madvr to limited in which case I don't want the display interpreting full range. Alternatively I may be trying to test whiter-than-white or blacker-than-black. In which case again I don't want the display interpreting full-range. Sadly the PC does not let me independently select the transmitted code range and then separately select the Q0,Q1 signal to the display.

He said the separate generators always are explicit about the Q0,Q1 range signal to the display so they don't have this problem.

Therefore the behavior that I observed of the Vertex always reporting limited is correct since I was always using a CE Video Format Timing picture mode which defaults to limited-range.

To fix this Ken offered to make a firmware change that (I assume) will let me override the default-range coming from the PC to be either full or limited without remapping the video brightness codes. I will followup as I learn more.
ConnecTEDDD likes this.
webdove is offline  
post #10828 of 11804 Old 06-11-2018, 02:00 PM
Member
 
webdove's Avatar
 
Join Date: Jul 2006
Location: NH
Posts: 59
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 6
Quote:
Originally Posted by webdove View Post
Thanks much. I will test. BTW, were you able to find out why 3.5.1.6 generated instrument errors when you opened a second window? It was as though in that version it was attempting to reopen the device, rather than using an existing already opened device descriptor for the second or subsequent window. I find that having multiple windows to the same device is helpful when doing adjustment comparisons.
Zoyd, 3.5.1.7 still shows a black screen with the dolby icon.
webdove is offline  
post #10829 of 11804 Old 06-11-2018, 02:00 PM
Newbie
 
Join Date: Jul 2014
Posts: 8
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 0
Question i1D3 + HCFR 3.5.1.4

Dear fellow movie fans

I purchased an LG C8 65" TV, and I feel it's worth calibrating this beauty...

But I struggle to get my i1D3 to work with the latest HCFR 3.5.1.4. I'm running windows 10 (through parallels) and whenever I connect the calibrator it automatically installs standard drivers. In device manager trying to point to folder "HCFR Calibration-Drivers-ArgyllCMS" it won't accept it. I've tried rebooting and disabled driver signing.

Does anyone know how I can get this working? I'm reluctant of paying 300$+ for a commercial application like Calman.

Thanks
eidos is offline  
post #10830 of 11804 Old 06-11-2018, 02:17 PM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,644
Mentioned: 126 Post(s)
Tagged: 0 Thread(s)
Quoted: 4652 Post(s)
Liked: 1725
Quote:
Originally Posted by eidos View Post
But I struggle to get my i1D3 to work with the latest HCFR 3.5.1.4. I'm running windows 10 (through parallels) and whenever I connect the calibrator it automatically installs standard drivers. In device manager trying to point to folder "HCFR Calibration-Drivers-ArgyllCMS" it won't accept it. I've tried rebooting and disabled driver signing.
You do not need to install tye ArgyllCMS driver for the i1D3, unlike the i1Pro. It uses one of the Human Interface Devices drivers that come standard with Windows.
Dominic Chan is online now  
Sponsored Links
Advertisement
 
Reply Display Calibration

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


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