MadVR - ArgyllCMS - Page 71 - AVS Forum
Forum Jump: 
 42Likes
Reply
 
Thread Tools
post #2101 of 2671 Old 06-10-2014, 02:47 AM
Member
 
Francois76l's Avatar
 
Join Date: May 2014
Location: FRANCE
Posts: 80
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 10

Hello all!

 

Yesterday, I tried a calibration of my plasma (Pana ST50). Amazing!

Before: https://drive.google.com/file/d/0Bwd7JbAM5r1KYXlROTN5emRzaEVhMjFPcWhIaElidDBLcTJR/edit?usp=sharing
After: https://drive.google.com/file/d/0Bwd7JbAM5r1KdUFnd0NzbDJ2Qjh4bkVCMXUtempkUjg4MFJz/edit?usp=sharing

 

Before:

 

After:

 

But there is something I don't understand...

My goal was to reach a perfect linear 2.4 gamma with BT1886 correction.

Before calibration, It wasn't so bad:

 

But after calibration I've this:

 

My setup was:

MadTPG: 8% windows with 25% gamma light background

DispcalGUI; CCSS for plasma colorimeter correction, no blackpoint correction, BT1886 Absolute 2.4 all the rest as on the first page.

any idea? I really want to have a 2.4 gamma on the plasma, It should be fantastic!

 

After this calibration and a lot of comparaison between my plasma and my JVC (plasma win of course....) I tried a new calibration (at midnight!) of the JVC after tuning my whitepoint in the service menu. Unfortunately, I got tired and wake up this morning  at 7am! (poor projector :) ). I will show my results tonight.


HiFi:Nad C320BEE+Marantz CD-63 MkII-Ki Signature+Boston VR2+DIY cabling
HC:Onkyo TX-SR875+Boston VR2+CR8+Chorus cc800v+Subwoofer MJ Acoustic Ref1 MKII
DISPLAYS:Panasonic P50ST50E/JVC DLA-X35+110" gain 0.8+HTPC (LAV+ffdshow+Avisynth+MPC-BE+MadVr)
Francois76l is offline  
Sponsored Links
Advertisement
 
post #2102 of 2671 Old 06-10-2014, 05:00 AM
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,568
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 149 Post(s)
Liked: 330
Quote:
Originally Posted by Francois76l View Post

But there is something I don't understand...
My goal was to reach a perfect linear 2.4 gamma with BT1886 correction.

BT.1886 will only be perfectly linear if you have a 0.0 black measurement. Even with very low blacks you can get the curve you displayed. You can try -Ib:2.4 when creating the LUT (recommended), and that entire curve will be shifted up so the 50% point crosses 2.4. This will also prevent any residual near black compression. Or you can get it perfectly linear by unchecking the BT.1886 option and using the source space Rec709_gamma24.icm but you may lose a little bit of shadow detail.

-----

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
zoyd is online now  
post #2103 of 2671 Old 06-10-2014, 05:15 AM
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,568
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 149 Post(s)
Liked: 330
Quote:
Originally Posted by gwgill View Post


If some displays clip on a per component basis rather than displaying WTW or clipping in a way that preserves hue, then yes, feeding color managed WTW values to the display is probably a bad thing. But that's not what I was told was the situation. Changing it so that the 3DLut clips the input value is an obvoius alternate to the current behavior, but this is certainly not passing WTW through.

Displays won't do this intentionally but it's often the case that clipping will occur in one channel as it "runs out of gas" before the other channels depending on how far the contrast is pushed. If the display allows WTW it's usually recommended to set contrast so that this channel does not clip well into the WTW region and then white balance the other channels to the weak one. Perhaps those that are seeing color shifts in WTW could check to see if one of the color channels is clipping early.

-----

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
zoyd is online now  
post #2104 of 2671 Old 06-10-2014, 05:28 AM
AVS Special Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 9,958
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 530 Post(s)
Liked: 549
Quote:
Originally Posted by gwgill View Post

So the CRT selection isn't suitable ?
I didn't think so. Isn't it for a direct view CRT with colored phosphors in a shadow mask?

Does projection CRT use colored phosphors, colored lenses, or both? I thought it used colored lenses in front of a white CRT, but I can't seem to find any confirmation either way. Regardless, wouldn't the CRT selection only be suitable if the RP-CRT has a similar spectral response? If the phosphors are different, or the lenses have a different wavelength, I would expect the CRT setting is not valid. I don't have a spectroradiometer, but the RGB primaries for my set are well outside the REC.709 triangle for Red and Green.
Stereodude is offline  
post #2105 of 2671 Old 06-10-2014, 06:01 AM
Member
 
Francois76l's Avatar
 
Join Date: May 2014
Location: FRANCE
Posts: 80
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 10
Quote:
Originally Posted by zoyd View Post


BT.1886 will only be perfectly linear if you have a 0.0 black measurement. Even with very low blacks you can get the curve you displayed. You can try -Ib:2.4 when creating the LUT (recommended), and that entire curve will be shifted up so the 50% point crosses 2.4. This will also prevent any residual near black compression. Or you can get it perfectly linear by unchecking the BT.1886 option and using the source space Rec709_gamma24.icm but you may lose a little bit of shadow detail.

Thanks zoyd for your help!

Can I use the arg "-Ib:2.4" directly in dispcalGUI in the window "Create 3DLUT" or am I suppose to do it via command line?

And we have to use this tweak only if the linear gamma is not reachable despite a good black level? Because on my JVC projector, I obtained a linear 2.4 gamma.


HiFi:Nad C320BEE+Marantz CD-63 MkII-Ki Signature+Boston VR2+DIY cabling
HC:Onkyo TX-SR875+Boston VR2+CR8+Chorus cc800v+Subwoofer MJ Acoustic Ref1 MKII
DISPLAYS:Panasonic P50ST50E/JVC DLA-X35+110" gain 0.8+HTPC (LAV+ffdshow+Avisynth+MPC-BE+MadVr)
Francois76l is offline  
post #2106 of 2671 Old 06-10-2014, 06:07 AM
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,568
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 149 Post(s)
Liked: 330
Quote:
Originally Posted by Francois76l View Post

Thanks zoyd for your help!
Can I use the arg "-Ib:2.4" directly in dispcalGUI in the window "Create 3DLUT" or am I suppose to do it via command line?

2.4 Relative on the 3DLUT page is what you want.
Quote:
And we have to use this tweak only if the linear gamma is not reachable despite a good black level? Because on my JVC projector, I obtained a linear 2.4 gamma.

I believe that the next release of ArgyllCMS will include a linking switch for linear transfer function targets with an input offset. Until then you have to use rec709_gammaXX.icm source spaces.

-----

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
zoyd is online now  
post #2107 of 2671 Old 06-10-2014, 06:20 AM
Member
 
Francois76l's Avatar
 
Join Date: May 2014
Location: FRANCE
Posts: 80
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 10

You mean here, on point 5?

Because -Ib:2.4  apply 2.4 BT.1886 curve effective gamma. But there is no precision if it's absolute or relative.

I'll check that after work.

 

Concerning my projector and my linear 2.4 gamma, you suggest to make another calibration with rec709_gamma24.icm source spaces?

 

Thanks again for your help...


HiFi:Nad C320BEE+Marantz CD-63 MkII-Ki Signature+Boston VR2+DIY cabling
HC:Onkyo TX-SR875+Boston VR2+CR8+Chorus cc800v+Subwoofer MJ Acoustic Ref1 MKII
DISPLAYS:Panasonic P50ST50E/JVC DLA-X35+110" gain 0.8+HTPC (LAV+ffdshow+Avisynth+MPC-BE+MadVr)
Francois76l is offline  
post #2108 of 2671 Old 06-10-2014, 06:27 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by madshi View Post

I suppose some displays might not clip until the very last processing stage. Which means that feeding them WTW/BTB might help get some more headroom in earlier processing stages (e.g. converting to YCbCr, deinterlacing, scaling, motion interpolation etc etc).
I strongly suspect that WTW display is simply incompatible with profiling. If you assume that white equates to R=G=B, then it doesn't matter if the display clips at some level above 235, as it will (presumably) clip evenly and white will remain white. But once profiled, white in general is R!=G!=B, so display clipping will always result in a tinted WTW.
So I don't see any point in allowing for this inherently flawed idea.
gwgill is offline  
post #2109 of 2671 Old 06-10-2014, 06:39 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by zoyd View Post

I believe that the next release of ArgyllCMS will include a linking switch for linear transfer function targets with an input offset. Until then you have to use rec709_gammaXX.icm source spaces.
I'm still contemplating the details of this, or if it is to be included in the next release at all. I've already had the sort of feedback I feared in relation to this feature, that of it being misused for displays with a noticeably non-black black point, and the resulting curve being not so smooth.

One option is to replace it with an output offset model. Another is to add another parameter to be able to choose something between input (BT.1886) and output offset, just like dispcal. I'm not sure people would know know what to do with that though - merely having BT.1886 with a variable power, never mind effective vs. absolute seems to be too much for many people to grasp.
gwgill is offline  
post #2110 of 2671 Old 06-10-2014, 06:45 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by Stereodude View Post

Does projection CRT use colored phosphors, colored lenses, or both? I thought it used colored lenses in front of a white CRT, but I can't seem to find any confirmation either way.
From an engineering point of view, I'd be really surprised if it wasn't colored phosphor - using white and filters would be very inefficient, throwing away 2/3 of the energy output of each tube, and CRT projectors need all the brightness they can get. Whether the phosphor is similar to a direct view CRT, I don't know.
Quote:
If the phosphors are different, or the lenses have a different wavelength, I would expect the CRT setting is not valid.
Glass typically has a fairly flat spectral characteristic compared to everything else that could affect things.
Quote:
I don't have a spectroradiometer, but the RGB primaries for my set are well outside the REC.709 triangle for Red and Green.
That is a hint that the phosphors may well be different from a typical sRGB/Rec709 like direct view CRT.
gwgill is offline  
post #2111 of 2671 Old 06-10-2014, 07:50 AM
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 4,568
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 149 Post(s)
Liked: 330
Quote:
Originally Posted by gwgill View Post

I'm still contemplating the details of this, or if it is to be included in the next release at all. I've already had the sort of feedback I feared in relation to this feature, that of it being misused for displays with a noticeably non-black black point, and the resulting curve being not so smooth.

One option is to replace it with an output offset model. Another is to add another parameter to be able to choose something between input (BT.1886) and output offset, just like dispcal. I'm not sure people would know know what to do with that though - merely having BT.1886 with a variable power, never mind effective vs. absolute seems to be too much for many people to grasp.

My personal opinion is that a pure power law option is a desirable one given the right display characteristics and any misuse of it (and resulting complaints) is on the user. I wouldn't include any additional offset options unless you feel there is something that is better than BT.1886 for certain situations.

-----

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
zoyd is online now  
post #2112 of 2671 Old 06-10-2014, 09:15 AM
Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 197
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 40 Post(s)
Liked: 34
Quote:
Originally Posted by Francois76l View Post

Because -Ib:2.4  apply 2.4 BT.1886 curve effective gamma. But there is no precision if it's absolute or relative.
Just to confirm, the equivalent of -Ib:2.4 (BT.1886 with effictive gamma of 2.4 at 50%) is "BT.1886 2.4 relative" in the GUI.
fhoech is online now  
post #2113 of 2671 Old 06-10-2014, 05:16 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by zoyd View Post

My personal opinion is that a pure power law option is a desirable one given the right display characteristics and any misuse of it (and resulting complaints) is on the user.
I suspect that few people are actually using a pure power law though, instead ending up with output offset curves. The collink -I G:x.x beta code is an absolute power law, but the question remains of what sort of curve to use to transition from that curve to the actual black level. On a theoretical basis, it's hard to find a justification for targeting an absolute power curve when the black level is not close to zero, and the question of the transition shape turns into the same question of what curve to use for a non-zero black level, the only reasoned answer that I'm aware of being input offset/BT.1886. If there were some understanding about why BT.1886 doesn't work in all cases, then it might be possible to propose a curve shape that addresses those concerns.
gwgill is offline  
post #2114 of 2671 Old 06-10-2014, 06:16 PM
Advanced Member
 
spacediver's Avatar
 
Join Date: May 2013
Location: Toronto
Posts: 840
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 71 Post(s)
Liked: 70
Graeme, I've been going through the dispcal documentation, in preparation for my next calibration run, and am having problems understanding the relationship between -J and -I

I'm on an FW900 with a great black level (<0.01 nits), and am using a DTP-94.

Typically, when working with the instrument, I periodically do a black point calibration through the device itself (when it prompts you to place it on a dark surface). I do this every ten minutes, but according to your documentation, this is probably only necessary once in a while. So my plan is to let the device warm up for a couple hours, and then do a single black point calibration using the -J switch. Would you say that this is a safe bet with the DTP-94?

As for the -I b switch, I'm guessing that once the instrument has had its black point calibration done, Argyll can measure the luminance of a black patch, and use that as a reference (assuming display doesn't drift during the run). But if indeed it's only necessary to do the black point calibration once in a while (see above paragraph), then would there be any need to do a periodic black test patch to compensate for drift? Or is this a different type of drift than in the above paragraph?

Also, does -q u automatically take care of -I bw?
spacediver is online now  
post #2115 of 2671 Old 06-10-2014, 07:46 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by spacediver View Post

Graeme, I've been going through the dispcal documentation, in preparation for my next calibration run, and am having problems understanding the relationship between -J and -I

I'm on an FW900 with a great black level (<0.01 nits), and am using a DTP-94.

Typically, when working with the instrument, I periodically do a black point calibration through the device itself (when it prompts you to place it on a dark surface). I do this every ten minutes, but according to your documentation, this is probably only necessary once in a while.
This depends on the type of instrument. Something like a i1pro Rev A-D black reference is quite sensitive to the temperature. Doing a black cal every 5 or 10 minutes is not very convenient, and can introduce positioning errors. The -Ib switch is an attempt to address this issue.

Generally the colorimeters use either temperature compensated L2F sensors, or they have a thermometer in them and apply temperature compensation to their black point in the firmware, hence the lesser need for black calibration. The ColorMunki spectro and i1pro Rev E both have shielded cells that allow compensating the black reference, but it's still advisable to let them acclimatise if the highest accuracy is desired, since the compensation isn't perfect for large temperature variations.
Quote:
So my plan is to let the device warm up for a couple hours, and then do a single black point calibration using the -J switch. Would you say that this is a safe bet with the DTP-94?
It's worth trying. A sanity check is to use spotread to measure it's black point when it's cold (ie. against a black surface), let it aclimatise and then repeat the measurement. This should give you an idea as to how well temperature compensated it is.
Quote:
As for the -I b switch, I'm guessing that once the instrument has had its black point calibration done, Argyll can measure the luminance of a black patch, and use that as a reference (assuming display doesn't drift during the run).
Right. When using -I b the assumption is that the display black is more stable than the instrument black.
Quote:
Also, does -q u automatically take care of -I bw?
No, they are unrelated. -q is mainly about number of measurements and precision target.

dispcal may well be unnecessary for a well behaved TV, as long as the native white point is near the profiling target.
gwgill is offline  
post #2116 of 2671 Old 06-10-2014, 08:11 PM
Advanced Member
 
spacediver's Avatar
 
Join Date: May 2013
Location: Toronto
Posts: 840
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 71 Post(s)
Liked: 70
Quote:
Originally Posted by gwgill View Post

This depends on the type of instrument. Something like a i1pro Rev A-D black reference is quite sensitive to the temperature. Doing a black cal every 5 or 10 minutes is not very convenient, and can introduce positioning errors. The -Ib switch is an attempt to address this issue.

Makes sense - I've actually just finished creating a "stencil" pattern for my DTP-94 that allows me to line up the instrument so that the sensor is almost perfectly centered on my screen. Good to know I might not need it as much as I thought I would.
Quote:
Originally Posted by gwgill View Post

It's worth trying. A sanity check is to use spotread to measure it's black point when it's cold (ie. against a black surface), let it aclimatise and then repeat the measurement. This should give you an idea as to how well temperature compensated it is.

aye, good idea, I was planning on doing something like this, taking alternating black and white readings over a long period of time and measuring drift. If it turns out that the instrument is stable, I'm guessing it would be ok to skip the -I bw switch, right?

Quote:
Originally Posted by gwgill View Post

dispcal may well be unnecessary for a well behaved TV, as long as the native white point is near the profiling target.

I'm using dispcal mainly for gamma tuning. After doing a hardware white point balance through the service port of my FW900 CRT, the grayscale tracks almost flawlessly, but the gamma is way too high (2.8 or so) when I turn down the G2 voltage to inky black levels - that's where your software comes in!

thanks for the detailed reply.
spacediver is online now  
post #2117 of 2671 Old 06-10-2014, 10:21 PM
AVS Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 5,471
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 61 Post(s)
Liked: 129
Quote:
Originally Posted by gwgill View Post

I strongly suspect that WTW display is simply incompatible with profiling. If you assume that white equates to R=G=B, then it doesn't matter if the display clips at some level above 235, as it will (presumably) clip evenly and white will remain white. But once profiled, white in general is R!=G!=B, so display clipping will always result in a tinted WTW.
So I don't see any point in allowing for this inherently flawed idea.

Well, I don't really have a good WTW solution in mind for profiling. So I guess maybe it's the easiest and safest way to simply clip the input. The idea to always throw away data does hurt the perfectionist in me a bit, though...
madshi is offline  
post #2118 of 2671 Old 06-11-2014, 12:19 AM
Senior Member
 
Audionut11's Avatar
 
Join Date: May 2008
Posts: 354
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 77 Post(s)
Liked: 52
Quote:
Originally Posted by gwgill View Post

Whether they are exactly consistent is somewhat irrelevant - profiling trumps all that. I would imagine though, that they are different because they are aiming for slightly different things.
The display calibration step is somewhat limited, because it is working with the limitations of typical computer displays. The biggest gain would probably be from making sure the display white point is near the target. Per channel calibration curves may help the accuracy of the final 3DLut, but if the display response curve and grey balance is good to start with, is probably unnecessary.

Thanks Graeme. I have a display with pretty good 20 point grey scale adjustment which results in a much better response curve.







I can only assume the calibration will be better with the bottom results. I certainly see better consistency in measurement results, when the display is very close to grey. The calibration process is quicker also.


I am curious about the results after calibration.
Measurement report.

Can someone provide some pointers as to why the gamma response drops at lower luminance levels? I assume this is black point compensation.
Also, why the color temperature rises at lower luminance levels. I assume it will be better to adjust the display pre-calibration, to better match the curves in the measurement report.

edit: That's rather an open ended question. I'll research some more myself.
Audionut11 is offline  
post #2119 of 2671 Old 06-11-2014, 09:15 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by Audionut11 View Post
Can someone provide some pointers as to why the gamma response drops at lower luminance levels? I assume this is black point compensation.
You have a choice when it comes to black. You can either try and correct it's color and suffer from a raised black level, or you can leave it alone for maximum contrast ratio, but it's color remains unchanged.

Log/Log values don't make any sense at zero, and vary much more with slight variations in the measured numbers as you approach 0 or 1. They are not a good way to judge the transfer curve accuracy, since there is no correlation with visual result.
gwgill is offline  
post #2120 of 2671 Old 06-11-2014, 10:34 PM
Senior Member
 
Audionut11's Avatar
 
Join Date: May 2008
Posts: 354
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 77 Post(s)
Liked: 52
It appears to be a bug.

Toggling BPC, adjusting Black output offset, and toggling the gamma tone curve between relative and absolute, produced no changes in the gamma curve.

I then adjusted the gamma to 1.8 absolute, with produced a nice flat curve @ 1.8. I then adjusted back to 2.4 absolute, which produced a nice flat curve @ 2.4. After adjusting some black level settings (sorry I don't recall which ones exactly), it then proceeds to produce a gamma curve like the one in my measurement report again. And again, this curve cannot be adjusted back to a flat state, without adjusting the gamma value.


Quote:
Originally Posted by gwgill View Post
You have a choice when it comes to black. You can either try and correct it's color and suffer from a raised black level, or you can leave it alone for maximum contrast ratio, but it's color remains unchanged.
A nice option would be one that works based on a user defined black level. Currently, unless I am missing something, there is no way to determine the effect on the black level. So for instance, you can adjust the black point correction, but how far exactly will that raise the black level, depending on the settings chosen.

I run at a reduced white level to ensure I have a low black level (I hate grey blacks, more then I like bright whites), so an black point correction based on the defined value I enter, would speed up the processing, rather then having to guess how far my blacks are going to rise.

ie: User sets black level to some defined value (one above the measured black level), and the black point correction does what it can to even the RGB levels, without raising the black level above this defined value.

Last edited by Audionut11; 06-11-2014 at 11:03 PM.
Audionut11 is offline  
post #2121 of 2671 Old 06-12-2014, 12:08 AM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
>It appears to be a bug.

>Toggling BPC, adjusting Black output offset, and toggling the gamma tone curve between relative and absolute, produced no changes in the gamma curve.

There is no such thing as BPC in ArgyllCMS, and I'm not sure what you mean by Black output offset - are you referring to colprof or collink ? What option ?

> I then adjusted the gamma to 1.8 absolute, with produced a nice flat curve @ 1.8. I then adjusted back to 2.4 absolute, which produced a nice flat curve @ 2.4. After adjusting some black level settings (sorry I don't recall which ones exactly), it then proceeds to produce a gamma curve like the one in my measurement report again. And again, this curve cannot be adjusted back to a flat state, without adjusting the gamma value.

I don't know what you mean by "adjust gamma level". collink provides BT.1886 power value override. Is that what you mean ?

> A nice option would be one that works based on a user defined black level.

Definitly not. The point of using an instrument is not to have the user invent color values. The point is to measure what the display is actually doing.

> ie: User sets black level to some defined value (one above the measured black level), and the black point correction does what it can to even the RGB levels, without raising the black level above this defined value.

That's not what people say they want - they want the darkest possible black.

[ Oh - and currently AVS really sucks. The font sizes are all over the place making it difficult to read, the navigation to threads posted to seems to be missing, jumping to threads dumps you at the start of the thread instead of where you last left off, and the message editor is complete broken. It's using a tiny font, quoting doesn't work, there seems no way of inserting images in line, it keeps sucking up any formatting I put in like lines or spaces, there's no preview for editing an existing message, previous in-line images are huuuge, and that's just what I've noticed after trying to compose 2 messaged. But it's got lots of smilies ! It seems a hasty and poorly tested change. I really don't feel like corresponding using AVS in its current form. ]

Last edited by gwgill; 06-12-2014 at 12:21 AM.
gwgill is offline  
post #2122 of 2671 Old 06-12-2014, 12:26 AM
Senior Member
 
Audionut11's Avatar
 
Join Date: May 2008
Posts: 354
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 77 Post(s)
Liked: 52
Quote:
Originally Posted by gwgill View Post
There is no such thing as BPC in ArgyllCMS, and I'm not sure what you mean by Black output offset - are you referring to colprof or collink ? What option ?.....................I don't know what you mean by "adjust gamma level". collink provides BT.1886 power value override. Is that what you mean ?
I describe the options in dispcalGUI. My apologies for not making this more obvious.


Quote:
Originally Posted by gwgill View Post
Definitly not. The point of using an instrument is not to have the user invent color values. The point is to measure what the display is actually doing.
Fair enough.


Quote:
Originally Posted by gwgill View Post
That's not what people say they want - they want the darkest possible black.
As do I. Is there a way to determine what the black level will be, after any black level correction for hue? This is my point. As far as I know, there is no way to determine this, until after the profiling.

So rather then the current functionality that says, "correct hue by this amount, and deal with the increased black level", I would like to see functionality that says, "hey, try and correct hue as you see fit, but do not increase the black level above this level I have set, I'll deal with any remaining hue issues".

Last edited by Audionut11; 06-12-2014 at 12:32 AM.
Audionut11 is offline  
post #2123 of 2671 Old 06-12-2014, 12:29 AM
Senior Member
 
Audionut11's Avatar
 
Join Date: May 2008
Posts: 354
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 77 Post(s)
Liked: 52
Quote:
Originally Posted by gwgill View Post
[ Oh - and currently AVS really sucks. The font sizes are all over the place making it difficult to read, the navigation to threads posted to seems to be missing, jumping to threads dumps you at the start of the thread instead of where you last left off, and the message editor is complete broken. It's using a tiny font, quoting doesn't work, there seems no way of inserting images in line, it keeps sucking up any formatting I put in like lines or spaces, there's no preview for editing an existing message, previous in-line images are huuuge, and that's just what I've noticed after trying to compose 2 messaged. But it's got lots of smilies ! It seems a hasty and poorly tested change. I really don't feel like corresponding using AVS in its current form. ]

At the top of the forums, select TOOLS, User CP, Edit Options (on the left), scroll to the bottom of that page and change "Message Editor Interface" to standard. This make it somewhat better.
Audionut11 is offline  
post #2124 of 2671 Old 06-12-2014, 10:33 AM
Member
 
Icaro's Avatar
 
Join Date: Jul 2007
Location: Italy
Posts: 164
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 15
Quote:
Originally Posted by Audionut11 View Post
It appears to be a bug.

Toggling BPC, adjusting Black output offset, and toggling the gamma tone curve between relative and absolute, produced no changes in the gamma curve.

I then adjusted the gamma to 1.8 absolute, with produced a nice flat curve @ 1.8. I then adjusted back to 2.4 absolute, which produced a nice flat curve @ 2.4. After adjusting some black level settings (sorry I don't recall which ones exactly), it then proceeds to produce a gamma curve like the one in my measurement report again. And again, this curve cannot be adjusted back to a flat state, without adjusting the gamma value.




A nice option would be one that works based on a user defined black level. Currently, unless I am missing something, there is no way to determine the effect on the black level. So for instance, you can adjust the black point correction, but how far exactly will that raise the black level, depending on the settings chosen.

I run at a reduced white level to ensure I have a low black level (I hate grey blacks, more then I like bright whites), so an black point correction based on the defined value I enter, would speed up the processing, rather then having to guess how far my blacks are going to rise.

ie: User sets black level to some defined value (one above the measured black level), and the black point correction does what it can to even the RGB levels, without raising the black level above this defined value.
In my opinion,
if you want to correct the Black point pre-calibration,
with our "domestic" probes,
the best way, is to act only lowering the control of luminance by Grey\gamma, and not via the individual components R,G,B or Hue.

Sorry for my bad English

Last edited by Icaro; 06-13-2014 at 02:27 AM.
Icaro is offline  
post #2125 of 2671 Old 06-12-2014, 05:41 PM
Advanced Member
 
spacediver's Avatar
 
Join Date: May 2013
Location: Toronto
Posts: 840
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 71 Post(s)
Liked: 70
A couple nights ago, I factory calibrated my FW900 CRT and brought the G2 voltage down very low to produce extremely inky blacks. As a result, the gamma was really high, so I used dispcal to generate a custom 1D LUT (instrument is a DTP-94).

The top image shows what things looked like before dispcal. Three main things:

1: Black level is lower than the instrument's ability to read.

2: Gamma is extremely high.

3: White balance is already excellent after hardware calibration.

Here is the string I used:

dispcal -dmadvr -m -Yp -qu -J -Ib -v3 -g2.4 -f0 -k0 -A16 name.cal

The bottom image shows the results of this LUT. The gamma is improved, but it still seems too high. Could this have something to do with the fact that the black level is too low for the instrument to read?

Either way, can you offer any suggestions on how to proceed?

and goddamn this new forum interface - seems we can no longer insert images directly and choose preview sizes.

Before Dispcal




After Dispcal

spacediver is online now  
post #2126 of 2671 Old 06-12-2014, 06:12 PM
Advanced Member
 
gwgill's Avatar
 
Join Date: Jan 2013
Location: Melbourne, Australia
Posts: 631
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 76 Post(s)
Liked: 79
Quote:
Originally Posted by Audionut11 View Post
At the top of the forums, select TOOLS, User CP, Edit Options (on the left), scroll to the bottom of that page and change "Message Editor Interface" to standard.
That seems to make little difference. It is simply not honoring the markup, as you can see in this post.
Quote:
> I describe the options in dispcalGUI. My apologies for not making this more obvious.
I'm afraid I you're on your own then. These options are not ones I've created or know the details of, and are not things I've tested or recommend for creating Video calibration 3DLuts.
Quote:
> Is there a way to determine what the black level will be, after any black level correction for hue? This is my point. As far as I know, there is no way to determine this, until after the profiling.
Not really, no. The device characterization is what lets us predict what RGB is needed to create any particular color.
Quote:
> So rather then the current functionality that says, "correct hue by this amount, and deal with the increased black level", I would like to see functionality that says, "hey, try and correct hue as you see fit, but do not increase the black level above this level I have set, I'll deal with any remaining hue issues".
Sorry, that's a lot of work for something that is not needed 99% of the time.
gwgill is offline  
post #2127 of 2671 Old 06-12-2014, 06:42 PM
Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 197
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 40 Post(s)
Liked: 34
Quote:
Originally Posted by Audionut11 View Post
Can someone provide some pointers as to why the gamma response drops at lower luminance levels? I assume this is black point compensation.
Yes. The effect of using BPC in a 3D LUT workflow is that the response will be offset and scaled to account for the non-zero black point.

Quote:
Originally Posted by Audionut11
Also, why the color temperature rises at lower luminance levels.
If you didn't use black point hue correction during calibration and then used BPC, this will mean that the response gradually crosses over from the whitepoint to the native blackpoint of the display. So the higher color temperature at lower luminance is an effect of the native blackpoint.

Quote:
Originally Posted by Audionut11
I assume it will be better to adjust the display pre-calibration, to better match the curves in the measurement report.
Maybe.
fhoech is online now  
post #2128 of 2671 Old 06-12-2014, 06:49 PM
Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 197
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 40 Post(s)
Liked: 34
Quote:
Originally Posted by gwgill View Post
That seems to make little difference. It is simply not honoring the markup, as you can see in this post.
It seems that the old form of quoting no longer works as intended, and you have to add the quoted user's name like so: [ quote=fhoech ] Lorem Ipsum [ /quote ]
(spaces only added to prevent the forum from parsing the code)
fhoech is online now  
post #2129 of 2671 Old 06-12-2014, 08:04 PM
Advanced Member
 
spacediver's Avatar
 
Join Date: May 2013
Location: Toronto
Posts: 840
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 71 Post(s)
Liked: 70
Graeme, any idea what went wrong with dispcal in my above post?

I'm gonna re-run tonight using the argyll test pattern generator instead of madvr, and not use the -Ib switch.
spacediver is online now  
post #2130 of 2671 Old 06-12-2014, 10:23 PM
Senior Member
 
Audionut11's Avatar
 
Join Date: May 2008
Posts: 354
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 77 Post(s)
Liked: 52
Thanks fhoech, but what about this?

Quote:
Originally Posted by Audionut11 View Post
It appears to be a bug.

Toggling BPC, adjusting Black output offset, and toggling the gamma tone curve between relative and absolute, produced no changes in the gamma curve.

I then adjusted the gamma to 1.8 absolute, with produced a nice flat curve @ 1.8. I then adjusted back to 2.4 absolute, which produced a nice flat curve @ 2.4. After adjusting some black level settings (sorry I don't recall which ones exactly), it then proceeds to produce a gamma curve like the one in my measurement report again. And again, this curve cannot be adjusted back to a flat state, without adjusting the gamma value.
Audionut11 is offline  
Reply Display Calibration

User Tag List

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