MadVR - ArgyllCMS - Page 49 - AVS Forum

AVS Forum > Display Devices > Display Calibration > MadVR - ArgyllCMS

Display Calibration

MSL_DK's Avatar MSL_DK
02:03 PM Liked: 10
post #1441 of 2851
01-20-2014 | Posts: 132
Joined: Nov 2011
Isn't this the right settings using dispcalGUI 1.7.1.6 Remembered madshi mentioned to leave this setting alone when using argyllcms


|Tch0rT|'s Avatar |Tch0rT|
06:38 PM Liked: 62
post #1442 of 2851
01-20-2014 | Posts: 629
Joined: Nov 2006
Quote:
Originally Posted by paulkono View Post

You have to enable ICCv4 in your browser and other applications otherwise the colors in pictures will appear "inverted" due to the swapped matrix. For Chrome you have to launch with the "--enable-monitor-profile" flag and for Firefox you have to set "gfx.color_management.enablev4" to true and "gfx.color_management.mode" to 1.

I'm currently using the ICC profile generated by the steps in the first post. Only thing I did different was untick "Low quality B2A table".

Thanks but is there anyway to convert it to ICC version 2? Or do I lose something doing that? Am I correct in saying that it kinda doesn't matter for madVR use since the video players just override the ICC profile anyway and madVR applies the same stuff with the generated 3DLUTs?
gwgill's Avatar gwgill
09:08 PM Liked: 88
post #1443 of 2851
01-20-2014 | Posts: 683
Joined: Jan 2013
Quote:
Originally Posted by |Tch0rT| View Post

Thanks but is there anyway to convert it to ICC version 2?

It's not clear where you got ICCV4 profiles from, because ArgyllCMS certainly didn't create them...
|Tch0rT|'s Avatar |Tch0rT|
10:54 PM Liked: 62
post #1444 of 2851
01-20-2014 | Posts: 629
Joined: Nov 2006
Quote:
Originally Posted by gwgill View Post

It's not clear where you got ICCV4 profiles from, because ArgyllCMS certainly didn't create them...

Ah sorry I assumed it was ICCV4 since they do a similar thing to pictures in my browser and windows picture viewer as the ICCV4 profiles that X-Rite's software makes.
paulkono's Avatar paulkono
11:22 PM Liked: 10
post #1445 of 2851
01-20-2014 | Posts: 23
Joined: Aug 2013
Yeah I meant ICCv2. If you tick "disable GPU gamma ramps" in madVR then it'll clear the videoLUT from the GPU. Then you can load the 3DLUT only, assuming you enabled BT.1866 gamma mapping when when created it.
VerGreeneyes's Avatar VerGreeneyes
01:58 AM Liked: 12
post #1446 of 2851
01-28-2014 | Posts: 112
Joined: May 2013
madshi released a new version of madVR recently which adds various OpenCL-only features, including a new dithering method called error diffusion, not enabled by default. This method has a lower noise floor, so it might be interesting to try the test pattern generator included with 0.87.4 to see if it changes our calibration and profiling results smile.gifNote that OpenCL is broken in the most recent Nvidia drivers, so if you have an Nvidia card, you'll need to use driver version 327.23 or older. Hopefully this will be fixed or worked around soon.
MonarchX's Avatar MonarchX
11:04 AM Liked: 30
post #1447 of 2851
01-28-2014 | Posts: 669
Joined: Jul 2012
Quote:
Originally Posted by VerGreeneyes View Post

madshi released a new version of madVR recently which adds various OpenCL-only features, including a new dithering method called error diffusion, not enabled by default. This method has a lower noise floor, so it might be interesting to try the test pattern generator included with 0.87.4 to see if it changes our calibration and profiling results smile.gifNote that OpenCL is broken in the most recent Nvidia drivers, so if you have an Nvidia card, you'll need to use driver version 327.23 or older. Hopefully this will be fixed or worked around soon.

LOL, I was about to post about the madVR update. Which nVidia drivers are you talking about? They just released 334.67 betas. How do I check to see if OpenCL is broken?
mightyhuhn's Avatar mightyhuhn
11:46 AM Liked: 125
post #1448 of 2851
01-28-2014 | Posts: 745
Joined: Nov 2013

334.67 is known to be broken. the problem is reproduced by nvidia so maybe next update/beta.


MonarchX's Avatar MonarchX
01:21 PM Liked: 30
post #1449 of 2851
01-28-2014 | Posts: 669
Joined: Jul 2012
Driver packages contain opencl32.dl_ , opencl64.dl_ , nvopencl.dl_ and nvopencl32.dl_ and . Those can extract actual .dll files using DOS command "expand -r filename.dl_" and be placed in nVidia drivers directory and Windows\System32 and Windows\SysWOW64 directory (on Windows 8.1). I will try to replace these files from an older package to see if it works. I've done similar stuff before and it worked.
MonarchX's Avatar MonarchX
01:47 PM Liked: 30
post #1450 of 2851
01-28-2014 | Posts: 669
Joined: Jul 2012
EDIT: Please read the post below - you can and you should make OpenCL work with the latest drivers as they use updated OpenCL files, one of which simply needs to be slightly renamed to work properly with madVR.
MonarchX's Avatar MonarchX
01:52 PM Liked: 30
post #1451 of 2851
01-28-2014 | Posts: 669
Joined: Jul 2012
I figured it out even further - no need to replace any files. nVidia simply did not name their drivers properly! All you have to do is go into Windws\SysWOW64, find nvopencl.dll and rename it to nvopencl32.dll! I realized that when I extracted the same file from the 327 drivers and noticed that one was nvopencl32 and the other nvopencl.dll while the newer drivers were both nvopencl.dll. I hope this helps someone!

EDIT: original post had a typo (opendl32.dll instead of opencl32.dll). Rename nvopencl.dll to nvopencl32.dll in Windows\SysWOW64 directory. I promise it will not do anything bad and will, in fact, correct OpenCL rendering for madVR and other OpenCL applications. These type of naming errors happen all the time - they happened a lot with AMD drivers back when it was known as ATi. I would advise to use 334.67 drivers as they have much larger updated OpenCL.dll files comparing to 327.23 drivers.

madVR works flawlessly now and I can tell that video fidelity has improved over the previous 86 version. I saw the change log and it appears quite a few things were not properly rendered in those older versions. It sure makes a difference. I'm about to re-measure grayscale and color gamut of my 3DLUT using OpenCL dithering to see I need to make another 3DLUT!
MonarchX's Avatar MonarchX
10:58 PM Liked: 30
post #1452 of 2851
01-28-2014 | Posts: 669
Joined: Jul 2012
I checked my calibration using new dithering technique and found no difference, which is more of good news than bad news. I do not know if it is a placebo effect, but new madVR seems to to render content much better - colors seem more vibrant for some reason.
mightyhuhn's Avatar mightyhuhn
08:50 AM Liked: 125
post #1453 of 2851
01-30-2014 | Posts: 745
Joined: Nov 2013

this change a top there just breaks opencl, so madvr isn't even trying to use it anymore, so no black screen but no opencl too just broken opencl plz make sure didn't try this anymore thanks.

 

reproduce in doom9 madvr thread.


James Freeman's Avatar James Freeman
09:03 AM Liked: 75
post #1454 of 2851
01-30-2014 | Posts: 260
Joined: Sep 2013

 

As mightyhuhn already stated, this fix actually does nothing.

 


MonarchX's Avatar MonarchX
10:11 AM Liked: 30
post #1455 of 2851
01-30-2014 | Posts: 669
Joined: Jul 2012
Quote:
Originally Posted by mightyhuhn View Post

this change a top there just breaks opencl, so madvr isn't even trying to use it anymore, so no black screen but no opencl too just broken opencl plz make sure didn't try this anymore thanks.

reproduce in doom9 madvr thread.

How do you know that? Anyway, at least I tried to do something and help... It did seem like it worked since green screen went away.

How does madVR look for the .dll files? Previous working OpenCL drivers used nvopencl32.dll for SysWOW64 directory and the new broken-OpenCL drivers use nvopencl.dll for the same directory - how does madVR know which files to use? Does it use the .dll file from system32 directory or SysWOW64 directory?

OK, I will try another fix I thought of earlier - using working-OpenCL 327 drivers to extract nvopencl32.dll (for SysWOW64 folder) and nvopencl.dll (for system32 folder), but renaming nvopencl32.dll in SysWOW64 folder to nvopencl.dll since madVR and new drivers specifically use look for that exact .dll name for OpenCL functions.

I'll try it and get back, but I would appreciate if someone were to tell me how to verify if it worked. Doom9 forums have been down lately... (for me at least).
James Freeman's Avatar James Freeman
10:22 AM Liked: 75
post #1456 of 2851
01-30-2014 | Posts: 260
Joined: Sep 2013
Quote:
Originally Posted by MonarchX View Post


How do you know that?

Tested by many @ doom9.org.

 

No increase of CPU/GPU usage at all.

NNEDI3 functions do not work at all (although do not freeze/black screen).

 

It just bypasses OpenCL completely, hence not apparent side effects.

The dithering you see is just normal random dithering with openCL bypassed although looks like OpenCL error diffusion is activated... Its not.


MonarchX's Avatar MonarchX
10:29 AM Liked: 30
post #1457 of 2851
01-30-2014 | Posts: 669
Joined: Jul 2012
Quote:
Originally Posted by James Freeman View Post

Tested by many @ doom9.org.

No increase of CPU/GPU usage at all.
NNEDI3 functions do not work at all (although do not freeze/black screen).

It just bypasses OpenCL completely, hence not apparent side effects.
The dithering you see is just normal random dithering with openCL bypassed although looks like OpenCL error diffusion is activated... Its not.

Ah...OK, sorry - I will try the other method I explained in my edited reply above. Could you please link me to the thread discussed?
James Freeman's Avatar James Freeman
10:37 AM Liked: 75
post #1458 of 2851
01-30-2014 | Posts: 260
Joined: Sep 2013

This is the Official (the one and only) thread:

http://forum.doom9.org/showthread.php?t=146228


slosvt's Avatar slosvt
11:50 AM Liked: 15
post #1459 of 2851
01-30-2014 | Posts: 473
Joined: Dec 2002
any help with the error below?


fhoech's Avatar fhoech
08:28 AM Liked: 51
post #1460 of 2851
01-31-2014 | Posts: 262
Joined: Apr 2013
Go to menu options -> additional command line parameters and check if you have set the -E parameter for dispcal/dispread. If so, remove it.
MSL_DK's Avatar MSL_DK
04:03 AM Liked: 10
post #1461 of 2851
02-05-2014 | Posts: 132
Joined: Nov 2011
Quote:
Originally Posted by N3W813 View Post


*I will be updating the 1st post soon

Is there a chance for an update?
N3W813's Avatar N3W813
11:57 AM Liked: 36
post #1462 of 2851
02-05-2014 | Posts: 1,140
Joined: Feb 2003
Quote:
Originally Posted by MSL_DK View Post

Is there a chance for an update?

Sorry, I haven't had time and will not have time until end of this month to work on updating this workflow.

Is there a specific step that you need instructions on? The workflow has not changed much since the last update except for the addition of profile verification with madTPG.
James Freeman's Avatar James Freeman
10:11 AM Liked: 75
post #1463 of 2851
02-06-2014 | Posts: 260
Joined: Sep 2013

There are combination of settings that give (me) a perfect picture *:

* No Elevated Blacks, No Black Dithering Noise, No blue whites, Perfect color, Perfect Gamma Curve.

 

 

 

 

I have tried all combinations of these:

 

2.2 Relative - Absolute Colometric  =  Blue Whites.

2.2 Relative - Absolute Colometric w/ White Pouint Scaling  =  Blue Whites, Extended Grayscale.

2.2 Relative - Absolute Appearance  =  Good (Colors/Gamut not perfect).

2.2 Relative - Relative Colometric  =  Elevated Blacks.

2.2 Relative - Luminance Matched Appearance  =  Perfect.

 

 

2.2 Absolute - Absolute Colometric  =  Blue Whites.

2.2 Absolute - Absolute Colometric w/ White Pouint Scaling  =  Blue Whites, Extended Grayscale.

2.2 Absolute - Absolute Appearance  =  Good (Colors/Gamut not perfect).

2.2 Absolute - Relative Colometric  =  Elevated Blacks.

2.2 Absolute - Luminance Matched Appearance = Blue Elevated Blacks...

 

 

If you guys have problems in MadVR with Dithering and noise in the blacks try the one setting.

Of course you can use any gamma you want (2.4) or Apply Calibration (I use windows so its always on).

 

Note that my calibration (Asus Gamma 2.2) is a pure power curve (Gama 2.2 Relative) with Black Offset 100%, Black Point Correction 0%, Profile is default Curves + Matrix.

 

I think you can calibrate to MadVR preset, BT.1886 & LUT Profile, I don't think it matters.

 

I do not use MadVR for calibration or the BT.1886 setting in argyllCMS, I found that the standard Gamma 2.2 Relative gives me the most pleasing and balanced results for my monitors without extra hassle (these settings are standard for a reason).

 

Try and tell me what you think.


James Freeman's Avatar James Freeman
02:36 AM Liked: 75
post #1464 of 2851
02-07-2014 | Posts: 260
Joined: Sep 2013

Double post... but things are going slow lately anyway.

 

 

I want to use the "Measurement Report" in DispCalGUI with MadvrTPG...there are few disabilities so that I can't get the results I expect,

My goal is to compare the MadVR 3DLUT or yCMS with the Rec.709 color space, THATS IT.

 

Is there a simple way to use MadvrTPG to show the colors with the already set Calibration parameters in MadVR settings?

For example: I want to compare the yCMS LUT I have created with rec.709 color space... Just the color space (Gamut)... But I CAN'T.

 

Thanks.


andreamusso.am's Avatar andreamusso.am
04:15 AM Liked: 10
post #1465 of 2851
02-07-2014 | Posts: 3
Joined: Apr 2010
Hello,

excuse my English , I write with the translator of google and also are just starting out, so bear with me for the very basic questions :
- I have a matrix of Zoyd for HCFR format . Thc ( which fits well to fill in cmmx files for madVR , I downloaded a matrix from the link already reported by Zoyd same , but the values ​​of the matrix. Thc , ​​leaving them devoconsiderare from top left to continue with the next to the right, or from the first birth at the top left and then go down again to the left , and so on ? ( I am attaching the representation of the matrix. thc )

- Reading the parameter ' -E ' Usage Scenarios, I do not understand , where do I find the above parameter , since my TV does not seem to accept the 0/ 255 (the TV is an LCD lamps Samsung C650 ) . I say that does not seem to accept the 0/255 because according to the guide word for word , to the extent that I have to create the 3D LUT , boxes and input ouput , I appear 16/235 .

- I currently have a Nvidia GTX 460 GPU , which should not be able to process it all with madVR ( use JRiver as a player set to Red October HD) , but the video shows no one shot, then I think that I see that the video is not rendered full with madVR , even if the opening, icon appear that precisely madVR is in operation. Where am I doing wrong ?

Sorry again and thank you infinitely in advance for your kind answers.



Cattura.PNG 49k .PNG file
Attached: Cattura.PNG (49.3 KB) 
James Freeman's Avatar James Freeman
09:49 AM Liked: 75
post #1466 of 2851
02-07-2014 | Posts: 260
Joined: Sep 2013

Well that does it...

 

I can't generate a MadVR 3DLUT that doesn't change my perfectly good calibration curve.

All that I've tried enforces me to use BT.1886 which I DON'T WANT !!!

 

I just want a gamut correction to rec.709 and leave my calibration (greyscale) alone.

Graeme, Florian, why is it still not possible?


Kukulcan's Avatar Kukulcan
10:39 AM Liked: 23
post #1467 of 2851
02-07-2014 | Posts: 134
Joined: Sep 2012
Quote:
Originally Posted by James Freeman View Post


I just want a gamut correction to rec.709 and leave my calibration (greyscale) alone.

1) Install the last updates of Argyll and dispcalGUI
2) Choose madvr settings in dispcalGUI
3) switch tone curve to gamma and select 2.2
4) flag black point compensation (even though I can't see its effect in the log file)
5) generate the 3dlut as suggested in the guide and with apply BT1886 flagged but select 2.2
6) enjoy your pure gamma curve.
N3W813's Avatar N3W813
11:09 AM Liked: 36
post #1468 of 2851
02-07-2014 | Posts: 1,140
Joined: Feb 2003
Quote:
Originally Posted by Kukulcan View Post

1) Install the last updates of Argyll and dispcalGUI
2) Choose madvr settings in dispcalGUI
3) switch tone curve to gamma and select 2.2
4) flag black point compensation (even though I can't see its effect in the log file)
5) generate the 3dlut as suggested in the guide and with apply BT1886 flagged but select 2.2
6) enjoy your pure gamma curve.

This method will not generate a pure 2.2 power gamma curved 3DLUT. The resulting 3DLUT will still be generated with a BT.1886 curve but with either an absolute or relative 2.2 gamma depending on the option you choose. AFAIK, there is currently no way to generate a pure power 2.2 gamma curved 3DLUT with ArgyllCMS that will work correctly with MadVR.
James Freeman's Avatar James Freeman
11:15 AM Liked: 75
post #1469 of 2851
02-07-2014 | Posts: 260
Joined: Sep 2013

Kukulcan,

 

I already have a perfectly good calibration+profile ICM with Gamma 2.2 Relative.

I now need to convert it to 3DLUT without changing the Gamma curve only the Colors.

 

Thank you, but it does not work. It still changes the gamma curve to some extent.

Turning off BT.1886 raises the gamma to extreme, making it super bright (as expected from the REC.709 profile).

On the other hand, when selecting BT.1886 2.4, the gamma is correct at the shadows but darkened at the middle.

There is no option to leave calibration as is.

 

When I load the 3DLUT into MadVR I see Gamma & Color change, I want to see ONLY Color change.


James Freeman's Avatar James Freeman
11:19 AM Liked: 75
post #1470 of 2851
02-07-2014 | Posts: 260
Joined: Sep 2013
Quote:
Originally Posted by N3W813 View Post


This method will not generate a pure 2.2 power gamma curved 3DLUT. The resulting 3DLUT will still be generated with a BT.1886 curve but with either an absolute or relative 2.2 gamma depending on the option you choose. AFAIK, there is currently no way to generate a pure power 2.2 gamma curved 3DLUT with ArgyllCMS that will work correctly with MadVR.

Thank you,

 

FWIW I don't want a "pure" gamma curve, I want MY gamma curve from the calibration I already did (carefully tweaking the Black Offset & Rate).

 

Meantime solution I use:  BT.1886 2.4 & lower the gamma to 2.05 in MadVR.



Reply Display Calibration

Subscribe to this Thread

Powered by vBadvanced CMPS v3.2.3