MadVR - ArgyllCMS - Page 19 - AVS Forum

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

Display Calibration

hungro's Avatar hungro
09:35 AM Liked: 63
post #541 of 2855
07-27-2013 | Posts: 638
Joined: Mar 2007
Finally managed to generate the 3dlut file for madvr. It is a little choppy but still watchable. My 965 intel integrated chipset is too slow. I am happy that atleast I was able to figure out command prompt and all the codes smile.gif
N3W813's Avatar N3W813
11:36 AM Liked: 36
post #542 of 2855
07-27-2013 | Posts: 1,140
Joined: Feb 2003
Quote:
Originally Posted by baii View Post

A quick question, say if I put the AVS HD 709 calibration disc/mp4 into calibrated madvr with 0-255 input and output.

Ideally, should I be seeing flashing below "reference black", and should I be seeing flashing above 235 white?

Right now, I do not see flashing under reference black, but I see flashing @white 238.

That is correct. If brightness is set correctly, you should not see any flashing bars below 16. If bars above 235 is flashing, that is fine.
N3W813's Avatar N3W813
12:07 PM Liked: 36
post #543 of 2855
07-27-2013 | Posts: 1,140
Joined: Feb 2003
Quote:
Originally Posted by lgreis View Post

Hi there to all, i am newbe to this calibration thing, so a lot of questions i have in my mind that maybe you can answer some of them eek.gif I have an Epson 5020 and an HTPC so i would also want to make the best of it... does the MadVR - ArgyllCMS can do what Lumagen Radiance Mini-3D can? An automatic calibration of the display in 125 color point (whatever this means!!!)? If does, which one makes the more acurate color and gama calibration? What are the diferences between them?
Thanks

Difference between ArgyllCMS/MadVR and Lumagen Radiance
  • On the HTPC, you are required to play all your media through MadVR in order to apply the 3DLUT correction. The Radiance can color correct ANY source (HTPC, BR Player, etc.) connected to it via HDMI.
  • Radiance must be used with Calman, ChromaPure, or LightSpace in order to use the auto calibration of the internal 3DLUT.
  • The Radiance allows settings at 5 points of luminance, saturation, and hue, for a 5x5x5 (125-point) cube . ArgyllCMS does not have that limitation.
  • Radiance 3DLUT correction workflow is more consumer friendly.
  • ArgyllCMS and MadVR are both free. biggrin.gif

lgreis's Avatar lgreis
05:04 AM Liked: 12
post #544 of 2855
07-28-2013 | Posts: 69
Joined: Sep 2012
Thats spectacular news the possibility of doing a perfect and an automatic calibration for free!!!biggrin.gif And from what i interpret of what you said, even more acurate than lumagen because of the 125 point limitation than ArgyllCMS doesnt have.
Thank you for the makers of the madvr and ArgyllCMS;)
zoyd's Avatar zoyd
11:15 AM Liked: 434
post #545 of 2855
07-28-2013 | Posts: 5,041
Joined: Sep 2006
Quote:
Originally Posted by gwgill View Post

Largely I've been relying on the tools to do the right thing - which they generally do, having been proven over many years.

But to address this issue more seriously, here is an update http://www.argyllcms.com/Win32_collink_3dlut.zip which includes some improved tools, support for the latest version of MadVR including madTPG, a revised and updated tutorial on application to video, plus a new tutorial section on verification for video systems. One of the interesting things I found in developing this was how much of an effect gamut clipping can have on the verification numerical result. I've included a number of suggestions as to how to recognize its effect.

Using your verification tutorial here are examples of the sort of improvements one can expect based on dE metrics.

Error vectors before and after


LUT off
Code:
C:\ArgyllCMS\bin>verify.exe -v -n -w -x -k ref.ti3 checkc.ti3
No of test patches = 219
Chose patch 1 as white, xyz 95.046000 100.000000 108.910000
White is from display luminance ref. xyz 131.395865 137.664792 150.154663
L*a*b* white reference = XYZ 95.046000 100.000000 108.910000
No of test patches in worst 10% are = 22
No of test patches in best 90% are = 197
Verify results:
  Total errors (CIEDE2000):     peak = 5.679932, avg = 2.922740
  Worst 10% errors (CIEDE2000): peak = 5.679932, avg = 5.085542
  Best  90% errors (CIEDE2000): peak = 4.539863, avg = 2.704253
  avg err X  2.902348, Y  1.921141, Z  1.507352
  avg err L* 1.816308, a* 5.663599, b* 2.618575

LUT on
Code:
C:\ArgyllCMS\bin>verify.exe -v -n -w -x -k ref.ti3 checkb.ti3
No of test patches = 219
Chose patch 1 as white, xyz 95.046000 100.000000 108.910000
White is from display luminance ref. xyz 127.308060 134.017432 147.093273
L*a*b* white reference = XYZ 95.046000 100.000000 108.910000
No of test patches in worst 10% are = 22
No of test patches in best 90% are = 197
Verify results:
  Total errors (CIEDE2000):     peak = 1.971956, avg = 0.488834
  Worst 10% errors (CIEDE2000): peak = 1.971956, avg = 1.287164
  Best  90% errors (CIEDE2000): peak = 0.886013, avg = 0.404177
  avg err X  0.264835, Y  0.293266, Z  0.400181
  avg err L* 0.337355, a* 0.620370, b* 0.599887


The HCFR saturation runs for the same case are:

LUT off


LUT on


With average dE94 values of:




Note that the current version of HCFR only supports saturation run targets using a gamma of 2.2, so you need to include a 2.2 gamma 3dLUT in your workflow for verification. I'm working on an update to use measured gamma targets and a separate verification tab that will run colors in addition to the saturation sweeps.
gwgill's Avatar gwgill
08:18 PM Liked: 88
post #546 of 2855
07-28-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by baii View Post

A quick question, say if I put the AVS HD 709 calibration disc/mp4 into calibrated madvr with 0-255 input and output.
Right now, I do not see flashing under reference black, but I see flashing @white 238.
There are certainly ways of getting whiter than white to pass through the system. On the individual R,G & B channels you will typically see early clipping (ie. < 235) if your display doesn't have sufficient gamut, and late clipping ( > 235) if you display has a larger gamut than the video colorspace. Similarly, if for some reason you have set a white point that is less than the display white, then you will see flashing > 235. You may also have made a mistake somewhere in the workflow though.
gwgill's Avatar gwgill
09:04 PM Liked: 88
post #547 of 2855
07-28-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by slosvt View Post

dispread: Error - new_disprd failed with 'Instrument Access Failed'
You probably haven't installed the system driver. See http://www.argyllcms.com/doc/Installing_MSWindows.html

You didn't need to do that for the i1d3 because it's an HID USB device.
gwgill's Avatar gwgill
09:33 PM Liked: 88
post #548 of 2855
07-28-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by lgreis View Post

I have an Epson 5020 and an HTPC so i would also want to make the best of it... does the MadVR - ArgyllCMS can do what Lumagen Radiance Mini-3D can? An automatic calibration of the display in 125 color point (whatever this means!!!)? If does, which one makes the more acurate color and gama calibration? What are the diferences between them?
Thanks

The Lumagen Radiance appears to have a very modest sized cLUT resolution of 5. I would imagine that careful per channel calibration and/or TV calibration is needed as well, to get the best out of it, given the very coarse resolution. By comparison, the typical desktop color management RGB profiles use a cLUT resolution of 33, while something like the eeColor uses 65. MadVR uses 256 by default.
lgreis's Avatar lgreis
04:03 AM Liked: 12
post #549 of 2855
07-29-2013 | Posts: 69
Joined: Sep 2012
I dont know what a cLut is but does this means that if the radiance as 5 points of luminance, saturation, and hue for a 5x5x5 cube (125 point calibration) The madvr could have 256 point of luminance, saturation, and hue for a 256x256x256 cube (16777216 point calibration)?
madshi's Avatar madshi
04:10 AM Liked: 140
post #550 of 2855
07-29-2013 | Posts: 5,483
Joined: May 2005
Yes, exactly. However, it's obviously not possible (due to time contraints) to actually measure every one of those 16 million calibration points. So basically what happens is that ArgyllCMS takes a specific number of measurements (you can choose how many exactly) which are spread over the whole data range, and the "missing" measurements in between are then interpolated (which means "guessed" by using some math algorithms).

Well, in reality ArgyllCMS creates a smaller than 256x256x256 3dlut and then in the last step scales it up to 256x256x256.
gwgill's Avatar gwgill
08:00 AM Liked: 88
post #551 of 2855
07-29-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by lgreis View Post

I dont know what a cLut is
cLUT = color Lookup Table = N dimensional interpolation table.
For RGB, N = 3. For CMYK N = 4. etc.
gwgill's Avatar gwgill
08:05 AM Liked: 88
post #552 of 2855
07-29-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by madshi View Post

So basically what happens is that ArgyllCMS takes a specific number of measurements (you can choose how many exactly) which are spread over the whole data range, and the "missing" measurements in between are then interpolated (which means "guessed" by using some math algorithms).
There are many ways to do this, although it depends on what you assume about the measured points. If you assume nothing about the location of the measurement points, you need a "scattered data point" interpolation algorithm. The one that Argyll uses is called regular spline interpolation, which has the property that it outputs a regular grid that balances interpolation accuracy and smoothness. There is a lot more to it than that, since it also creates per channel input and output curves from the data as well as the cLUT.
Quote:
Well, in reality ArgyllCMS creates a smaller than 256x256x256 3dlut and then in the last step scales it up to 256x256x256.
I presume you meant to say "... 65x65x65 and then scales it up to 256x256x256"
RobGreen's Avatar RobGreen
08:50 AM Liked: 10
post #553 of 2855
07-29-2013 | Posts: 8
Joined: Sep 2003
Hi guys,

I just wanted to 'thank you' madshi, N3W813, gwgill and everyone for either creating these great tools (and sharing them for free!) and supporting this thread. I have just completed my first calibration using the method from the first post and managed to generate a 3DLUT for my PC monitor (as a test run) which gives some very nice results.

For those interested I'm using a fairly standard HP LP2475w IPS monitor which has never been calibrated. The pre-calibration readings were:





And post calibration:





The colour gamut looks pretty much spot on. I would like to throw the question out there about the gamma curve to ask if this in the expected result?

Next up is my JVC X30 / RS45 projector. I have been 'playing' at calibrating that for the last few days using on my meter (i1 Display Pro 3) and Chromapure. It seems to have some severe gamma issues in its uncalibrated state and also suffers from a narrower that rec.709 gamut in normal mode and a wide gamut in 'wide' mode. This is where I am hoping madvr / Argyll can help me.

Another question for the experts. What steps whould I run through on the projector prior to generating the 3DLUT? Here I mean should I set gamma and white balance on the projector first then leave madvr / Argyll to correct the gamut? I noticed on previous attemptes to manually calibrate the JVC that achieving near perfect white balance and gamma came at the expense of red and blue luminance. I'm still too new to this to know what that means for me and my work on calibrating the device.

Anyway, thanks again for those who made this software and supported the thread. It is greatly appreciated. If I can send a few $ your way I will gladly do so, just send me a link via PM.

Regards

Rob

P.S - In the guide on page 1, step 6 of the guide says to use the switch -Yn to 'skip meter placement confirmation'. dispread complained about this to me so I left that switch out of the command and confirmed the meter placement myself.
madshi's Avatar madshi
09:24 AM Liked: 140
post #554 of 2855
07-29-2013 | Posts: 5,483
Joined: May 2005
Quote:
Originally Posted by gwgill View Post

There are many ways to do this, although it depends on what you assume about the measured points. If you assume nothing about the location of the measurement points, you need a "scattered data point" interpolation algorithm. The one that Argyll uses is called regular spline interpolation, which has the property that it outputs a regular grid that balances interpolation accuracy and smoothness. There is a lot more to it than that, since it also creates per channel input and output curves from the data as well as the cLUT.

Yes, I know. I intentionally wrote a somewhat simplified explanation because not every user is able to understand things at the level you're writing at... biggrin.gif

Quote:
Originally Posted by gwgill View Post

I presume you meant to say "... 65x65x65 and then scales it up to 256x256x256"

Well, I wrote "smaller than 256x256x256" instead of "65x65x65" because I wasn't sure if it's 65x65x65 in all ArgyllCMS tools, with all quality settings.

Quote:
Originally Posted by RobGreen View Post

I would like to throw the question out there about the gamma curve to ask if this in the expected result?

Can you tell ChromaPure to expect a BT.1886 gamma curve? The form of the gamma curve looks like BT.1886, but whether it's exact or not I can't say. I think ChromaPure by default expects a pure power gamma curve, which isn't really what I would calibrate to, unless you display has perfect blacks.

Quote:
Originally Posted by RobGreen View Post

Next up is my JVC X30 / RS45 projector. I have been 'playing' at calibrating that for the last few days using on my meter (i1 Display Pro 3) and Chromapure. It seems to have some severe gamma issues in its uncalibrated state and also suffers from a narrower that rec.709 gamut in normal mode and a wide gamut in 'wide' mode. This is where I am hoping madvr / Argyll can help me.

Another question for the experts. What steps whould I run through on the projector prior to generating the 3DLUT? Here I mean should I set gamma and white balance on the projector first then leave madvr / Argyll to correct the gamut? I noticed on previous attemptes to manually calibrate the JVC that achieving near perfect white balance and gamma came at the expense of red and blue luminance. I'm still too new to this to know what that means for me and my work on calibrating the device.

I'd suggest to use "wide" mode. I'm not sure if you should set your JVC to 6500 or whether you should set it to maybe 7500 or 8500 and let ArgyllCMS correct it. Maybe both will end up with almost the same results, maybe not, don't know, you could also try both and compare...
RobGreen's Avatar RobGreen
09:39 AM Liked: 10
post #555 of 2855
07-29-2013 | Posts: 8
Joined: Sep 2003
Thanks madshi. I can tell ChromaPure what gamma to target (numeric or BT.1886) but I'm not sure that is relevant here as Chromepaure hasn't been used to set the gamma, madvr and Argyll have (unless I'm misunderstanding). I suppose my question was: Does the gamma curve look like the result you would expect madvr and Argyll to target?

To reply to my own comment about instruction step 6, the switch '-Yp' works perfectly instead of '-Yn' which failed.

I am now calibrating my JVC (as I type) and have started with vanilla settings and no pre-calibration using Chromapure. I went with 'wide 2' colour mode, 6500k and 2.4 gamma in the settings. The pre-calibration numbers were pretty horrid so let's see what madvr and Argyll can do. I will post some before and after calibration reports once it is done.

Cheers\
Rob
madshi's Avatar madshi
09:59 AM Liked: 140
post #556 of 2855
07-29-2013 | Posts: 5,483
Joined: May 2005
If you set ChromaPure to BT.1886, it might adjust the graph layout so that you can better check if what you got is near to the desired result or not.
RobGreen's Avatar RobGreen
10:04 AM Liked: 10
post #557 of 2855
07-29-2013 | Posts: 8
Joined: Sep 2003
Quote:
Originally Posted by madshi View Post

If you set ChromaPure to BT.1886, it might adjust the graph layout so that you can better check if what you got is near to the desired result or not.

I see what you are saying. I will try this once madvr / argyll have finished. On 1261 out of 2000 measurements as I type.
N3W813's Avatar N3W813
12:48 PM Liked: 36
post #558 of 2855
07-29-2013 | Posts: 1,140
Joined: Feb 2003
Quote:
Originally Posted by RobGreen View Post

To reply to my own comment about instruction step 6, the switch '-Yp' works perfectly instead of '-Yn' which failed.

Thanks Rob. Graeme updated the switch settings in the latest binaries. I have updated the 1st post with the correct parameter.

Also took Nezil's recommendation, removed the old workflow. smile.gif
N3W813's Avatar N3W813
01:30 PM Liked: 36
post #559 of 2855
07-29-2013 | Posts: 1,140
Joined: Feb 2003
My latest calibration on my Sharp Elite with Local Dimming setting set to Off.




cmd /c dispwin.exe -v -c > 0.dispwin.clear.log

cmd /c dispcal.exe -v -dmadvr -Yp -qm -m -t 6504 -g2.2 -X "correction.elite.ccmx" "elite.p2.2" > 1.dispcal.log

cmd /c targen.exe -v -d3 -G -f5500 -g128 -s64 "5500_g128" > 2.targen.log

cmd /c dispread.exe -v -dmadvr -Yp -K elite.p2.2.cal -X "correction.elite.ccmx" 5500_g128 > 3.dispread.log

cmd /c colprof.exe -v -qh -aX 5500_g128 > 4.colprof.log

cmd /c collink.exe -v -3m -qh -et -Et -Ib -G -iaw -a elite.p2.2.cal ref\Rec709.icm 5500_g128.icm output\elite.5500_g128.iaw.b22.icm > 5.collink.b22.log
gwgill's Avatar gwgill
07:08 PM Liked: 88
post #560 of 2855
07-29-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by madshi View Post

Can you tell ChromaPure to expect a BT.1886 gamma curve? The form of the gamma curve looks like BT.1886, but whether it's exact or not I can't say. I think ChromaPure by default expects a pure power gamma curve, which isn't really what I would calibrate to, unless you display has perfect blacks.
Note that BT.1886 defines a reference electro-optical transfer function, but nowhere does it mention how this should be applied to color image reproduction.

One way of applying BT.1886 would be to apply it independently to the Red, Green and Blue channels. If the black point is not the same color as the white, this will result in a gradual shift in hue down the neutral axis.

Another way of applying it is to use it to govern the Luminance response, and to treat the color independently. This is what the collink BT.1886 implementation does, maintaining a neutral axis color consistent with the white point to a low level, before "bending" the hue to match the black point color.

Comparing one to the other will show a discrepancy.
MSL_DK's Avatar MSL_DK
02:42 AM Liked: 10
post #561 of 2855
07-30-2013 | Posts: 132
Joined: Nov 2011
N3W813>

Thank you for sharing your settings and results!
dispcal -gs>-g2.2 solved the problems I have had with raised black level (Someone who can explain why?). But I have another problem with white, which is yellow when I use collink -a

I have not adjusted the white balance on my TV, I have adjusted the contrast and brightness.

If I use collink -H, there is apparently no problems. If I use collink -a is white something close to yellow.

If I set the white balance on my TV and use collink -a do I achieve a better result, I think;-) white is certainly no longer yellow. Is there a logical explanation for my problem?

My workflow is the same as you show here
MSL_DK's Avatar MSL_DK
06:45 AM Liked: 10
post #562 of 2855
07-30-2013 | Posts: 132
Joined: Nov 2011
Is this the right way using madTPG?

dispwin -v -c
dispcal.exe -v -dmadvr -Yp -qm -m -t 6504 -g2.2 -X "correction.elite.ccmx" "elite.p2.2"



dispread.exe -v -dmadvr -Yp -K elite.p2.2.cal -X "correction.elite.ccmx" 5500_g128



collink.exe -v -3m -qh -et -Et -Ib -G -iaw -a elite.p2.2.cal ref\Rec709.icm 5500_g128.icm output\elite.5500_g128.iaw.b22.icm
N3W813's Avatar N3W813
11:18 AM Liked: 36
post #563 of 2855
07-30-2013 | Posts: 1,140
Joined: Feb 2003
Quote:
Originally Posted by MSL_DK View Post


dispread.exe -v -dmadvr -Yp -K elite.p2.2.cal -X "correction.elite.ccmx" 5500_g128



dispread.exe -v -dmadvr -Yp -K elite.p2.2.cal -X "correction.elite.ccmx" 5500_g128

Since the command already incorporates the videoLUT calibration settings (-K), enable "disable VideoLUTs" in MadTPG.
nezil's Avatar nezil
01:22 PM Liked: 31
post #564 of 2855
07-30-2013 | Posts: 117
Joined: May 2013
You might need to re-start madTPG, as I've found that changing the 'disable VideoLUTs' button only seems to take effect on restart.

I've also found that if you start madTPG, then disable VideoLUTs manually, with dispwin -v -c, madTPG doesn't seem to notice this change, so pick one method and use that only. In any case, Newbie is correct that if you're using -K, you shouldn't be using VideoLUTs.
gwgill's Avatar gwgill
08:00 PM Liked: 88
post #565 of 2855
07-30-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by N3W813 View Post

Since the command already incorporates the videoLUT calibration settings (-K), enable "disable VideoLUTs" in MadTPG.
Don't fiddle with the buttons in MadTPG, or you will mess things up. Both the VideoLUTs and 3dluts should be in their default (enabled) state.

Use the recommended sequence of command lines, and it will all operated correctly without such fiddling, since the ArgyllCMS tools tell MadTPG exactly how things must be displayed in each situation.
nezil's Avatar nezil
10:47 PM Liked: 31
post #566 of 2855
07-30-2013 | Posts: 117
Joined: May 2013
Quote:
Originally Posted by gwgill View Post

Don't fiddle with the buttons in MadTPG, or you will mess things up. Both the VideoLUTs and 3dluts should be in their default (enabled) state.

Use the recommended sequence of command lines, and it will all operated correctly without such fiddling, since the ArgyllCMS tools tell MadTPG exactly how things must be displayed in each situation.

Graeme,

I think you're correct that if you don't touch anything, that ArgyllCMS and madTPG will do the right thing, the problem is, in my experience. If I want to test the .cal file that I've just created, and see it's results in ColorHCFR, then I'll need to load the file using dispwin -v [filename.cal] and then load madTPG. Sometimes on loading, madTPG will still unload the 2dlut curves, even if disable VideoLUTs is not selected, and I then have to re-run the dispwin command once madTPG is running. Only then is it possible to check the result of the .cal file using ColorHCFR.

Once I've done that, all of the automated ArgyllCMS / madTPG options seem to go astray, and I have to manually enable or disable the relevant 2d or 3d buttons as appropriate.

I think if users only make use of ArgyllCMS and madTPG, and don't use madTPG with ColorHCFR, then the buttons in madTPG can be left alone. Sadly this isn't the case for most users, as many will want to use ColorHCFR to validate the results in a format that they're used to.

Personally, I think it's a little confusing that ArgyllCMS trys to take control of the madTPG buttons, but doesn't give any indication that it's doing so. Is there a way to improve on this, and either make it more transparent, or stop it from trying to take control altogether and just have the user make the settings as required?

I hope I'm not confusing the issue... apologies if that's the case, but it seems confusing to me.
madshi's Avatar madshi
01:26 AM Liked: 140
post #567 of 2855
07-31-2013 | Posts: 5,483
Joined: May 2005
Quote:
Originally Posted by MSL_DK View Post

N3W813>

Thank you for sharing your settings and results!
dispcal -gs>-g2.2 solved the problems I have had with raised black level (Someone who can explain why?). But I have another problem with white, which is yellow when I use collink -a

I have not adjusted the white balance on my TV, I have adjusted the contrast and brightness.

If I use collink -H, there is apparently no problems. If I use collink -a is white something close to yellow.

If I set the white balance on my TV and use collink -a do I achieve a better result, I think;-) white is certainly no longer yellow. Is there a logical explanation for my problem?

My workflow is the same as you show here

Did you use the same files for both "collink -a" and "collink- H"? Or did you redo the full calibration for both tests?

Quote:
Originally Posted by nezil View Post

I think you're correct that if you don't touch anything, that ArgyllCMS and madTPG will do the right thing, the problem is, in my experience. If I want to test the .cal file that I've just created, and see it's results in ColorHCFR, then I'll need to load the file using dispwin -v [filename.cal] and then load madTPG. Sometimes on loading, madTPG will still unload the 2dlut curves, even if disable VideoLUTs is not selected, and I then have to re-run the dispwin command once madTPG is running. Only then is it possible to check the result of the .cal file using ColorHCFR.

Once I've done that, all of the automated ArgyllCMS / madTPG options seem to go astray, and I have to manually enable or disable the relevant 2d or 3d buttons as appropriate.

I think if users only make use of ArgyllCMS and madTPG, and don't use madTPG with ColorHCFR, then the buttons in madTPG can be left alone. Sadly this isn't the case for most users, as many will want to use ColorHCFR to validate the results in a format that they're used to.

Personally, I think it's a little confusing that ArgyllCMS trys to take control of the madTPG buttons, but doesn't give any indication that it's doing so. Is there a way to improve on this, and either make it more transparent, or stop it from trying to take control altogether and just have the user make the settings as required?

I hope I'm not confusing the issue... apologies if that's the case, but it seems confusing to me.

I'm not sure if you're testing with the latest madVR and Argyll versions? With the latest versions, in the moment when Argyll starts remote controlling madTPG, the buttons "disable VideoLUTs" and "disable 3dlut" should reflect what is really happening. E.g. if you don't have either of those buttons pressed and then call "dispcal -R -dmadvr" the "disable 3dlut" button should be shown pressed, as long as the measurements are being done. Also while the measurements are taken, you can't change the state of those 2 buttons, anymore. So basically madTPG shows exactly what it's doing while doing the measurements. I don't see how this could be made more transparent than this!

You're not supposed to use "dispwin" behind madTPG's back, while madTPG is already running. If you do that, you're on your own and you have to live with unexpected behaviour.

madTPG should most definitely *not* disable the VideoLUTs unless:

(1) ... you have the "disable VideoLUTs" button pressed.
(2) ... or Argyll tells madTPG to disable the VideoLUTs during the measurements.
(3) ... or you have a 3dlut active which comes with a linear VideoLUT attached to it.

If you can reproduce a situation where madTPG disables the VideoLUTs even though no measurements are active, the "disable VideoLUTs" button is not pressed and no 3dlut is loaded, please let me know how to reproduce it and I'll be happy to look into it.
gwgill's Avatar gwgill
01:48 AM Liked: 88
post #568 of 2855
07-31-2013 | Posts: 685
Joined: Jan 2013
Quote:
Originally Posted by nezil View Post

I think you're correct that if you don't touch anything, that ArgyllCMS and madTPG will do the right thing, the problem is, in my experience. If I want to test the .cal file that I've just created, and see it's results in ColorHCFR, then I'll need to load the file using dispwin -v [filename.cal] and then load madTPG. Sometimes on loading, madTPG will still unload the 2dlut curves, even if disable VideoLUTs is not selected, and I then have to re-run the dispwin command once madTPG is running. Only then is it possible to check the result of the .cal file using ColorHCFR.
If you want to test the loaded calibration curves, then MadVR doesn't currently have a facility to do so independently of a 3dLUT, so what I'd suggest is to set MadVR to "disable calibration controls for this display" to take that out of the picture, and load the .cal using dispwin etc.
Quote:
Personally, I think it's a little confusing that ArgyllCMS trys to take control of the madTPG buttons, but doesn't give any indication that it's doing so.
ArgyllCMS doesn't "take control" over the madTPG buttons - they are an independent input to the TPG, but madishi has also used them as indicators, to show what's going on. The MadTPG API has the functions to load calibration curves and/or disable the 3dLut that ArgyllCMS needs to do its job.
Quote:
Is there a way to improve on this, and either make it more transparent, or stop it from trying to take control altogether and just have the user make the settings as required?
Definitely not - it would be a nightmare trying to tell users to press/unpress buttons at various steps, with all the confusion and mistakes that would result, and it's not how it works using ArgyllCMS's own patch display tools. Disabling the buttons indicator functionality might reduce superficial confusion, but it would be reducing transparency, not increasing it!

I'd also recommend refreshing the ArgyllCMS tools from http://www.argyllcms.com/Win32_collink_3dlut.zip, since I've tweaked the documentation slightly, and fixed a bug that may affect usage with MadTPG.
RobGreen's Avatar RobGreen
04:21 AM Liked: 10
post #569 of 2855
07-31-2013 | Posts: 8
Joined: Sep 2003
Hi all,

Quick update from me on the success (in my eyes) of my latest calibration efforts.

Before:





Calibrated gamma (2.3 target) and white balance using projector controls. This achieved the following:





And then following MadVR and Argyll I get this:





I must say that I am very happy with this outcome. Thank you again Madshi and Graeme for creating and collaborating on these tools and to others who have supported this thread.

Cheers

Rob
madshi's Avatar madshi
08:21 AM Liked: 140
post #570 of 2855
08-01-2013 | Posts: 5,483
Joined: May 2005
@Graeme,

the very latest collink build doesn't seem to be able to create eeColor box 3dlut files. It gets stuck at "Writing out file" and the file grows very large very fast. I now aborted it after half a minute or so, and the file was already 2GB big!

collink -v -qh -G -iaw -Ib -3e -et -Et -a dispcal.cal ..\ref\Rec709.icm dispread.icm iaw.icm

Reply Display Calibration

Subscribe to this Thread

Powered by vBadvanced CMPS v3.2.3