CurtPalme.com Home Theater Forum

[David_Web]

http://www.argyllcms.com/

After another search to find an app that supports my probe Huey I found this little awesome app.
It can create ICC profiles or gamma ramps and a lot more.
I had to go beta for it to work but it seems stable.
It supports most probes and seems to be extremely accurate.

A normal display calibration takes a few minutes and does a surprisingly good job.
You could also make a complete measurement of the display with error detection and everything. 800+ readings takes a while though to I have not tested yet.

It is command line only and takes a while to figure out.
I had to specify the x,y of my wanted whitepoint for it to take.

It also helps you to set black and white level if you want to.

to run a display callibration with a square in the middle
-y c is CRT (-y l would be LCD)
-q l (l,m or h for different quality, h takes longer to run)
-d 1 (-d 2 would be secondary monitor)
-p 0.5,0.5,0.3 (first two is x ans y with 0.5 being in the middle. The last is size.)
dispcal -v -y c -q l -B -d 1 -w 0.313,0.329 -p 0.5,0.5,0.3 filename

To load the curve into the LUT
-d 1 (-d 2 would be secondary monitor)
dispwin -v -d 1 filename.cal

some sensors should be calibrated, refer to help doc for that.

No one was reading it so I changed the title to be a little more telling.

I tested this on my current M8000 with great results.
I also added -B to the argument to make the background black and added some more arguments.

I have yet to see any software come close in terms of accuracy. Although I have not tested any of the one included with other sensors but it beat huey with a huge margin. The Huey sensor literally went from crap to good.
I recommend any one to test it as it supports most sensors and is really easy to use.

[David_Web]

Version 1.0.0 is now released.

I highly recommend this little app if you use a HTPC and want it calibrated properly.

A calibration of the display will only get you so far so why not give it a little extra.

However do it with analogue RGB only as a 8 bit digital signal will show banding.

I can't wait to try it with HD145 lenses.

My laptop now has correct grayscale and it really gives the little extra, even if you loose a little brightness.

[rajdude]

Ok I am downloading this app and will report back.
Sounds pretty good!

[rajdude]

David you must be very tech savvy or a Unix administrator to be able to say it is easy!!

[rajdude]

Ok, so this Argyll does calibration for your "desktop" in a HTPC/any computer.

But as we all know the desktop usually varies from the video. The video is dependent on the video player software. For example I had an issue with Vista Media player crushing black when everywhere else blacks were fine. Also the display is set to 0-255 and the video is usually set to 16-246

How does one handle these issues?

Do we calibrate the color and grey scale for the desktop first (using a CMS like this) and hope all is well for the video?

OR

do we calibrate the projector to the best of its capabilities using usual procedures and THEN use CMS to improve the image?

[David_Web]

I use Haali Video Renderer http://haali.cs.msu.ru/mkv/ (included in the same package as the splitter) instead of VMR9 or overlay
It can convert 16-246 to 0-255 PC scale. Easy to setup using media player classic home cinema.
As the calibration is loaded into the LUT of the graphics card the video is properly converted as well.
Using VMR9 should also be corrected but might be iffy at times.

I would calibrate projector first but is should not be critical unless it have to make extreme corrections as windows don't accept too much correction. (for example lowering red output to just over half will get an error, at least when I have playd with making my own LUT)
I would get the white point as good as it gets with the built in wizard and make sure mt G2 was ok. No real need to battle any humps as they are digitally corrected. Not even defocus blue. Do note that you will loose total brightness as you have to compensate by lowering output of R and G. White point should be set at 100IRE as it determines total light output.
As most graphics cards use 10bit LUT there is minimal chance for banding anyway. (which Argyll tries to avoid btw)

No I'm as new to command line as any other. It's easy if you just add commands after the program. It's pretty easy to understand the different arguments in the dispcal manual.
He is planning to add a GUI and more features like graphs and stuff so it should be easier to use in the future. No idea when though.

[David_Web]

I coded my own simple LUT viewer to see a better view of the LUT curve then what powerstrip offers.

I noticed some stair steps in the curve meaning it was not smooth and would show banding.
I reloaded the LUT through Argyll and it was smooth as butter.

This means that PowerStrip don't store the curves with enough precision meaning DO NOT USE POWERSTRIP TO STORE COLOR CORRECTION LUT WITH!!!

btw here is the program I made to see the LUTs. Do note that it is ugly and flickers a LOT. And not really accurate (or at least not tested for it) You need to redraw the LUT every time the window is resized.
Auto refresh runs at 10hz

[rajdude]

David, which video card do you use?

I heard that powerstrip does not work for newer cards. I have a nVidia 8600GTS and a 8500GT.

I remember older nvidia cards came with drivers which would let you view and change some kind of graphs manually.....I think they were gamma curves. Is that the same thing as LUT?

Alas ! The newer cards don't have that functionality

IMHO LUT is this:
http://en.wikipedia.org/wiki/Lookup_table and
http://en.wikipedia.org/wiki/3D_LUT

[David_Web]

My laptop has an ATI MOBILITY FireGL v5200

A LUT or look up table is what it sounds like.
A table to see what value (in 10bit digital format in this case) it should send out for any given input.

They are usually called gamma curves as they are used to correct for gamma. But the proper term is IMHO LUT as you are not inputting a gamma exponent. Yet the function called is GetDeviceGammaRamp. Buy that's only my opinion.

[rajdude]

Thanks for the explanation David.....this makes it much clearer to me.

BTW, I just installed the software which came with my iOne Display LT (Eye-One Match 3). After running through its steps on my LCD monitor it created a ICC profile. DUDE ! I am amazed at the difference it made!

I ran a before and after calibration using HCFR......again I was amazed how much difference this ICC made to the measurements. The RGB levels graph is almost flat now! Take a look at the before and after chc files.

I am going to try out the Argyll with my mouse now. If it works similarly........the Eye-One Match 3 is surely the way to go! Just click next next ....wait a minute or two and you are done! AWESOME!


I did learn something about applying ICC on XP Pro using the display control panel:
You gotta restart the machine to really apply the new icc.......otherwise nothing happens.

But how come the Eye-One Match 3 is able to apply ICCs on the fly?

[David_Web]

Looking nice.
Could you do a calibration with Argyll and compare?

Preferably with high quality option.
Should be something like this

dispcal -y c -q h -B -d 1 -G 2.2 -w 0.313,0.329 -p 0.5,0.5,0.3 filename

As far as I know ICC works only in ICC aware apps like Photoshop and similar. That way you get the full color space conversion so it would look the same regardless of the gamut of the monitor (as long as it's bigger than the target)
This can't be done with a simple LUT based approach. You need a full 3x3 conversion matrix for that.

So if you see it applied instantly to the desktop it's most likely a LUT correction. But I might be wrong and have not heard about any alternatives so...


I have never tried to profile my screen with Argyll but in the documentation 1000-3000 readings is not unusual if you want to be very accurate. It does a lot of error correction and testing just to make sure the result is as good as it can be.
targen is used to generate the values that will be measured. dispread is the app to use for the measurements. colprof to generate the ICC profile.
note, as far as I know you should really calibrate to gamma 2.4 if the room is light controlled. The 2.2 is for a lit up film set meaning the blacks need to be more visible. But the film is intended to be displayed in a monitor with gamma 2.4.
This is a good read about gamma http://www.poynton.com/papers/IST_SPIE_9801/index.html

The Argyll docs also takes up this. It's under the gamma section for dispcal.

A gamma of 2.4 will give you more punch anyway and a CRT pj is fully capable of resolving the shadow detail when corrected properly.

[rajdude]

David, I have not yet tried the Argyll, but I did try the Eye-One Match 3 on my Plasma yesterday. I notice that it does not do as good here. I will attach more files later but basically the Lumiance and gamma is WAY off. The RGB curves are spot on. Gotta start a seperate thread for that problem

I notice that the ICC is applied to everything.........the desktop included.

After playing around with Eye-One Match 3, I feel that a Lumagen type 14 point gamma and LUT correction capability is needed for manual calibration. Argyll and Eye-One Match 3 are both automatic........I cannot find a way to tweak the LUTs........unless the ICC is really a text file. is it?

There is another gamma app on AVS which I am going to try out soon.


BTW, how do you manage gamma using Argyll?

[David_Web]

Yes it is a text file. Not sure if it is easy to edit though.
You should not need to change anything if you set the arguments right in Argyll.
Read the doc for dispcal as there are some options regarding how it handles black level.
Mainlt -f and -k
You are still only applying corrected values to the graphics cards LUT. I think ICC profiles holds this info as well.

You choose the gamma with the -G flag.

I tested that app on AVS and it was horrible to work with. Not sure if he have updated it yet.

My patch made in VVVV works a lot better but it has no use now when I found Argyll.
I might clean it up a little and post it on my userpage at VVVV.org
Might be a little hard to use for people not used to vvvv though.

If you want you could use the -e flag with the number of times Argyll should keep refining the calibration.

Try the plasma with Argyll and see what happens. Be sure to check out the -f and -k flag as it will have a huge impact on how Argyll handles gamma.

[rajdude]

David, not sure which app from AVS you tried but I just spent a couple of hrs playing with the app in this thread:
http://www.avsforum.com/avs-vb/showthread.php?t=1042160

one word.........AWESOME!

That is EXACTLY what I am looking for...........gives the control in OUR hands...............not some software. That app + HCFR works wonders on my laptop.
Can't stop smiling!

[David_Web]

The UI is still horrible.

But it's a useful app if he gets it to work properly.

But I don't see how entering values manually will work better than a proper app that makes multiple passes to make sure you get the right color on screen.

You could use VideoEqualizer to edit the LUT later as it loads the current LUT.

[rajdude]

David.......frankly I am still shying away from Argyll because of the command line only thingy.

Anyway let me explain why manual is important for me.

When I am doing calibration of my HTPC first, I dont have a USB cable long enough to connect the eye-one probe to my HTPC, second I dont want to install any extra software which does curves etc. Imagine yourself doing a calibration for a customer.......will he want you to install apps on your HTPC?

I run the calibration program (HCFR) on my laptop, take readings and adjust the LUT with VE as needed. It is quick and simple.

By the way after calibration and using the VE and LUT etc etc I have some issues with the calibration not looking good........basically even with the calibration looking good in measurements I have black crush and people looking sunburnt (red-push?) and also a lack of pop on the screen.....bacsically low brightness.

Judging from how little people read curt's forums I guess we should get more ppl involved by going to AVS, there is a whole section there:
http://www.avsforum.com/avs-vb/forumdisplay.php?s=&daysprune=&f=139

Lastly, why do you think that VE is not working properly? It works fine for me!

[David_Web]

Well he does have a road plan for GUI but it's not priority at the moment. Command line is ugly but at least it's available at all.

USB extension cables are cheap and works great.

What software do you have to install?
Argyll don't need to be installed and as far as I know you don't need the device driver installed.


My problem with VE its that the points don't move as expected. The interpolation applied by each point is ugly at best (I get a pointy hump when "dragging" one point. It should be Bézier or similar.
Brightness and contrast is implemented wrong (bump it up and then back down).

The point of the app is nice but it's not refined enough yet.

I might make my own app in the end to so the same but as I want it but that will be later. Maybe.

[rajdude]

Actually I never drag a point in VE.

I click on an empty space just above or below the point which I want to move..........the point moves a little bit....depending upon how far away my clicking position was.

This way it is easier to move the points.

I totally agree that VE is far from perfect...........but at this time it works for me.........better than Eye-One Match since it gives control in my hands.

[rajdude]

After playing around with VE for a few days I think I'm going to give Argyll a try.

VE just takes too much time. an automated means is better. I just ordered two 16' active USB extension cables.

[rajdude]

I keep getting this error:

dispread failed with 'Instrument Access Failed'

Driver issue? Any tips?