Monitor and printer calibration

[ajohnw]

I have a bit of a thing about this since one package showed different colours to the others I use. I've checked shots on other machines. They vary a bit but seem to be ok to me so I have never bothered going any further.

As a colour calibrator is a little pricey for home use I have bought a 2nd hand one sold on the basis of needing a new one to suite a mac upgrade. Hope that is the only reason. £21 isn't to bad for home use if it works. It's a Monaco X-Rite Optix XR - Monitor Colour Calibration Device.

I intend to use it with an open source software called Argyll. That's capable of working with many brands of calibrator. One disappointment is that the unit should have come with an IT8 reflective card. The seller says they will look for it so there is some hope but I need to sort out what I need to do if it doesn't arrive.

The calibrator is supplied with the card so that the software can be used to calibrate a scanner and subsequently a printer. Basically the scanner is calibrated via the card, a file that matches the card is printed and then scanned so that a profile can be generated for the printer as well. My problem is that I have the file to be printed but have no idea what type of IT8 card it is.

I did find an IT8 Agfa tiff file and notice that the white is on the warm side. I would expect Kodak to be more neutral and Fuji to encourage Brit's to look tanned. If I buy a card this leaves me wondering what card should be used for calibration? Could be that the one with the unit is out of date as well. Any ideas?

I'm not sure if Argyll will calibrate a printer this way but it is a very flexible package so for the moment live in hope. However if of use people who work on open source software will often extend it. Biggest problem with Argyll is finding a decent from novice tutorial. Most of what I read about it near gobble de gook to me.

John

[ajohnw]

Update
Seems that the gentleman that sold it will be sending me the card soon having found it.

A source of reasonably priced IT8 7/2 cards is mentioned on Argyll's web site. These come with a standard data file so if I have problems obtaining the data for the other card or it looks like it has too few colour patches I shall be ordering one of those. The gentleman has been selling them for a long time. He will even do individually calibrated ones. The ordinary ones are calibrated on a batch basis. There are instructions about for using the reflective cards for camera calibration on the web as well but must be a bright clear sky day at noon. Some I believe use flash and arrange for the reflections not to hit the camera lens. Haven't found much info on that method. A matt card should be used but are a lot more expensive.

Argyll will run on Winows, Mac PC, Mac OS and Linux. There is also a neat GUI mentioned on the site which saves driving the Argyll utilities from the command line ( Dos Box). It will calibrate scanners and printers as mentioned but no GUI for that,

The GUI is called dispcalGUI and there is some fairly easy to follow info on it's site. So far I have only run a check on my monitor. Gamma high but it seems that this is to be expected on a home monitor. The various strip tests on the web indicated that as well also variations with colour. I assume that full calibration will take care of that as it seems it takes a long time.

It will also measure ambient conditions with an ordinary calibrator but warns that they could be out. Going one what it said about mine I am going to change the lighting and see what happens. An ordinary tungsten bulb should make a reasonable check. A led light maybe as well.

John

[sleist]

I use the Spyder3 Express puck with Argyll and dispcalGUI in Windows 7. My monitor is the Dell U2410 wide gamut IPS.
This combination of hardware and software work amazingly well. Far better than the Spyder3 software that came with the puck. There was an excellent post at dpreview that explained how to do things and gave some additional links which were also helpful. I ended up doing it several times with increasing degrees of monitor pre-adjustment and calibration of RBG and white point settings. Finally nailed it quite well. Took about 2 hours to run. I highly recommend Argyll and dispcalGUI.

[ajohnw]

I'm hoping running the uncalibrated monitor check and optimising the dynamic range and black and white points etc will help with final calibration time. Might work. Don't really know. It's possible to set for black level and white level drift checks and take even longer.

There is an interesting read on monitor problems some one posted for me here

http://ninedegreesbelow.com/photogra...r-profile.html

I only found out I had a problem because I used a package called Photivo ( Linux,Windows, Mac(?) ) that uses it's own internal colour management and I wondered why everything looked different in it. Running open source means that I have several options. Gimp takes care of this if set up correctly. It isn't. Not sure what happens with others but suspect that a system wide profile will cure them. Photivo has a selection of profiles including proofing ones at 5000K and sRGB. Adobe options as well but I haven't taken much notice of that as it's for printing really.

I set my monitor in sRGB mode but it didn't make any difference. Colours differed. Today I printed a jpg on an HP printer and the colours came out more like Photivo. The puck shows that my TFT monitor is actually at 5,500K and gamma 2.55. :-) May be worn out. I have had it a long time. The high gamma it seems could be down to normal monitors been set high on purpose so that they show slightly more contrast under the assumption that they will be used in higher light levels than a studio monitor set at 2.2. Its seems 2.4 is often used. It seems usually means I have read it once on the web.

The combined effect taking 2 views of a jpg using Photivo and another was that some white areas in Photivo were "over exposed". The shot included a white notice board with black writing on it below the level of legibility. I had carefully adjusted the shot locally so that what was there looked reasonable rather than weak and wishy washy Some of it couldn't be seen at all viewed in Photivo. All down to my monitor. It seems. Some whites were whiter in Photivo and I have also noticed a sort brownish cast to light shadows in it that are much nearer black in other packages - probably due to white balancing in them in the none corrected packages.

If I do manage to sort this out it will be interesting to see what my carefully adjusted shots look like. It seems that there will always be some doubt in any case due to the state of other peoples monitors.

Some people are using HD TV's as monitors. Many have a suitable connection and also show sRGB video etc. I'm wondering if that is a sensible option as TV makers have to get colours reasonably correct.

John

[ajohnw]

The results of calibrating my monitor are interesting. The shot that started me wondering now looks the same how ever I view it. It's a very processed shot concentrating on getting colours right and max detail. Most of the others haven't changed much other than one tone mapped one of flowers that I took in the rain. It has a little more punch than intended. One I posted on here were I had to dodge in detail in blacks looks slightly over done.

I checked the 2nd hand calibrator with a led desk lamp from Ikea. They have a very small head and give very even illumination. Some people use them for lighting. It came out at 5000k which is correct. I also pointed it out of the door and that made sense too.

I used DispcalGUI with Argll. It's very easy to use. When an icc file is generated it comes up with a display to trim up rgb and brightness on the monitor with ideal target values marked. Turns out I had the monitor way too bright and blue couldn't keep up. Without repeated black and white level drift checks the process took about 1/2hr and DispcalGUI installed the icc file on the system for me. It's a nice package.

One thing I now notice though is a very slight bluish tint to paper white web pages. Could be down to room lighting. No real light from outside and a strip light.. That is cool white but the monitor only receives reflected light. There are few things to play with here. It's possible to add the effects of room lighting. The DispcalGUI web site also points out that contrast setting in the monitor matter and to use the default. There is a scale on mine from 0-100, currently at 75 so am going to try doing another with it set at 50.

It offers several types of profile. I chose to use Web Plus Office. This looks to give a mix of sRBG and Adobe RBG. Gamma comes out at 2.34 slightly high but from the web that is normal for ordinary monitors and is often 2.4 to account for brighter than studio style ambient lighting. I would be interested in any comments from people on this area. It can also set up photo proofing profile files at 5000k but that isn't what I need.

John
PS The slight blue tint is what put me off sRGB. It's not to bad on here, hardly noticeable but more so on some web sites.

[Boatman]

I have a Pantone Huey Pro that has been giving me fits ever since I rebuilt my system OS about three months ago. Up until then it was fine, now it creates a pinkish cast that others have posted is caused by a bad Huey calibarator. The timing is too perfect to belive that the Huey actualy broke.

I noted the Argyll software mentioned in this thread. The Argyll web page does not say much about the features of this package, but I downloaded it and there are files in it for Huey. Does this package work with the Huey device? If so, perhaps it will work better than the Pantone software. Any thoughts?

PS - I found tihs post, http://forums.dpreview.com/forums/re...ssage=39337909, that clearly states that the Huey will work with Argyll. If I can figure out how to install Argyll (I'm not a Linux or command line guy) I'll give it a go a post results.

[ajohnw]

If you also install dispalGUI the monitor calibration is more or less straight forwards. The package is mentioned on Argyll's web page. Info on that is sort of tucked away under documentation including a list of supported devices and solutions to common problems.

If you run dispcalGUI there is an option to calibrate. This brings up an rgb and brightness display with taget settings for adjustment before generating an profile file. I had to use the preset option on my monitor to adjust r g and b individually and the usual brightness control. Great fun as all inter react. There is a note on the gui web site about setting the monitors default contrast. You can also measure ambient conditions with a calibrator. I check mine that way as it was 2nd hand - use the sky, tungsten lamp, warm white led etc.

DispcalGui http://dispcalgui.hoech.net/

Argyll docs http://www.argyllcms.com/doc/ArgyllDoc.html

These are I believe available for linux, windows, mac and mac pc

It's worth trying Argyll even if the calibrator isn't mentioned as it will recognise it automatically if it can via it's hardware. If you are running Linux your distro update package is bound to offer Argyll and probably the gui as well and it will install them for you. You can check that the set up is working sensibly with the ambient measuring capability. All easy with the gui. I suspect I will generate several profile files for my monitor with the contrast at various levels as I have no idea what the default is. The gui also installs the profile on the system. I could only install local user and not system wide. I think the argyll web site mentions how to get round that but I couldn't even do it as root. Makes no odds really as I am the only person that uses this PC.

If you have to compile from source unpack and read the readme file. They nearly always tell you what to do. Otherwise it's action - open console here in the directory of the files, and in the console ./configure -------------- make ------------ su and password or sudo ------ make install. That's for kde to get to the console. Gnome should be similar. Sometimes alternate methods are used but they are always mentioned in the README or INSTALL file. If ./configure comes up with errors it's usually because library files are missing. It tells you which ones so it a case of a web search to find them to install and then ./configure again. It's easier to use your package manager as it will pull in what's needed. Sometimes it wont be the latest version. In that case as I run opensuse I go to the opensuse build service - find the package and see what versions are available and can be run on my version of opensuse ( 11.4 ) The latest versions are usually there in rpm format so I download and install with yast the package manager. If not I compile or take the most recent version. None linux users may wonder where a name like YAST comes from. It stands for Yet Another System Tool. There have been loads of them. It happens to be a good one fortunately.

John

[Brocken]

Searching Google Images for the name of a very popular painting gives a good idea of the scale of chrominance and luminance deviation commonly found in image representations on the internet.

I chose Turner's "Fighting Temeraire" as my first test search, and found the famous sky represented as everything from almost monochrome to the most lurid over-saturation.

I then tried Van Gogh's "Starry Night" with similar results.

Of course, the representation of the original work and its arrival on my monitor results from a series of discrete steps, only the last of which are within my control. Did the apparently discoloured images look good to the person who launched them onto the web? Did they have good colour vision? Do I? Who can say?

[ajohnw]

I've found calibration made most difference to paler colours and also shots that had had black brightened or detail bought out, A very noticeable difference on pale colours. The reason I started wondering about my monitor is all down to a package called Photivo that is colour managed. It can drive the monitor with a few different colour profiles. One is sRGB and the other 2 are 5000K types for print proofing. Switching the monitor to sRGB mode didn't help nor did other adjustments. Web graph tests were no problem other than gamma. I found those to be difficult to use but seemed to show a gamma around 2.5 or so.

Looking around the web and choosing sites that seem to know something about colour and aren't just repeating what others have put there are still some problems. It seems a typical sRGB monitor is set up with a gamma of 2.4 to increase contrast over the spec for sRGB which is 2.2 and intended for a studio situation. At the moment Photivo still shows a very slight colour difference from packages that are using the profile I have produces myself. That could be down to it's sRGB profile being off for my monitor or the fact that I am using a web/office profile which is a mix of sRGB and Adobe RGB and gives a broader range of colours. It turns out that the main problem with my monitor was brightness adjustment. That threw the blue output way off. I had set the monitor up on grey and the usual colour scales. Now it's calibrated it still passes those and with a little more clarity. My ambient lighting is about 3500K as it's reflected and I can notice a slight blue tint to some paper white web pages at times. Could be down to the ambient lighting but on the other hand the only real difference on a screen between 5000k an 6500 is the blue content. The 5000k emission is similar to sunlight in shape. 6,500k has a blue spike and fools our eyes.

Colour vision - mine was 20/20 when it was tested. Is it still probably. Many people don't meet 20/20 colour vision but get very close. Others are really colour blind and don't even know it.

Having seen what a calibrator can do and how easy it is to adjust a monitor I might even consider buying a new one of I need too. 2nd hand is more palatable. Chances are though that many monitors are way out and the owners will never buy calibrators to correct them. I believe it's possible to buy software that processes pictures specifically for the web but put it all together and it's obviously a bit pointless. Some use adobe rgb in the camera too - really intended for printing as I understand it.

John

[Boatman]

I recently purchased an Asus PA238Q and that has pretty much solved my screen to print issues. I found some posts that suggested using the 'User Mode' and changing the color temperature from 6500 to 5500. I tried that and did some other minor adjusting winding up at Br87, Con83, Sat50, Hue50, ColorTemp5500 and Gamma2.2. Using Lr4 print preview adjustments yields prints that are as close as I think you can get to matching the screen, though I still have some issue in the blue-green range.

I loaded Argyll and dispalGUI and ran them with the Pantone Huey. The software seemed to work fine and created monitor .iccs without too much difficulty. However there is no visible difference between the screen image without the .icc and the screen image with it. That leads me to believe that either the Huey isn't working at all or I have my screen already optimally set, which I doubt.

Post script. I got some more time to work with this. I ran the DispCAL calibrate tool on both monitors and let it create profiles. I then opened Lr4 a pulled up an image of a MacBeth color checker cart, which I expanded in Lr so that it showed in both displays. I also put a printed MacBeth chart in my lightbox. I then went into Windows controls to the color management tool. There you can select the monitor and its corrected calibration. Clicking the "Use my settings for this device" toggles the screen between native and corrected color. Results? Both screens show slight changes when corrected and both screens are close to identical in coloration. The Asus screen seemed a little dull compared to the printed MacBeth chart so I increased the saturation from 50 to 58, which seemed to correct it nicely. The other screen does not have this adjustment. I think I may have my colors under control!

I have not had time to run Argyll on another monitor that has known color problems.

[ajohnw]

My monitor was way out. Most shots haven't changed much and still look ok. Where I do see a definite difference is on a few shots that I have taken where the tone and lighting range are extreme and I have had to process them more than usual. They look over done. The worst example is the shot of some black Victorian machinery I posted a link to on here where the detail in the machines had to be burnt in. On other shots tone mapping is more apparent. This looks to be down to my monitor having more contrast now.

Next thing I want to do again via DispcalGui is to calibrate accounting for ambient lighting. There doesn't seem to be any problem simply inverting the calibrator in front of the monitor to read light levels and temperature. Also to produce a straight sRGB profile rather than the extended one I am using at the moment which is a cross between sRGB and Adobe. It makes sense to use the GUI utility rather than driving Argyll directly. It even installs the icc files in the correct place. For system wide I suspect it has to be run in admin mode or as root on linux.

I have finished up in an odd situation when using a colour managed software package. I suspect the video look up tables have been modified when the package is run and that it doesn't correct that aspect. It can't really without colour calibration. I've read that similar problems can occur with some Adobe packages and that it's important to run these packages with the new calibration file.

John

[revi]

Is there any relation between screen/printer profiles (the .icc file you create with Argyll/DispCalGUI)
and the workspace used in editing etc. (AdobeRGB, ProPhoto, sRGB, ...)?

@john:
What might happen from what I read in your post, is that you install a colour correction profile for use
by the system (and thus applied to all screen output) and then you use colour managed software, again
with colour correction active. I'm not sure that software is aware that your system is already correcting
the screen output, so it might apply the correction again, giving you wrong colours again.

You could try to turn off colour correction for the screen (either globally or in the specific software, but
not in both , of course) so you get exactly one colour correction applied.

[ajohnw]

Maybe the colour managed software should now use the ICC profile I have created really not the default one that assumes I have a real sRGB monitor. It's hard for me to be sure.

The management software does take over the package's display. There is no doubt about that because there is a change in images when I load one into it. There is more vibrancy. I suspect this is because installing the icc file has modified the video's cards look up table so that it is linear. The other part of the ICC installation then applies the gamut as it seems to be called - horribly complex subject. The icc profile I generated is the web-office one. I now need to generate a pure sRGB one and see if images then appear the same in all including the managed one. Otherwise it will be hours on the web trying to find out what goes on. They tend to miss basics like how it all actually works.

The earlier comments about prints are interesting. Generally prints are balanced to 5000K, screens to 6,500K which makes any sort of comparison difficult while looking at one against another. When I get round to doing the printer I intend to set it up for 5000K.

If anyone does use dispcalgui remember that the display must be calibrated before the icc profile is generated. It provides 3 sliders to adjust r g and b. They inter react on my monitor so it took 10 to 15mins to get it correct.

John

[revi]

Maybe the colour managed software should now use the ICC profile I have created really not the default one that assumes I have a real sRGB monitor. It's hard for me to be sure.

That's an easy one: always use the profile you created, and not the factory-provided one: each display is slightly different (even within an series of given brand and type).
Your monitor is a 'real sRGB' monitor most likely, but like all devices has some differences between what it shows and what it should show (that's why we profile).

The management software does take over the package's display. There is no doubt about that because there is a change in images when I load one into it. There is more vibrancy. I suspect this is because installing the icc file has modified the video's cards look up table so that it is linear. The other part of the ICC installation then applies the gamut as it seems to be called - horribly complex subject. The icc profile I generated is the web-office one. I now need to generate a pure sRGB one and see if images then appear the same in all including the managed one. Otherwise it will be hours on the web trying to find out what goes on. They tend to miss basics like how it all actually works.

I think you don't quite have the system in place (in your mind).
Basically, we have two things to deal with:
1 - the way we represent 'real life' colours in our images (how do we encode a colour in three channels?)
2 - any device will have errors in how it displays colours (i.e. the colour we see is not quite the colour we should see for a given code)

The first part (encoding) is tackled by the colour space we use (in this case sRGB, but there are others). That's a standard, pre-defined way of describing colours, and we as photographers don't change anything there (or no one else would be able to see the image the same way). The only thing we do is tell the software what colour space we use, and this information is normally stored in the image file (sRGB is a very common choice). Most colour spaces don't cover all the colours we can see (and they differ in what colours they cover), and the range of colours a colour space or device covers is called the gamut.

The second part is purely a correction of device errors, and this correction is specific to the device (display, printer, ...) for which it is made. That correction is the profile you generate with argyll/dispcalgui.
For instance, a given yellow colour is encoded in sRGB as <128,128,10>. Dispcalgui knows what colour should be shown, and measures the displayed colour. As it happens, your screen needs to get a code <127,128, 5> to show that particular yellow. So we need a correction (here <-1, 0, -5>) to the colour as stored in the image to get the correct display. This correction is stored in the profile. This is repeated for 100's of colours covering the whole of the display gamut to create the final profile.

So, what the system does to display an image is:
1 - check if the colour space of the image is the same as the colour space for the device;
2 - if not, translate to the device colour space (and deal with 'out-of-gamut' colours in one way or another), otherwise, do nothing;
3 - then, correct the colours using the profile for the given device;
4 - send the image to the device.
Step three can be done either by each individual programme wanting to display an image, or by the underlying operating system. It should never be done by both (think on what would happen then, remembering that computers are stupid...)

And, it's absolutely normal that you see a change in colours between a display from an non-colour managed programme, and one from a programme that does do colour management. That's the whole point of using colour management... And you don't need to generate a profile for each programme you use, profiles are device specific, once you have one, all programmes can use it. And there is no such thing as a 'pure sRGB profile' that you could generate, sRGB is a colour space, not a (colour correction) profile.
(The profile you generate starts from a known colour space to decide what colour should be displayed, so in that sense a profile is linked to a colour space, sRGB for most displays).

The earlier comments about prints are interesting. Generally prints are balanced to 5000K, screens to 6,500K which makes any sort of comparison difficult while looking at one against another. When I get round to doing the printer I intend to set it up for 5000K.

That's a different subject, so let's not get into that here.

[/quote]If anyone does use dispcalgui remember that the display must be calibrated before the icc profile is generated. It provides 3 sliders to adjust r g and b. They inter react on my monitor so it took 10 to 15mins to get it correct.[/QUOTE]

Normally, dispcalgui should indicate that you have to calibrate before profiling. But the really important part is not to calibrate after generating a profile: the profile would become worthless.

Remco