CRT Monitor Calibration and Gamma curves. (For those who still have CRT Monitors)

[robski]

OK I've decided to put my engineers hat on again to investigate some quirks with CRT Monitor calibration. mainly to understand a number of points brought out by various websites giving advice on the subject. Initially to look into the pitfalls of using adjust by eye systems such adobe gamma. At some point I intend to get myself a calibration tool such as Spyder 2 and compare the results.

In a previous life I was involved in Television engineering. Gamma encoding was done at source ( TV Camera ) so we only had to worry about CRT cut-off (dark level ), gray scale, geometry and convergence adjustments.

I intend to try and keep it simple so that you can follow the gist of the subject. I have no experience of flat screen setup and I am sure they have their own set of problems.

The 2 areas I want to focus on are Gamma curves and the Brightness control.

To start with some background information as to why we need gamma correction. The CRT is an electronic device that goes back to the days of steam radio. In the UK we called them Valves in other parts of the world they are known as vacuum tubes. The little glass bottles that glowed in the dark in granny's radio. As a schoolboy I used to build valve radio projects given in the Practical Wireless magazine. They were the days.

Next a 30 second electronics course. One of the simplest Valves is a triode. As the name suggests it contains 3 basic components.

1, a heated cathode which emits electrons, connected to the negative voltage supply.
2, an anode connected to high a positive voltage supply to attract the electrons towards it.
3, a control grid that is positioned between the cathode and anode. By applying a small negative voltage with respect to the cathode the flow of electrons can be controlled.

Valves were typically used as amplifiers. The input signal applied to the control grid and an amplified signal taken from the anode.

In a CRT the electronic part is called an electron gun. The main difference being the device has a couple more anodes. In a colour CRT the anode that is used for focusing has typically 3000 Volts applied and the final Anode has 25,000 Volts applied. The idea is to accelerate the electrons so that they bypass the anode and hit the face of the CRT. The face of the CRT is covered with phosphor dots which illuminate when struck by the electrons. The screen brightness is dependent on the volume of electron flow. The volume is controlled by the video signal and brightness control circuit.

Unfortunately the electron flow is not linear over the whole range of operation with valve and transistor devices. Typically the first 40% of control is curved and becoming more linear ( approximating to a straight line ) for the second 60%. The second 60% is the region that audio and video amplifiers are interested in to avoid distortion.

The attached CRT Characteristic image illustrates a typically transfer curve. The input being the controlling signal and the output being the amount of light produced. 2 points to note. First we don't get into the linear section of the curve until we have pasted the 20% light output stage. Secondly the first 50% of input produces only 20% of the light and the last 50% of input produces 80% of the light.

If we were to adopt a linear approach as with audio designs our dark point would be at 20% of the CRT brightness. i.e. the black would be a dark gray. For our blacks to black on the CRT we need zero electron flow and we have to live with this non-linearity. This is where gamma encoding is used to compensate. The dark parts of the image are boosted in such a way to linearize the brightness range. The CRT itself is the gamma decoder.

The shape of the curve has a logarithmic function and the clever mathematicians can approximate these curves with a function of x^y i.e. x raised to the power of y. x being the input signal and y would equal 2.2 to give this shape curve.

2.2 being the gamma factor.

It would appear that everybody ( including Mac users ) have settled on using a gamma value of 2.2 as the standard.

I trust you have all manage to keep up with the technical stuff.

The next attachment (Video brightness levels) illustrates the effect of applying the gamma encoding by video card driver. On the left is a step wedge of equal pixel brightness steps (original image data) and on the right show the steps with a gamma 2.2 encoding applied. The darker levels are boosted and the brighter levels are reduced in amplitude. The effect seen on the screen should a step wedge of equal brightness steps.

In an ideal world the first level of pixel brightness should coincide with the brightness control black level cut off point. If you have a black desktop and the brightness control set to maximum you may see a faint glimmer from the phosphor dots. The black level cut off point is where the phosphor stop glowing as you reduce the brightness. In theory this is the darkest black the monitor can produce. Any brightness seen will be the result of ambient room light reflected back off the face of the screen. Some Manufacturers have developed screens to reduce the reflection factor to give richer blacks and improved contrast ratio.

Part 2 to follow

[robski]

As mentioned in the previous post in an ideal world the first level of pixel brightness should coincide with the brightness control black level cut off point.
The first attachment shows the ideal where the first level of pixel brightness coincides with the brightness control black level cut off point. The lower part shows the video data gamma encoded.

Running the Monitor Black Point Check from this Link with no Gamma correction applied and brightness set to maximum.
http://www.drycreekphoto.com/Learn/C...itor_black.htm

It is common not to see any activity until the pixel luminance level reaches 8 or more which equates to a luminance level of over 3%. So in the real world the first pixel luminance level is below the CRT black level cut-off and therefore will never be seen.

Now if we were to use an excessive gamma encoding to achieve the viewing of the 1% box in this link we can end up with a host of problems.
http://www.pbase.com/image/67769565

The second attachment illustrates applying excessive gamma encoding to achieve viewing shadow detail in these dark areas.

In the last attachment I have plotted the shape of different gamma encoding curves.

Gamma 1 is a straight line.
Gamma 1.8 out dated Apple Mac standard
Gamma 2.2 PC now accepted as the de facto standard
Gamma 2.6 just to illustrate the effect of increasing the value

Note the steepness of the initial part of the curve for the last 3 curves.

The difference between gamma 2.2 and 2.6 is a 24% gain in pixel brightness for input pixel brightness of 5%. The mid grey (50%) would be 3.5% brighter. When applying excessive gamma correction the viewed image will exhibit some of the following symptoms. Posterisation, banding and increased noise artifacts in the shadow detail. The image will appear too bright and lacking contrast with an over sensitivity to small brightness edits.


Part 3 will study the workings of the Adobe Gamma method

[robski]

Colour Management became the buzz word at the turn of the century. Computer operating systems and hardware began to support the concept of profiling using ICC profiles and CMM (colour matching method) ICM (image colour matching) engine. A number companies such as Kodak, Colosync and Adobe who are members of the ICC profile consortium developed these engines. Initially they were used only in applications such as Photoshop. Apple and Microsoft then took these engines on board and used them within their operating system.

So if your still running Windows 95 or Ancient Hardware tough luck colour management not supported. I can't remember how much Windows 98 supported it. So you need Window 2K or XP onwards and If I recall correctly Mac OS 8 onwards. But beware I did suffer a Windows Service pack that disabled colour management in my graphic card driver. No rhyme or reason and plenty of people complaining about it, I had to get a different graphic card to solve the problem.

A lot of photographers are using Adobe Photoshop. When Version 5 and above is installed it installs the Adobe Gamma into the Windows Control Panel. This utility allows you to calibrate your monitor. The more you study this utility you soon realise how crude and clunky it is. But Hey it is better than nothing. Some Graphic cards provide their own adjustment utility but next to no help of how to correctly calibrate.

If you have bought a High end monitor you should have been supplied with ICC profiles for it or be able to download from the manufactures website. The profile should have some setting your are unlikely to know or have to resort to the product manual for.

Adobe have supplied some detailed instructions on how to use the Adobe Gamma utility.
http://www.adobe.com/education/pdf/c.../ps7_cib17.pdf

I don't want to go into such detail but just to summarise and highlight a few points

After opening the utility the first stage is to select a monitor ICC profile. If you were supplied with one for the monitor then use that one.

Otherwise the two most common profiles are sRGB IEC61966-2.1 and Adobe RGB1988

This link goes into the merits of each version http://www.cambridgeincolour.com/tut...obeRGB1998.htm

Basically Adobe RGB has a wider colour range and is best suited where images are destined for a CMYK printing press.
The sRGB has a narrower range of colour and has become the de facto standard for web and photo lab work.

So if you are clueless pick the sRGB profile.

Next set the contrast control on the monitor to maximum. The contrast setting has little effect on the gamma adjustment.

The next step is to adjust the monitor Brightness (black level) control. The instructions on this are not very precise and crude.
The brightness setting has a much bigger interaction with the gamma adjustment than the contrast setting.
Basically the instruction is to make the square dark but not too dark !!! whatever that means.
What they are in fact doing is presenting you with a black square (0r,0g,0b) and a dark square (38r,38g,38b) and asking you to adjust the brightness to make sure that one is seen lighter than the other. Purely subjective as to how much brighter you make it.

If you have used a profile supplied with the monitor the phosphors used should be know within the profile. The monitor manual may contain the information. Otherwise select one of the two most common ones used are EBU/ITU and Trinitron as the average person will not have an instrument to measure the phosphor colours. An easy way to tell if a Trinitron CRT is used is to carefully look 25% down from the top of screen and 25% up from the bottom. You will see a very thin faint black line. This is a very thin wire that is used to support the shadow mask behind the glass. The first attachment illustrates the phosphor colour positions in the sRGB colour space within the ICC profile.

The next step is too select the desired Gamma value. Normally 2.20.
What happens here is that the mid grey center square is encoded with a Gamma value. A mid grey (128r,128g,128b) becomes (186r,186,g,186b). The outer square is made up of lines which are black and white. The average of which will be mid grey (128r,128g,128b). These lines are not affected by the adjustment when you adjust the slider to match the center square to outer box.
Again this adjustment is crude and subjective, only the one point (50%) on the curve is matched. Adjusting individual colour gamma by eye could be problematic and lead to colour shifting through the brightness range.

Lastly the white point adjustment is somewhat subjective and again you are unlikely to have an instrument to measure it. The last attachment illustrates the white point in the sRGB colour space within the ICC profile.

[andy153]

Thanks Rob for all this effort - it explains a lot and shows me that anything like a Spyder 2 Pro or similar calibration device can make things a lot simpler. By the way, I'm on a Mac and installation of the Spyder 2 Pro automatically disabled the Adobe Gamma.

[robski]

Thanks Andy

I have a Spyder on order - it will be interesting to see how it compares.

In the meantime in an idle moment I have knocked up an extended gamma checker to cover 20% 33% 50% & 66% luminance levels for a Gamma value of 2.2

You may need to stand 2 or 3 metres away from the screen to use it properly. You could view this whilst doing the Adobe Gamma adjustment.
The inner squares should view the same as their surrounding area.

In my case all tones seem to view well after a minor tweak.

[robski]

With these 2 you may need to stand back 3 or 4 metres to see if they look OK.

10% and 5% gamma 2.2

Don't go by the thumbnails as the scaling has distorted the tonal range

[robski]

I am now experimenting with my new Spyder calibration tool and will report back once I've got to grips with it and tried out on several PCs.

[barrypriddis]

Rob I will be very interested in your results. I have a flat screen monitor and had problems matching my prints with what my monitor was showing. I have purchased a spyder 2 and used it to calibrate the monitor, which did give an improvement. However, I find my prints are still a bit darker than the monitor which is somewhat annoying.
I will be changing my monitor soon, and would appreciate any advice on what would be a good buy. My budget is limited to around £250.

[Gidders]

Originally Posted by barrypriddis

Rob I will be very interested in your results. I have a flat screen monitor and had problems matching my prints with what my monitor was showing. I have purchased a spyder 2 and used it to calibrate the monitor, which did give an improvement. However, I find my prints are still a bit darker than the monitor which is somewhat annoying.
I will be changing my monitor soon, and would appreciate any advice on what would be a good buy. My budget is limited to around £250.

In answer to your monitor question, have a look at the Colour Confidence web site - there are a couple of Eizo entry level monitors that they recommend that are just about within your budget depending on what size screen you are looking for. I used to have a (cheap) flat screen but it was pants for colour critical work so I went back to CRT ... but I'm saving my pennies for an Eizo 24" wide screen which start at £586

To get your prints to match your monitor output you have to calibrate not only your monitor, but also your printer. The problem with calibrating printers is that it is both ink and paper dependent ie if you change either your ink or the paper, the results will be different. The solution is to

1. Always use cartridges from the same manufacturer, and I would suggest that should be OEM - true their inks are more expensive but they are colour consistent from batch to batch, where as, in my experience, cheap after market inks are not.
2. Have a profile written for each different paper that you use

Sometimes you'll find that one profile works well for a couple of papers by the same manufacturer for example I find I can use my profile for Jessops Gloss for Jessops Satin ... but not for Jessops Matt. Similarly my profile for Ilford Galerie Satin works for their Gloss & Permajet Papyrus works for a couple of their other fine art papers. So how do you get printer profiles.

Colour confidence can sell you a spectrophotometer which will do the job for you ... for about £500. The alternate if to use a profiling service. This involves printing a test card at know defaults and then sending it to someone with a spectrophotometer who creates the profile for you. Most of these seem to cost ~£30-£40 per profile. I have a contact who used to offer a very good service at a much more competitive rate & I've emailed him to see if he is happy for me to pass his details on. Another source is Native Digital who I bought my Spyder 3 from. They actually dissuaded me from buying a spectrophotometer on the basis that, unless I wanted to do loads and loads of different papers, they offer a remote service for £15 (I think it was) per profile - & you can have a lot of £15 profiles created for £500 That said I haven't actually tried them yet but planning to send a couple off this weekend

[robski]

I guess the viewing angle affecting the image with flat screen monitors is the biggest hassle. It would be interesting to know what real differences are between the Spyder3 and Spyder2 sensors. I suspect improvements to cope with flat screen problems.

As Clive has mentioned the printer profiling maybe part of your problem. If you are using Photoshop you should be able to view the effects of the printer profile as a soft proof on screen. All ICC profiles have a forward path for conversion. e.g image colour to printer colour and image colour to monitor colour. The printer profile has reverse path that can be used for soft proofing on screen.


In Photoshop CS goto the View menu

Proof Setup -> and select the custom option. In the pop up dialog navigate to your installed printer profile and select. You would normally use the relative intent. (Note some profiles use the same data for all intents instead of creating individual intents. In which case your not see any difference with different settings)

You can turn the view On or Off by using Ctrl + Y or ticking or un-ticking proof colours under the View menu.
Note the extra info on the image menu bar when it is switched on.

Another factor are you using the colour management option correctly when printing ?



I am currently comparing the options available within the 3 versions of Spyder2. (Spyder2 Express, Spyder2, Spyder2 Pro)