Standard Observers – L* Compensation

[Lorna Wiles]

Is an L* compensation gamma applied to files before they are displayed on a monitor?

Our brains, assuming we are all standard observers, effectively apply an L* profile to the light entering our eyes. So if you profile a monitor with linear gamma (native gamma) and send a file to the monitor with also has a linear gamma (gamma 1.0), the apparent result should look correct - file displayed with linear gamma (gamma 1.0) eyes apply L*, result, tonally correct looking image. BUT it doesn't look correct it actually looks (mathematically) like a gamma 1.0 file. This means that under the hood somewhere an L* compensation gamma is applied before, during or after the output profile passes the file off to the monitor for display.

The discussions over what a monitor should be profiled to in order to reduce banding would seem not to take this effect into account. Could anyone explain?

 

[Lukas Engqvist]

Banding occurs usually because of colour conversions having to round off numbers. I don't see how it has to do with gamma.

Where do you get native gamma being linear? That no gamma conversion takes place does not make the gamma linear, only the CONVERSION linear. So for no banding to occur one could argue not to convert because then you will not get any rounding off of values, but the natural gamma of a monitor can vary, not unusual for an LCD to have a natural gamma of 2 or more.

I don't quite understand what you are trying to achieve. The story of linear gamma is almost the same as that of a linear plate, or a linear print. Actually it is interesting that a gamma of 1.8 (traditional gamma for Mac) is almost the same as a TVI of an offset press, but that is just curious.

 

[Lorna Wiles]

Thanks for the reply Lucas, it’s appreciated. I will attempt to explain my confusion a little more clearly. Instead of considering the output of the monitor I’ll try and explain in terms of the output of the press.

When profiling a press printing in a single spot colour, I use an untagged greyscale target. I then output the target (either untagged or assign a gamma of 1.0 or dot gain of 0 – net result no change) measure the printed result and create a profile to compensate for the dot gain. If I then again print the greyscale target using the new dot gain profile, the result should now be a greyscale target that ‘looks’ correct – using the new profile the press is now linearised.

However my understanding of the way our eyes see in L* is that an 18% grey should ‘look’ like a 50% tint. It doesn’t, either in the printed result or on the monitor. It ‘looks’ mathematically correct, as if our eyes were not viewing in L*. Am I doing something wrong here?

Thanks

 

[Lukas Engqvist]

I don't know if I am understanding you correct, but I remember being confused myself several years back. Please don't take offence if I am stating what you already know, if I missunderstand you maybe someone else can clarify.

A 50% tint by definition (due to the baggage of tradition) is assumed to be 50% at film, and in modern workflow is assumed to be 50% at the plate. There is no place in the procces exept when doing mathematical colour functions in photoshop that one would find a gamma of one (to my knowledge).

You are talking of both spot colour and of "grey"?

You will probably find what you want is NOT a linear press but a press that matches the standard dotgain for the process you are matching.

What dot gain are you getting? If you are getting a dot gain of say 10% @50% your target is 50%+16% and so you will need to add 6% in your curve (well it will probably be a little less because you will need to plot your curve find out where you measured 56% and then back trace it). So it doesn't have to do with how our eyes see it, but rather what we have decided to call 50%. (you can compare it to how there is a difference in where 0°C and 0°F is if you like…*it is a linguistic problem of 50% being an abstract that no one actually intend you to see in print and not 50% of the paper covered in ink)

 

[Lorna Wiles]

Lucas, I would not take offence at your explanations, in fact the complete opposite - I’m grateful for your replies!

To answer your questions;

The dot gain we get on press is high, around 30% as we are screen printers.

Apologies for confusing the terms “spot colour” and “grey”. If working on a ‘single colour’ image in PS, I usually do this in greyscale mode as opposed to a single spot channel simply because more tools / menu options are available.

When working in greyscale mode and assigning greyscale profiles PS does not offer selections for custom gamma / dot gain, only the ones it comes preloaded with. However it does allow the current work space profile to be assigned. By first loading a custom greyscale profile into the work space you can then assign (or convert) any profile you choose, for instance the press dot gain profile, or even a gamma 1.0 or a 0 dot gain!

In my example the reason for assigning a gamma 1.0 or 0 dot gain is so that you are ‘seeing’ the original untagged file in it’s untagged form, the raw numbers, as opposed to what ever default gamma your work space in PS is set to, perhaps more commonly 1.8 or 2.2

So, with a linear greyscale target (all tint percentages reading correct K % with the dropper tool and info pallet in PS) which has a linear gamma (gamma 1.0) printed with an accurate dot gain profile for the press (linearised press) one would assume that an 18% printed halftone should ‘look’ like a 50% tint due to L* of the observers eyes. It doesn’t, and I don’t understand why.

Apologies if I am not explaining myself clearly, and thanks for your input. As previously mentioned, it is appreciated.

 

[Lukas Engqvist]

In Photoshop you can set, in the colour settings the dot gain for the grey and for spot colour.

If you select (in colour settings) you can even create a custom dot gain (or if you prefer custom gamma for grey, wich you can save and then use for spot…*but spot is usually not measured in gamma) and save as a curve.

If you are trying to get to grips with your process, it is important to measure each step so you can know where to back track if you make a mistake.

I do think "seeing" the 1.0 or zero dot gain is of no use to you. It is like placing a "midway" target that you shoot through on the way to the final target. IMHO it has no practical value. If you have a dot gain of 30% and you want to leave those values unchanged you will best simulate on screen by choosing in your colour settings 30% dot gain in the "spot color" field in the color settings.

If you do want 50% of a colour to print at 50% with a dot gain of 30% your plate will need to imaging 25%. That means that all your tints between 0-50% will be in your file 0-25%. 50% on plate (screen) will be 80% in print, at gamma 1.0 in print that means 80% in the file needs to be reduced to 50% on the plate (screen) so that it swells to 80%.

It is therefore better that you just say "how will 50% look in real life?" Apply a screen simulation of the 30% dot gain and evaluate, lighter or darker. (don't worry bout gamma or L* if your monitor is calibrated it will do the right maths) The monitor DOES NOT show the same as 50% ink coverage even though the pipette will show you 50%, it shows 80% (50+ dot gain).

The other option is to say in your flow. "I want to be able to use a standard profile!", the advantage being that your CMYK values will look about the same as other output devices (Well assuming the pigments are compatible, and they mix in a similar way). In this case you may decide you want to target 13% for CMY and 16% for K and Spot. This means that in your rip you would make a compensation.
If we tale the example of the 50%. You are now in a spot colour wanting 16%. So in Photoshop you will choose this as your Spot color profile. You measure 50% with the pipette, and the screen simulates 50+16%= 66% ink coverage. But if your plates (or screen) are outputting 50% then you will still be getting 80%, your screen is showing a lighter image than you are getting, so the RIP will need a transfer curve that will calculate the difference. Aprox 40% on plate would give you 70%, so if to thumb it you will need a curve saying that at 50% you want 14% less dot on plate than in the file.

Now it does get a little confusing when you are talking about 18% looking like 50%. When you say 50% are you talking light or area of the substrate covered with ink, or the "mid point" between white and black?

 

[Lorna Wiles]

Thanks Lukas

I’m going to need a little time to work my way through your explanations. I’m beginning to realise that I may be confusing the process that the RIP performs as opposed to that of the editing software – in my case Photoshop.

The 18% I’m referring to is the "mid point" between white and black, the point at which the tone ‘looks’ like a 50% tint.

I think I need to work carefully through your advice whilst profiling the press, and as you say ‘measure each step so you can know where to back track if you make a mistake’

Thank you for your help. I am grateful to knowledgeable people such as yourself that freely give of their time on this forum to help others.

 

[rich apollo]

Lorna, I think you're talking about "middle gray".

"In photography, painting, and other visual arts, middle gray or middle grey is a tone that is perceptually about half way between black and white on a lightness scale;[1] in photography, it is typically defined as 18% reflectance in visible light.[2]…

As early as 1903, middle gray was defined as the geometric mean intensity between a white and a black intensity that are in a ratio of 60:1.[5] That is equivalent to 12.9% of the white intensity." Middle gray - Wikipedia, the free encyclopedia

Middle gray doesn't correspond directly to screen values. Neither does the concept of gamma. Gamma, and middle gray, are much more applicable to density, when talking about print.

In monitors, gamma describes the ratio of power, or pixel value, to brightness. More voltage produces more brightness, but the relationship is not 1:1. The native gamma of a monitor is generally not 1.0.

In Photoshop build a grayscale step wedge. Set your info pallet to read Lab. Now, assign different profiles to the image and see where L* of 50 falls, noticing how many levels of gray are to each side.

 

[Lorna Wiles]

Rich, thank you. Now I am starting to get my head around this.

It’s a case of the more that you think you know the more you realise that you don’t. It obviously requires a bit more research on my part to broaden my understanding.

Between your good self and Lukas you have pointed me in the right direction for which I am grateful.