Welcome to TINA CHeP :: View topic

DrNo · Joined: 03 Jan 2009 Posts: 1063

This is actually a suggestion for the TC-RTC.

What we try to do in this competition is achieving the best
graphics experience.

However we have the calibration issue. What one sees
is not what others see. The calibration can even be part
of either concept, artistic, and technical. It is much more
important than one can think at first.

Some of us had this problem in the last round. I wanted
"Earth-born" (and Chris wanted "Pocket Dragon") to be very
dark. But if it happens that the reviewer has a very high
brightness, the visual result would not look the same at all.

Of course it is impossible for us to be perfectly synchronized.
I cannot post an image here and say "please synchronize
your monitor according to that image", because I may not
have a proper calibration myself.

However whatever calibration would do it, because the
point is only for us to be synchronized. But it is also better
not to have to change our settings too much.

So I propose the following solution. A simple calibration section
could be added to the TC-RTC site. In this section, there would
be about 10 calibrations of a given image (a photograph or
whatever). Let's say the darkest gamma up to the brightest one
(0.5 to 3.0) from image #1 to image #10.

And then people will simply select (through a pop-up menu)
the image that suits best the monitors they are used to work
with. After 10-20 votes, we will get the perfect average
calibration. If the democratic choice is image #4, then
people who didn't choose that image would have to adjust
their monitors for this image #4 to be their new best.

If for me image #8 was the best, then I would probably
need to change the brightness (and/or gamma and/or contrast)
a lot to be compatible with image #4 (which was way too
dark for me).

One interesting point here is that we won't necessarily get
the same calibration. Indeed a bias is introduced: the taste!

If one likes slightly brighter images, this fact will be
considered somehow in the process. As if I personally
like darker images, then tuning them according to my
taste will automatically fit everybody's taste.

Because I tuned my monitor according to image #4,
but also according to my taste in the first place. Since
everybody did the same, we have a perfect match.

What do you think?

I don't know how easy it is on Windows, but on Macintosh,
it is straightforward to generate a custom monitor profile.

I would like to create a TC-RTC profile, and tune all my
further entries according to that profile. And, as a result,
I would enjoy to stop worrying about it.

I think a good calibration reference sample could be
"Fire and Ice" from Angela in the Contrast round.

Only registered users can see links on this forum!
Register or Login on forum!

I currently don't see the stars in this image, do you?

_________________
DrNo

Ive · Joined: 18 Jul 2009 Posts: 34

DrNo wrote:

This is actually a suggestion for the TC-RTC.

What we try to do in this competition is achieving the best
graphics experience.

However we have the calibration issue. What one sees
is not what others see. The calibration can even be part
of either concept, artistic, and technical. It is much more
important than one can think at first.

(...snip a lot of stuff)

The WWW and JPEG images (without icc profile) are per definition in sRGB color space and if someone is not able or willing (for whatever reasons) to adjust his system to this fact what makes you think he would calibrate his monitor to some private and incompatible 'color space' only used in the small niche called TINA CHeP?

I would certainly not because my system is already calibrated as well as possible.

DrNo wrote:

Only registered users can see links on this forum!
Register or Login on forum!

I currently don't see the stars in this image, do you?

I do.

-Ive

DrNo · Joined: 03 Jan 2009 Posts: 1063

Ive wrote:

The WWW and JPEG images (without icc profile) are per definition in sRGB color space...

That helps a lot.

In my system, the following profiles are available:

1 - Adobe RGB (1998)
2 - Generic RGB profile
3 - iMac calibrated
4 - sRGB IEC61966-2.1

I used to work with "iMac calibrated" which seems
to be tuned with gamma = 1.

It is probably too dark. So when I adjust my images
according to this profile, they appear too bright for
most people.

On the other hand, there is no visual difference between
the sRGB and Adobe RGB profiles, so I guess they
are more "standard".

Unfortunately my pictures don't look very good with
these profiles, but it is too late.

At least, your help was enough for me to make the decision
of always using the sRGB profile from now on (does the
IEC61966-2.1 ring a bell?).

I actually see more of the stars of "Fire and Ice" with this
profile, but it is still too dark. I really feel that Angela's
monitor was too bright. So it looked perfect only for her.

In this "competition", what is important is only the authors
and the reviewers to be synchronized. So instead of a
"democratic" calibration settings, we can simply go with
a standard calibration (which is already democratic anyway),
but always the same: let's say sRGB.

Then I propose to simply add a TC-RTC FAQ pointing to a
page addressing the calibration issue. This page would
contain a few images showing some extreme situations,
such as "Fire and Ice" and "The Challenge" (from the
same round).

However, such images (or any other) may not be tuned properly
in the first place.

So the plan would be:

1 - we choose a few templates, and some of us take a few
minutes to tune them using the sRGB profile.
2 - if we get roughly the same visual results, we use
these templates as references.
3 - then people would be asked to choose a display profile
such that the templates look best to them.

So the FAQ would look like:

Q: Is there a monitor calibration for me to have the
best graphical experience?

A: Yes. First select the sRGB profile from your display preferences.
Then adjust your monitor brightness such that the images
of the calibration page look best to you. If it happens that
you don't see any sRGB in the profile list, the templates can
still help you for choosing a proper calibration or creating
your own.

As simple as that. The more people will be on the same page,
the more we will see what the authors want us to see.

_________________
DrNo

clipka · Joined: 04 Nov 2008 Posts: 254

DrNo wrote:

I used to work with "iMac calibrated" which seems
to be tuned with gamma = 1.

I'd imagine that to be 1.8 actually.

Stephen · Joined: 03 Nov 2008 Posts: 566 Location: London

DrNo wrote:

Ive wrote:

The WWW and JPEG images (without icc profile) are per definition in sRGB color space...

That helps a lot.

In my system, the following profiles are available:

1 - Adobe RGB (1998)
2 - Generic RGB profile
3 - iMac calibrated
4 - sRGB IEC61966-2.1

I used to work with "iMac calibrated" which seems
to be tuned with gamma = 1.

What I do, is to copy each image into my paint program and adjust the gamma until it looks right for me. Then I score it.
OK it is a bit of work but so is scoring. I can also rescale so I don’t have to pan the screen with big images.

_________________
Best Regards
Stephen

Ive · Joined: 18 Jul 2009 Posts: 34

Gamma correction, monitor calibration, color management, oh well I'm so bored of it all!

Ok, this will become a rant, but since almost 15 years I'm so frequently confronted with these issues and have thousands of times tried to explain them, mostly without any success. I indeed remember exactly 5 people (which makes 1 person per 3 years!) who did listen, understand and we were finally able to solve problems.
The strange thing is, you'll find misconceptions about these issues even in the high-end professional computer graphics industry and, believe me, things get really ugly in the printing business. Even otherwise intelligent people can become quite stubborn when it comes to color science. Meanwhile my believe is: besides of (color-) blind people human vision and the perception of color is so fundamental to humans that everybody thinks he already has some kind of fundamental knowledge how vision/color works. But in fact this 'knowledge' is in many cases based on wrong assumptions.
Also the usual "do a google search" will not help in this matter because about 80% (just an educated guess) of the sides you'll find do not state things right. Well usually they are not plain wrong but contain a mixture of lot's of true facts combined with some false ones. And if one does not already know the subject well how could he decide what's right and whats wrong?

Now your question might be: why should you trust me? Well, hmm, given my own statements from above I can give you in fact no good reason why I should be more trustworthy than others. Simply saying because I know what I'm talkin' about will not help because everybody seems to claim that. Anyway, here are a few links to pages of people *I* do find trustworthy - with one of them I do have a frequent e-mail correspondence, one other I do know personally:

First some quite scientific ones, so be warned:

Only registered users can see links on this forum!
Register or Login on forum!

The entry page and here is a direct link to some pdf documents where in this context mostly the ones under Computer Vision are of interest:

Only registered users can see links on this forum!
Register or Login on forum!

German, but here are direct links to the English versions of 'Farbenlehre 1 & 2'

Only registered users can see links on this forum!
Register or Login on forum!

Personally I do find this one extremely amusing but it is only available in German:

Only registered users can see links on this forum!
Register or Login on forum!

And now one page that might assist you in calibrating your system:

Only registered users can see links on this forum!
Register or Login on forum!

Note that there are many pages that do claim to do so but also many of them are just crap.

Some random notes:

@Christoph: I've just seen your discussion with Warp in the POV-Ray NG. I had quite similar ones with him at least two times and he's still repeating false claims and simply does not listen, so welcome to the club.

@Normand: Your submitted JPEG images do include sRGB ICC profiles but this should cause no problems when exchanged over the web because the WWW is per definition of the W3C in sRGB color space. Assuming 3 kinds of software we have:
a) a 'brain dead' piece of junk that does not recognize the profile but this does *in this case* not matter.
b) an 'intelligent' application that does recognize ICC profiles and applies a transformation from sRGB to sRGB which will result in exactly the same image data but is quite computing intensive (and might introduce small rounding errors).
c) something really smart that identifies the profile as sRGB and decides therefor no transformation is needed at all.
No need to mention that *most* software out there would be category (a).

@Thomas: I think to remember you are using PaintShop Pro. So just in case: the 'monitor calibration' part of this software is one of the crappiest I've ever seen and will simply give plain wrong results. Do not trust or use it. (Just to be clear: I'm not talking about PaintShop Pro itself, which is - given it's price - really good, but *only* about the 'calibration' part).

@Christoph: you are using the terms whitepoint and blackpoint quite often but you are using them wrong! Both do refer to a color temperature based on Planck's blackbody radiation (or the similar CIE D illuminants being slightly different because the perceived 'daylight' is different from an ideal blackbody) and NOT to the intensity. What you mean is simply maximal/minimal brightness or luminance. Blackpoint and blackpoint compensation is usually only of interest in the printing business. Please use these terms correctly, it will make look your arguments better Wink

@Normand: I'm pretty sure your system/monitor does also use a gamma 2.2 or sRGB setting as any other contemporary Mac.
The Mac 1.8 gamma is one of the myths that seems to live on forever. Well it once was true, but so many, many, many years ago that I do not even remember when exactly things did change. If you think you still have problems with your calibration feel free to mail me directly and I will try to assist.

-Ive

DrNo · Joined: 03 Jan 2009 Posts: 1063

What can we make of these tests?

Only registered users can see links on this forum!
Register or Login on forum!

Is the "black point" refering to the right thing?

_________________
DrNo

Ive · Joined: 18 Jul 2009 Posts: 34

DrNo wrote:

What can we make of these tests?

There is nothing wrong with them. Quoting the side my monitor would be "A top quality monitor using an excellent calibration system" so I'm not sure how useful those tests are in assisting you in case your system is not already well calibrated.

DrNo wrote:

Is the "black point" refering to the right thing?

They say: "If you see a color cast to the darker patches, this indicates a problem..." so yes, they are also talking about the 'color' of the blackpoint but they do not show something that would help you to identify problems with the color of the blackpoint. Therefor I would prefer to simply call it black level or min. luminance.

My impression: it will help you in indicating miscalibration but will assist not that much in actually selecting the best brightness and gamma values for your system.

I've just seen that the side I was referring to for monitor calibration is currently unreachable so here is another one that has at least a good gamma correction pattern:

Only registered users can see links on this forum!
Register or Login on forum!

DrNo · Joined: 03 Jan 2009 Posts: 1063

Ive wrote:

DrNo wrote:

Is the "black point" refering to the right thing?

They say: "If you see a color cast to the darker patches, this indicates a problem..." so yes, they are also talking about the 'color' of the blackpoint but they do not show something that would help you to identify problems with the color of the blackpoint. Therefor I would prefer to simply call it black level or min. luminance.

I think I got it. The black (or white) point says it all. It is the
RGB values of black (or white). So when we talk about intensity,
it refers only to the "V" of the HSV model.

One can decide that the black point is a dark brown, or a dark
purple, etc. Similarly, the white point could be a light yellow, etc.

So using "black level" or "min luminance" is indeed much less
confusing I agree.

Ive wrote:

Only registered users can see links on this forum!
Register or Login on forum!

Of course! Now I got it perfectly.

The gamma correction allows the monitor to display
50% gray for it to look exactly like half black/half white.

Because 127 (out of 255) may not look like 50% gray.

So the test simply uses a dense checkered pattern, half
the pixels are 0 (black), while the other half is 255 (white).
This pattern will look exactly like 50% gray, because
the eye sees the full intensity half of the time.

Then we compare this perfect 50% gray with another
one, constant for all the pixels.

The gamma will fix that uniform value which could be
127, or 112, or 146, etc...

It was about time I figured that out Embarassed

_________________
DrNo

Ive · Joined: 18 Jul 2009 Posts: 34

DrNo wrote:

I think I got it. The black (or white) point says it all. It is the
RGB values of black (or white). So when we talk about intensity,
it refers only to the "V" of the HSV model.

One can decide that the black point is a dark brown, or a dark
purple, etc. Similarly, the white point could be a light yellow, etc.

Err, yes (and I do not want to add confusion, really... Confused

) but actually the white/black point refers to its position within the CIE xyz space being a device independent color system, where RGB (and derivate like HSV) or color separations (like CMYK) are always device dependent.

I've written a little demo for POV-Ray (part of lightsysIV) that does a nice job in visualizing the relationship between blackbody color temperature, CIE D illuminants (often used as a shortcut for whitepoints) and sRGB values. The output image looks like this:

Only registered users can see links on this forum!
Register or Login on forum!

The CIE daylight is not defined for temperatures below 4000° Kelvin. And as a side note the surface temperature of our sun is about 6500° Kelvin.
Note that this image refers to sRGB color system and the rgb values for e.g. Adobe RGB would be different. As already mentioned, RGB is always device dependent.

DrNo wrote:

Of course! Now I got it perfectly.

DrNo wrote:

So the test simply uses a dense checkered pattern, half
the pixels are 0 (black), while the other half is 255 (white).

No, it doesn't. As the page itself states a checkered pattern is never a good idea, not even for lcd displays and it will not work on a crt at all.

I also do disagree with his reasoning for a 1.8 system gamma and also the myth of a Mac 1.8 gamma is stated again (one thing they did get right on the sides you did post links to).

What I do find helpful is the pattern itself - but use only the 2.2 gamma one!

-Ive

DrNo · Joined: 03 Jan 2009 Posts: 1063

Sure... whatever 3D color space...

And sure... interlaced instead of checkered...

_________________
DrNo