About gamma correction ...

[AltayHan]

Hi to all ...

I would like to ask about gamma correction ...

On the "understanding gamma correction" page, it is showed that system gamma is 1 (that is, it is linear) ... But, in the beginning, we have a raw file with gamma 1 (a linear file) ... after including the color space profile and then the display gamma correction, we're again coming to our starting point ... a linear curve.

I thought that the purpose of gamma is to give the non-linear response of the human eye to our raw files. But in this case it seems that we're not reaching to that point ... we're just standing at the our starting point.

I have spent so much time for understanding this gamma issue on the net, so I would be very thankful, if a professional explains the story about the gamma.

Thanks a lot ...

[Dave Humphries]

Hi there,

Have you seen the tutorial here?
https://www.cambridgeincolour.com/tu...correction.htm
It may help.

Could you do me a favour please?
Could you Edit your Profile and put your first name in the Real Name field and where you are (roughly) in the Location field? - thanks.

Welcome to the CiC forums from ...

[AltayHan]

Hi Dave,

Let me change my profile first of all ...

And ...
Ok now,
We can go on ...

I've already read the page you sent maybe 3 times and some other stuffs on other sites.

Everything is clear except one point ...
And this is what I tried to ask with my previous message.

On the page you mentioned, it is said that (under the "GAMMA WORKFLOW: ENCODING & CORRECTION" title, there is a diagram) display gamma is the opposite of the color space gamma we include it when we choose the color space and then total system gamma is equal to one. So, when we look at our monitors, we're seeing a linear image as a result, according to that diagram.

As you know, human vision is not linearly correlated with the luminance values around us. Our vision has logaritmic curve. This means that, as you know again, we're more sensitive to changes in dark tones than lighter tones. So, we have some gamma encoded in our eyes.

The point I would like to ask was that where is the nonlinear human vision in this system mentioned in the diagram on that page?

Are we just placing that gamma thing (when going from raw to JPEG, for example) for correcting the reverse gamma included in our display systems? If so, what is the reply of digital imaging to the logaritmic (nonlinear) vision of the human eye?

Thanks a lot ...

[revi]

Don't forget that your eye is looking at the image, so the non-linear response of the human eye comes into play there, and not in the image encoding.

The image on screen should reproduce the luminance values in the original scene. So there should be a 1:1 correspondence (so a gamma=1, ideally) between luminance values on screen and in the original subject. That is what is shown in image 3 in the tutorial that was referenced.
That means that there is no need to encode the gamma of human vision in the image.

The reason to use gamma encoding anyway is that it stores the data more efficiently (see the three gray scales in the tutorial): basically, the camera can store a lot more bright tones than we can distinguish, and less dark tones. Gamma encoding allows to correct that, and use less bits to store all the required information. The display then 'decodes' that to give us back the original tones. Note that a gamma-encoded image with 8 bits/channel corresponds to a linear encoded image with 11 bits/channel. Camera RAW files use typically 12 bits/channel... (11 bits/channel might be hard on the decoding software on your PC ).

Again, the exact response curve of the human eye isn't important in all this, that's why there have been several display gammas in use (old macs seemed to use gamma=1.8, PCs 2.2) which both showed perfectly natural images.

(Also note that the recommendation/habit to edit in 16 bits/channel whenever possible is not to get more levels in your final image, but to limit the risk of image degradation while editing: strong curves or levels corrections applied to an 8-bits image can lead to posterisation, where a 16-bit image wouldn't show problems)

[AltayHan]

That is ... you're saying that we're looking at linear images (gamma=1) on our monitors?

[AltayHan]

Remco, your explanation is reasonable ... and maybe true.

But please look at that document by Bruce Fraser http://wwwimages.adobe.com/www.adobe...near_gamma.pdf
It shows us two histograms ... one of them belongs to a linear image (gamma=1) and one of them belongs to a gamma encoded image. I think that I have never seen an image on my computer with a histogram like the first one (on the first page). I'm always seeing histograms like the second one on my computer. So, I think that my system gamma is not equal to 1.

[dje]

Hi Altay

I look at it in simple terms as follows

Gamma encoding basically counteracts the inverse gamma characteristic of the display device. This is what an overall system gamma of 1 means. The eyes don't come in to it. They are not part of the system in this terminology.

The image as seen on the display should match the original image so that our eyes and brain interpret them both in the same way.

Dave

[AltayHan]

Hmm, it is ok now )
Thank you for your explanation, Remco.
Please forget about my previous post ...

I would like to ask a second question now ...

As we know, photographic films have non-linear response curves, so their curves is more similar to our eyes ... that is, they're not recording the original luminance values in the scene. In this case, can we think that all analog photographs are always showing us a little bit overexposed version of the scene if we compare it with a digital image of the same scene at the same exposure level?

[AltayHan]

Thanks Dave ... it is clear for me now.

[AltayHan]

Hmm, it is ok now )
Thank you for your explanation, Remco.
Please forget about my previous post ...

I would like to ask a second question now ...

As we know, photographic films have non-linear response curves, so their curves is more similar to our eyes ... that is, they're not recording the original luminance values in the scene. In this case, can we think that all analog photographs are always showing us a little bit overexposed version of the scene if we compare it with a digital image of the same scene at the same exposure level?

Maybe, photographic papers has reverse gamma like our monitors?

[revi]

Well, as photographic papers use basically the same type of emulsions as film, they would have a similar non-linear response curve. But the source (the negative) has the luminance values inverted (more or less)... In practice it works out correctly (as Ansel Adams managed to show already )

We still get back to the same thing:
the image we look at must reproduce the original scene (within the physical possibilities of the medium).

Wrt the document by Bruce Fraser: if you look at the image associated with the gamma=1 histogram, you'll notice it's very dark. You normally don't see images like that, as the systems/programs we use immediately gamma-encode the image. There are some programs that can produce images with a gamma=1 from a standard RAW file (dcraw for instance), and those image look indeed very dark.

[AltayHan]

Visually, we see gamma=1 images on our computers, but the histograms belong to gamma (2.2) encoded images.

I mean ... when a JPEG comes to our monitors (gamma=2.2), its pixels have already gamma included values, so the program which shows us the histogram of the image shows these gamma=2.2 included pixel values in the graph. But, visually, because of the display gamma, we're looking at the gamma=1 image and then our eyes include a 2.2 (or something like this) gamma ... as a result, the histogram graph on our monitors belongs to the image in our brains )) very nice

Thanks a lot Remco,
Thanks a lot Dave.