Help me! Color management...

[george013]

Simon,
I missed your post. On another page.

When I talk about colors in this way, I mean the primary colors R,G and B. Those can have a maximum value of 255 in a 8-bit colordepth. And a screen should emit light with a wavelength for those primary colors within its colorspace.

As you can see from the diagram of colour spaces above, Adobe RGB can represent a greater range of colour than can sRGB. If you use 8-bit encoding, then each colour space uses 3 values in the range 0 to 255 to represent colours. Using 3 8-bit numbers to represent colour gives around 17 million possible combinations. In the case of Adobe RGB, those 17 million combinations represent a larger range of colour.

But when you use a monitor with colormanagment, your Adobe picture will be translated to sRGB. And since PP is mostly based on visual judgement, I can't see any difference.


From post 49.

I think it can be beneficial to shoot raw and post-process in a wide color space but use a standard-gamut monitor (which most people shooting in raw probably do). IMHO the benefits of processing in a wide color space are:

to avoid distortion caused by clipping from processing in a narrow color space
future proofing - one day I might want to use a wider gamut monitor, or print on a printer capable of wider gamut

The only way to avoid your first problem seems to me a larger bit depth. That's why never shoot in JPG. Nikon NEF is 12 or 14 bits, creating a raster file with the same bit depth in Capture. Than you're working with smaller steps. I still wonder, if I divide a larger range, let's say Adobe in 255 steps, than those steps will be bigger as when I divide a range of sRGB in 255.

You're right. I'm learning.

George

[xpatUSA]

I don't know what you exactly mean. When I speak of colors in my post I mean the R,G and B channels/colors.
And to me a range is something between 2 limits..
It's the sRGB color space and not gamut.

About your picture. If you mean a 4-bit color depth, then there're 4096 colors, 16 per channel, together 4096.

A picture doesn't have a gamut. That belongs to the screen. If you mean the colorspace, I wouldn't know.

George

Sounds like I got everything wrong, George, the way you tell it

I did make a typo under the image I posted. "4-color depth" and I meant it to read "4-bit color depth" which, as you point out, is not quite right . But what I really meant to say was 16 possible colors per screen dot or triad or whatever, including black and white.

I'm off to delete my incorrect and unhelpful post now . . .

[Simon Garrett]

Simon,
I missed your post. On another page.

When I talk about colors in this way, I mean the primary colors R,G and B. Those can have a maximum value of 255 in a 8-bit colordepth. And a screen should emit light with a wavelength for those primary colors within its colorspace.

Not exactly. Very vew monitors (if any) emit primary colours consisting of monochromatic light of a single wavelength. A typical monitor of near-sRGB colour space emits three primaries each of which consists of a narrow band of wavelengths in combination.

But when you use a monitor with colormanagment, your Adobe picture will be translated to sRGB.

No, that isn't what happens! The colour space of the image (Adobe RGB, sRGB, ProPhoto RGB or whatever) is translated into the monitor's colour space, which almost certainly isn't sRGB.

And since PP is mostly based on visual judgement, I can't see any difference.


From post 49.


The only way to avoid your first problem seems to me a larger bit depth.

Not really, although one may use a larger bit depth too.

The first problem I think you refer to is "...distortion caused by clipping from processing in a narrow colour space", which as it says is a problem of narrow colour space, not bit depth. If one carries out post-processing in sRGB, there is a chance that numerical computations may overflow the range of colours of sRGB, resulting in clipped colours - even if though the image may consist entirely of colours within sRGB. The simplest example is if you increase the saturation. If there are colours near the edge of sRGB, then increasing the saturation may result in colours that would be outside sRGB, but by processing in sRGB get clipped. If you increase saturation, then subsequently decide you've increased it too much and reduce it, you may end up with distorted colours if you edit in sRGB.

The problem is unlikely to occur except with major post processing edits.

That's a good reason for using what Adobe calls "Vibrance" controls rather than saturation. Vibrance decreases the amount of change to highly saturated colours.

That's why never shoot in JPG. Nikon NEF is 12 or 14 bits, creating a raster file with the same bit depth in Capture. Than you're working with smaller steps. I still wonder, if I divide a larger range, let's say Adobe in 255 steps, than those steps will be bigger as when I divide a range of sRGB in 255.

George

If you convert from Adobe RGB in 8-bit to sRGB in 8-bits, the steps (the change in colour resulting from a change of 1 unit) will be smaller.

An analogy: suppose you measure distance in inches on range of 0 to 9. You can measure up to 9 inches, with a step size of 1 inch. If you convert a measurement to centimeters, also with a range 0 to 9 then any values greater than 9 cm (about 3.5 inches) get truncated to 9 cm. But within the range 0-9 cm you now have a step size of 1 cm (about 0.4 inches). Lower range, greater resolution.

Similarly, if you convert from Adobe RGB to sRGB, both in 8-bits, you lose range but get greater resolution. From the link here (see the diagram top right - click it to enlarge) you can see that the red and blue primaries of sRGB are the same as those of Adobe RGB, but the green primary is different, and the colour space is stretched in the green direction. So sRGB has similar range and resolution (compared to Adobe RGB) for reds, blues and magentas, but lower range and higher resolution in greens, and to a lesser extent for cyans and yellows.

[george013]

Simon,

Thanks. I've got something to think about.

George