sRGB vs. RGB Locked

Started Apr 22, 2013 | Discussions thread
This thread is locked.
John King
John King Forum Pro • Posts: 14,941
Re: Why use sRGB for JPEGs...

Gidday Victor

Victor Engel wrote:

John King wrote:

Gidday Victor

Victor Engel wrote:

Bit depth has little to do with color space.

To a certain degree you are right.

Notice I didn't say it had nothing to do with it....

However, it does rather depend on how coarsely or finely you wish to display colour nuances. I have a test image that I have used extensively for testing what I have said here when printing. My infamous red headlight:

Looks silver to me (with blue reflections of the sky, apparently). 


This prints very differently in 8 or 16 bit sRGB, but far less difference when using 16 bit aRGB or PPRGB.

You must be doing something wrong, then. Adobe RGB and sRGB are very similar at the red corner of the color space.

True about aRGB and sRGB, but when one looks at the colour numbers, the reference point in PPRGB has G=36 (in 8 bit colour numbers). In sRGB, this value drops to G=2. So neither the displayed image, nor the print has enough green "ink" to properly colour balance the image. Hence the colour shift particularly noticeable in both the red and blacks, but also throughout the whole image.

There is a particular speck of sap above the bonnet join on the right of the image. Measuring just below this speck shows that the green channel value in 16 bit sRGB drops from 62 (in 8 bit colour numbers) in PPRGB to 2 in sRGB. This difference (not reflected in the R/B channels) changes the image from bright red to a sickly pinkish orange. The blacks also tend to grey, and the whites are wrong also.

If the whites and blacks are significantly different, you have not set things up properly. White and black should be identical in most color spaces.

I PP all of these images from the same RAW file, using the same settings. The only thing that I have changed is the bit depth and colour space (gamut). Printed from 16 bit PPRGB, this image is all but a perfect match for the original car paintwork.

I see the same effect in lots of other images, even when not a really saturated colour such as this. Many natural and man made colours are out of gamut for aRGB, and way out of gamut for sRGB.

I have a particular rose whose flowers have the most saturated red I think I've ever seen, except, perhaps from LEDs. I don't think it's out of gamut in Prophoto RGB, but it is out of gamut in sRGB and Adobe RGB. It's one of those colors that almost hurts to look at.

A color space is usually a three dimensional volume. The bit depth comes into play when approximating the continuous space. The higher the bit depth, the closer neighboring colors are, but 6 bit, 8 bit, and 16 bit sRGB models all cover the same volume.

Yes. But that sRGB volume is both deficient and defective, as I said before.

How is it defective?

As before: the R, G and B colour axes are not equal length. Particulalrly on a green planet, being unable to reproduce greens accurately is a somewhat major failing, IMNSHO. Hence defective.

Higher bit depth gives one more editing latitude before one colour number becomes the same as another, but different colour number.

Huh? If it's the same, how can it be different?

Not what I was saying. Sorry. Hard to explain. Say a point in one colour channel has a number of 16,384 in a 16 bit, wide gamut colour space. This number may be able to move up to 16,389 before it combines with another colour which has a colour number of 16,390. The 'distance' that the number can travel before this occurs is greater in a 16 bit wide gamut colour space than it is in an 8 bit space such a 8 bit aRGB.

Using a 16 bit colour space allows the reproduction of more finely nuanced prints and other images, without any form of posterisation.

Sure, but that doesn't have anything to do with gamut, except that if you use a wider gamut, consecutive numbers will be farther apart. You are more likely to see posterization in an 8-bit ProPhoto RGB image, for example, than an 8-bit sRGB image.

Neither I, nor anyone else I have ever read, suggests using a gamut as wide as PPRGB at 8 bit depth. For exactly the reason you mention - posterisation.

PPRGB is necessarily a creature of high bit depth colour spaces.

I have written a short (6 page) paper on this and related subjects. It explores many of these issues at an extremely shallow level. Its working title is "Basic colour management for photographers". Done to save others from the immense amount of reading and researching I have done on this WRT capture, editing and printing/displaying, and the necessity for me to explain it over and over again to those who ask me personally about it.

Personally, I am extremely interested in these and related subjects. Most are not, and just want simple rules they can follow that will give them acceptably good, accurate, reproducible results. That is the target audience I wrote the paper for. However, I'm not here to "educate" the world, or convert anyone to my way of thinking or doing things. I have found out the very hard way what works, and what doesn't, and what cannot; so I am just sharing that with people here, who are free to dismiss it as rubbish, or agree wholly or partially. It works exceptionally well for me. It is merely offered FWIW. I have no desire to fight about it with anyone.

-- hide signature --

Regards, john from Melbourne, Australia.

 John King's gear list:John King's gear list
Olympus E-1 Olympus E-510 Olympus E-30 Olympus E-M1 Olympus Zuiko Digital 14-54mm 1:2.8-3.5 II +17 more
Post (hide subjects) Posted by
(unknown member)
(unknown member)
MOD Footski
MOD Footski
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow