Bit depth vs gamut

[gordo]

Could some one explain this in lay person's terms?

Assume we're editing images in PShop.

The ProPhoto profile has a larger gamut (more colors) than the Adobe 1998 RGB profile.

But you're editing the image at a bit depth of 8 bits per channel - i.e. 24 bit RGB.

How does gamut relate to bit depth?

thx, gordo

 

[mattbeals]

At the risk of sounding like a complete idiot I always thought of it this way; Gamut describes the range of colors and bit depth referred to the amount of information (but not resolution).

 

[gordo]

Does this analogy work?

Gamut is like a balloon - it has volume. You can blow more air (color) into the balloon and it gets larger (bigger gamut).
Bit depth is like drawing circles around the balloon's surface. For 8 bit that would be 256 circles. As you inflate the balloon it gets larger so the distance between the circles gets larger - but there's still only 256 circles.
Because the spaces between the circles is larger there is more chance of seeing the steps - i.e. shadestepping.

Something like this:

best, gordo

 

[mattbeals]

yeah, you're right. that actually makes sense.

 

[Lukas Engqvist]

Loveley analogy

I have used a picture to explain different profiles the analogy of tents. A family tent and a hiking tent. In the hiking tent i can't stand up…*but it's light and i can take it anywhere, this is like the sRGB. AdobeRGB is like a family tent, I can stand up and move freely, there is more space. (Guessing ProPhoto is a circus tent here ;P )
I use this picture to visualise that if you use the wrong profile then it will be like mixing the tent poles. Using an Adobe profile on an sRGB will make it "break at the seams", and using an sRGB on an AdobeRGB immage will make it sagg/drab. (I'm now thinking circus tent on a hiking tents poles)

Now this analogy does not explain what gordo does so nicely with the balloon…*and I probably will borow that analogy in the future.

 

[TerryWyse]

Good analogy Gordo.....although I've never liked the smell of balloons.....and don't even get me started on clowns.

"bit depth" has nothing whatsoever to do with either gamut or dynamic range....it only has to do with the number of steps used to encode a given color gamut (or image really as most RGB color spaces/profiles are already 16bits...it's the image bit depth we're talking about here)

Your analogy shows why bit depth DOES matter though when it comes to gamut volume and image editing. In the small balloon (sRGB?), 8 bits-per-channel (bpc) might be perfectly adequate to use as an editing space since the steps are relatively close together and are less likely to result in banding/posterization as these steps are "pulled apart" as a result of a curve edit or other adjustment.

However with a space/gamut volume as large as ProPhotoRGB, 8bpc would reveal banding pretty quickly since the steps are farther apart than something like sRGB or even AdobeRGB.....the reason many professionals in the photography world (where ProPhoto is more prevelant) would recommend working with 16bpc images when using ProPhotoRGB as the color space "container" or "balloon".

So...the larger the gamut volume/color space, the more bit depth you'd need to avoid visible stepping/banding in an image.


Music recording/playback is a good analogy as well.....the higher the sampling rate (bit depth) the smoother the recorded sound is upon playback.....as you start dropping the sampling rate, the music will start sounding "grainy" or "noisy" (banding/stepping in images) because of the "gaps" left in the music from the too-low sampling rate. As in music as it is in images, the degree to which this becomes noticeable is largely a function of the source material. Rock-and-roll music or other forms that may apply a degree of distortion would likely tolerate lower bit depths/sampling rates than something with pure tones and a wider dynamic range such as classical music.

Regards,
Terry

 

[TerryWyse]

Does this analogy work?

Gamut is like a balloon - it has volume. You can blow more air (color) into the balloon and it gets larger (bigger gamut).

Hey gordo, how about a squeezebox/accordion as an analogy?

Your accordion has 256 pleats in it (8 bits-per-channel)......when it's closed/small, you barely notice the pleats in the bellows.....but open it up to get a BIG sound (larger gamut), all of a sudden one can see all the pleats in the bellows. If you had 16,384 pleats in the bellows (16 bits-per-channel), you probably wouldn't notice them as much and you could concentrate on the monkey holding the tin cup.



Terry

 

[meddington]

Your accordion has 256 pleats in it (8 bits-per-channel)......when it's closed/small, you barely notice the pleats in the bellows.....but open it up to get a BIG sound (larger gamut), all of a sudden one can see all the pleats in the bellows. If you had 16,384 pleats in the bellows (16 bits-per-channel), you probably wouldn't notice them as much and you could concentrate on the monkey holding the tin cup.

That's just sheer genius Terry, though I doubt very much Dan Margulis would see the value in a 16-bit accordion. You'd be making sounds that nobody could hear.

 

[gordo]

That's just sheer genius Terry, though I doubt very much Dan Margulis would see the value in a 16-bit accordion. You'd be making sounds that nobody could hear.

If it was a polka - that might be a good thing LOL!

Thanks all for elaborating on this topic.

best, gordo

 

[TerryWyse]

That's just sheer genius Terry, though I doubt very much Dan Margulis would see the value in a 16-bit accordion. You'd be making sounds that nobody could hear.

Perfect Mike....you haven't lost YOUR genius. Just remember, as a "genius" you're only one "false profile" away from madness.

;-)

tw

 

[Lukas Engqvist]

Perfect Mike....you haven't lost YOUR genius. Just remember, as a "genius" you're only one "false profile" away from madness.

Is that an "assign" or a "convert" relationship between "genius" and "mad" and do I use relative or psycometric conversion?

 

[TerryWyse]

Is that an "assign" or a "convert" relationship between "genius" and "mad" and do I use relative or psycometric conversion?

Well, if you're talking about Mike Eddington, he will not understand the difference between "assign" and "convert" and mistakingly assign a profile, resulting in a psycometric conversion from genius to madness...and his head will explode ("ultra wide gamut cranium conversion").



tw

 

[meddington]

Well, if you're talking about Mike Eddington, he will not understand the difference between "assign" and "convert" and mistakingly assign a profile, resulting in a psycometric conversion from genius to madness...and his head will explode ("ultra wide gamut cranium conversion").

Its happened before. There was color everywhere...like Walt Disney threw up.

 

[David Kunkel]

Another wacky analogy, based on music: Low bit depth is like AutoTune used in a song, and low gamut is like cheap earbuds. With cheap earbuds, you might not notice that the singer used AutoTune. High gamut is like 5.1 SurroundSound, and high bit depth is like Pavarotti. You can hear every nuance of Pavarotti's genius on the 5.1 SurroundSound system.

 

[David Kunkel]

Dang, Terry covered most of what i was thinking while I was typing.

 

[MrRWright]

It's like a box of crayons

It's like a box of crayons

Another way to think about this topic is a box of crayons. The higher the bit depth, the more crayons in your box.

Rich

 

[David Kunkel]

MrRWright, that's a better description of gamut (range of colors available) than bit depth (ability to draw more shades of a color)

 

[michaelejahn]

oh, that old chestnut - the 16 bit challenge

oh, that old chestnut - the 16 bit challenge

That's just sheer genius Terry, though I doubt very much Dan Margulis would see the value in a 16-bit accordion. You'd be making sounds that nobody could hear.

For anyone wondering what Dan Margulis and 16 bit might have to do with this discussion, here is a link to what this was all about...

Welcome to Bruce Lindbloom's Web Site

@ Gordo - i think this page might help you better understand a few things - to me' in lamans terms means "pictures" - and that is what you will see if you follow this link and scroll down to "Gamut Projections"

Welcome to Bruce Lindbloom's Web Site

Hope this help.

 

[Lukas Engqvist]

Welcome to Bruce Lindbloom's Web Site

The gamma had so much effect on number of unique colours very interesting.

Would have liked to see with Adobe or ECI RGB and compare 1.7 and 2.2 gamma…*just for reference

 

[Chuck Koehler]

The Volume of a Bit and the Depth of a Gamut

The Volume of a Bit and the Depth of a Gamut

Interesting Subject and I can never resist trying to visualize these concepts so that I better understand them. There are a few concepts explained in the post that are very good, so I thought I would try to add my two sense worth.

I have been intrigued with Gamut Volume for years, and although I think I get it, I don't get it to the point to explain it as well as I want to so here goes:

The original request:

Could some one explain this in lay person's terms?

Assume we're editing images in PShop.

The ProPhoto profile has a larger gamut (more colors) than the Adobe 1998 RGB profile.

But you're editing the image at a bit depth of 8 bits per channel - i.e. 24 bit RGB.

How does gamut relate to bit depth?

thx, gordo

Gamut is like a container (the accordion fully expanded )

Gamut Volume is like how many particles fill the container

Bit would be the description "particle"

Bit Depth would be the amount of different possibilities for describing the characteristics of the individual particles

Now comes the interesting part:

Assume that we have two Identically Sized (gamuts) accordions that can hold 1 liter total volume when fully expanded.

One accordion has 100 pleats (lower bit depth) and one has 1000 pleats. (higher bit depth)

And let's also assume that the particles that we put into the containers are different sizes, (different Gamut Volumes) let's say one is filled with mililiter sized particles and one is filled with centiliter sized particles. OK, so one gamut holds 1000 particles and the other holds 100 particles. Easy enough.

So now each container could hold an equal amount of particles, and also has the ability to describe each particle that is contained in it. Even the lower Bit Depth has the ability to describe all of the possible colors in the gamut with the higher volume, or even the colors of a larger gamut for that matter.

So, in the end - Gamut and Bit Depth do not (practically) affect one another at all. The only time there would be an issue would be if you manipulate an image in a smaller color space and then convert it to a larger color space, but in my opinion that is 100% to do with the gamut and has little to do with the bit depth.

The importance of Bit depth is immediately and drastically recognized when manipulating Images. Take two identical images (the Raw Capture and the Raw Capture converted to Jpeg) and then try and fix a bad exposure, or perform any sort of significant modification. The JPEG has maybe 1/10th of the latitude of the Raw Image.

Well, that was my feeble attempt at a description.

CK