Destruction by jpeg

[george013]

I've sent you a PM with links to the websites.

Would they shoot raw if they had more time? I spent yesterday evening with one of the photographers (Ray) and we got onto the topic and the answer was NO. He does not feel that shooting raw would result in a better product for his clients, but he added that if they insisted on raw (for some of the non-wedding / portraiture work / event work), he would have no problem doing so, but his price would go up due to the additional work involved.

Now just think about it; if your skill level were to the point that you shoot jpeg and spent 30 seconds per image (this includes culling down to 50-60 keepers), you'd be done in 2 hours, rather than 15...

What I read here, is that he thinks the JPG out of the camera are good eneough for this kind of photography, but if you want some more, more critical, than the RAW is coming up.

George

[Saorsa]

What I read here, is that he thinks the JPG out of the camera are good eneough for this kind of photography, but if you want some more, more critical, than the RAW is coming up.

George

There is nothing wrong with that at all. A wedding photographer, for example, is going to need to present some large number of images for selection for an album and JPEGs would be perfectly acceptable as proofs and album sized prints. I would guess that when the bride wants the 30x40inch beautifully framed image to hang over the fireplace that some processing will be needed and the raw would be nice to have or at least a workflow that keeps or can return to the original.

[Manfred M]

What I read here, is that he thinks the JPG out of the camera are good eneough for this kind of photography, but if you want some more, more critical, than the RAW is coming up.

George

Having seen their large format prints from a jpeg, I would have to disagree with your assessment. Prints use a CMYK+ colour space, which is even smaller than sRGB.

[xpatUSA]

Doesn't SWOP CMYK have a wider gamut in some places than sRGB but less in others?

[Manfred M]

Doesn't SWOP CMYK have a wider gamut in some places than sRGB but less in others?

Straight CMYK is definitely the smallest of the standard colour spaces. Unfortunately the printer manufacturers do not generally publish what the additional inks contribute over and above the basic CMYK. Numbers I've seen suggest 350,000 to 750,000 distinct colours is a reasonable range from pure CMYK to the 12 ink printers. This is a lot less than the 16million for basic sRGB.

[xpatUSA]

Straight CMYK is definitely the smallest of the standard colour spaces. Unfortunately the printer manufacturers do not generally publish what the additional inks contribute over and above the basic CMYK. Numbers I've seen suggest 350,000 to 750,000 distinct colours is a reasonable range from pure CMYK to the 12 ink printers. This is a lot less than the 16million for basic sRGB.

I had replied to this before but deleted my post as being too technical.

Still it is interesting that the "size" of a color gamut could be expressed in terms of the number of colors. Could you elaborate on that?

I see that sRGB is said to have 16million colors. That looks like the total number of colors available for any RGB file irrespective of it's color space - so I'm clearly misunderstanding something in that regard.

I've been accustomed to thinking of the "size" of a color space in terms of it's area on a CIE chromaticity diagram and, from that point of view, there are some printer spaces that are much bigger than sRGB.

[george013]

I had replied to this before but deleted my post as being too technical.

Still it is interesting that the "size" of a color gamut could be expressed in terms of the number of colors. Could you elaborate on that?

I see that sRGB is said to have 16million colors. That looks like the total number of colors available for any RGB file irrespective of it's color space - so I'm clearly misunderstanding something in that regard.

I've been accustomed to thinking of the "size" of a color space in terms of it's area on a CIE chromaticity diagram and, from that point of view, there are some printer spaces that are much bigger than sRGB.

The 16 million colors is the result of a 8-bit colordepth: 256x256x256. Forget about the amount of colors, that's only a result of a division, independant of a color, colorspace, colormodel, gamut or anything else.

George

[xpatUSA]

I see that sRGB is said to have 16million colors. That looks like the total number of colors available for any RGB file irrespective of it's color space - so I'm clearly misunderstanding something in that regard.

I've been accustomed to thinking of the "size" of a color space in terms of it's area on a CIE chromaticity diagram and, from that point of view, there are some printer spaces that are much bigger than sRGB.

The 16 million colors is the result of a 8-bit color depth: 256x256x256. Forget about the amount of colors, that's only a result of a division, independent of a color, color space, color model, gamut or anything else.

George

Which leaves open the question of how is it decided that one color space is "bigger" than another? If we don't count the # of colors, what do we do? I have a feeling that gamut diagrams provide the answer, especially 3D ones . . .

. . . quite a strong feeling, actually

[Manfred M]

Which leaves open the question of how is it decided that one color space is "bigger" than another? If we don't count the # of colors, what do we do? I have a feeling that gamut diagrams provide the answer, especially 3D ones . . .

. . . quite a strong feeling, actually

Yes and no. The diagrams are only part of the story and if you only look at them, it is easy to suggest that they are more or less the same size as sRGB and in fact are capable of reproducing a wider colour range. As I recall, that was the rationale behind the AdobeRGB colour space.

Now the problem of course is the ability of the devices to reproduce these discrete colours. This is dependent on the "efficiency" of blending the four different inks (hint -theory says you don't need black (CMY), but in order to actually get a decent print you do). Regardless of the actual printing process; offset press, dye sublimation, colour silver halide photo paper, laser jet, ink jet don't come anywhere close to the millions of tone of even sRGB. This is why the printing industry uses "spot colours" when a colour has to be bang on, like in a company logo.

Looking just at an inkjet pure CMYK printer that most people are familiar with. Four dyes are used to create a shade using discrete dots of ink. Nozzle size and dot sizes determine how colours are deposited. The lower end printers have a single dot size. The higher end ones have at least two dot sizes per nozzle which results in far fewer possible combinations than a 24 bit display or a 30 bit display. Printer manufacturers do not publish this type of information. Sources I have seen (I wish I could find these references now) suggest that real life printers reproduce around just over 300,000 distinct shades / colours while top end ones are probably pushing somewhere in the 700,000 or even 800,000 distinct colours.

I know you love splitting hairs. I guess I can do so too, but one sometimes needs to look at the practical side of things, not just the theoretical.

[xpatUSA]

Yes and no. The diagrams are only part of the story and if you only look at them, it is easy to suggest that they are more or less the same size as sRGB and in fact are capable of reproducing a wider colour range. As I recall, that was the rationale behind the AdobeRGB colour space.

Now the problem of course is the ability of the devices to reproduce these discrete colours. <> Printer manufacturers do not publish this type of information. Sources I have seen (I wish I could find these references now) suggest that real life printers reproduce around just over 300,000 distinct shades / colours while top end ones are probably pushing somewhere in the 700,000 or even 800,000 distinct colours.

I know you love splitting hairs. I guess I can do so too, but one sometimes needs to look at the practical side of things, not just the theoretical.

Manfred, after a bit of research, I found quite a few gamut diagrams for printer profiles, in various color spaces even. I'll put just one of them that shows a printer profile or two:



From what you're telling me, I can not claim that Epson 2200 Premium Luster has a larger or wider color space (well, gamut really) than sRGB - because the way the inks get splattered on the paper reduces the number of distinct shades or colors available for printing. Have I understood your point of view correctly?

And that view is quite distinct from a monitor output where it is the chromaticity of each of the RGB primaries that determines the "size" of the color space. Indeed, if I were to somehow set the monitor to display only 16 shades/colors (remember the Commodore 64?), would the size of the color space be changed or not? Answer: It would not; therefore, with regard to monitors, TVs, tablets, cell phones, etc., there is no relationship between the number of shades/colors and the gamut thereof.

[xpatUSA]

Looking just at an inkjet pure CMYK printer that most people are familiar with. Four dyes are used to create a shade using discrete dots of ink. Nozzle size and dot sizes determine how colours are deposited. The lower end printers have a single dot size. The higher end ones have at least two dot sizes per nozzle which results in far fewer possible combinations than a 24 bit display or a 30 bit display. Printer manufacturers do not publish this type of information. Sources I have seen (I wish I could find these references now) suggest that real life printers reproduce around just over 300,000 distinct shades / colours while top end ones are probably pushing somewhere in the 700,000 or even 800,000 distinct colours.

Found one for you from the Epson site - a 3880 profile:



Mainly of interest is a term "Gamut Volume" said to be 994,069 but with no units. I see also some Lab color space values quoted and I also see mentions of black ink so I'm not even sure what this profile is for, you would know better than I by the sound of it. Does say color too, though.

Anyway I downloaded it just as an example and opened it in ColorThink and it is way "smaller" (in my terms of 2D area or 3D volume) than sRGB, even with a "Gamut Volume" of almost one million.

[george013]

Which leaves open the question of how is it decided that one color space is "bigger" than another? If we don't count the # of colors, what do we do? I have a feeling that gamut diagrams provide the answer, especially 3D ones . . .

. . . quite a strong feeling, actually

I said before I don't know much about colors and that a digital photo doesn't have colors, just as it doesn't have a metric size. Colors and metric size are provided through the output device. So that's where you should focus on I think.

I think the diagrams, the horseshoe, is based on wavelength values. https://en.wikipedia.org/wiki/Spectral_color

It still leaves me with a lot of quetsions.

George

[Manfred M]

Interesting data, Ted. I'm going to have to wait until I get home and have a proper keyboard to type on before I try to reply. Typing on a tablet in a B&B is a PITA. With the diagram and chart, you are onto what I have been trying to say. The boundary conditions on the colour space diagrams are part of the answer as is the granularity of the data within the lines and of course a proper white and black point.

Part of it is theoretical / mathematical modelling and part of it is device dependence.

[Saorsa]

The camera is essentially an A/D converter. It provides for capture and storage. Once captured the stored data will be subject to many transformations before it is again made available for viewing. The ultimate D/A converter is our eyes and brain.

In creating a viewable image from a digital capture there are many, many factors which will impact what is presented. I may use a calibrated monitor to edit and create a common file format but if it is presented on an uncalibrated monitor, it will look different and different again sent to a television. Try printing on an ordinary bond paper, then a high gloss photo paper and then a matte presentation paper without changing the paper profile of your printer and you will see three entirely different presentations of the same image.

Ultimately, a raw file has greater precision than a compressed format like jpeg. That's just a mathematical fact. Since our means of presentation and/our ability to perceive what is presented have their own limitations then an image can be accurately presented regardless of precision.

The manufactures of all our processing devices, cameras, monitors, projectors, printers deal with the processing for their devices. Our contribution is our processing we choose, our workflow and the care we give the images.

[xpatUSA]

I said before I don't know much about colors and that a digital photo doesn't have colors, just as it doesn't have a metric size. Colors and metric size are provided through the output device. So that's where you should focus on I think.

I think the diagrams, the horseshoe, is based on wavelength values. https://en.wikipedia.org/wiki/Spectral_color

It still leaves me with a lot of quetsions.

George

You might find this article interesting:

http://www.tomshardware.com/reviews/...er,2919-3.html
.

[george013]

You might find this article interesting:

http://www.tomshardware.com/reviews/...er,2919-3.html
.

That's what I mean. You adjust the image for the output device. Here it's a printer with different papers. There are different ways to deal with the colors that are out of the printers range.

That's why I don't understand why people use Adobe colors even when their monitor can't show them.

George