Color management

[Macro guy]

Ok, I've taken one of my images (this one) because it's got the blues, the yellows and the greens and ran it through the paces.



I ran it through the RAW converter, setting the color space to wide gamut and I converted it to a 16bit tiff. I set the working color space in my image processor (I use Corel paintshop pro) to wide gamut and I opened the tiff in the app. It looked somewhat different even on my sRGB monitor. The yellows looked brighter and there was more green and yellow highlights. I'm going to take that file to the lab tomorrow and have it printed. I have a problem converting this file to sRGB to save it as a jpeg. I've tried exporting the file, setting the colorspace back to sRGB and opening the file again to no avail. Does anyone know how to do this?

I probably don’t understand, but you want to convert to sRGB and save it as a jpeg?? Stupid question, but why?

For web viewing

 

[mujana]

Ok, I've taken one of my images (this one) because it's got the blues, the yellows and the greens and ran it through the paces.



I ran it through the RAW converter, setting the color space to wide gamut and I converted it to a 16bit tiff. I set the working color space in my image processor (I use Corel paintshop pro) to wide gamut and I opened the tiff in the app. It looked somewhat different even on my sRGB monitor. The yellows looked brighter and there was more green and yellow highlights. I'm going to take that file to the lab tomorrow and have it printed. I have a problem converting this file to sRGB to save it as a jpeg. I've tried exporting the file, setting the colorspace back to sRGB and opening the file again to no avail. Does anyone know how to do this?

I probably don’t understand, but you want to convert to sRGB and save it as a jpeg?? Stupid question, but why?

For web viewing

Indeed stupid question from my side. Sorry. I thought you wanted to take that (converted to sRGB file) file for printing.

 

[Macro guy]

Ok, I've taken one of my images (this one) because it's got the blues, the yellows and the greens and ran it through the paces.



I ran it through the RAW converter, setting the color space to wide gamut and I converted it to a 16bit tiff. I set the working color space in my image processor (I use Corel paintshop pro) to wide gamut and I opened the tiff in the app. It looked somewhat different even on my sRGB monitor. The yellows looked brighter and there was more green and yellow highlights. I'm going to take that file to the lab tomorrow and have it printed. I have a problem converting this file to sRGB to save it as a jpeg. I've tried exporting the file, setting the colorspace back to sRGB and opening the file again to no avail. Does anyone know how to do this?

I probably don’t understand, but you want to convert to sRGB and save it as a jpeg?? Stupid question, but why?

For web viewing

Indeed stupid question from my side. Sorry. I thought you wanted to take that (converted to sRGB file) file for printing.

I still can't figure out how to do it

 

[Rich42]

Ok, I've taken one of my images (this one) because it's got the blues, the yellows and the greens and ran it through the paces.



I ran it through the RAW converter, setting the color space to wide gamut and I converted it to a 16bit tiff. I set the working color space in my image processor (I use Corel paintshop pro) to wide gamut and I opened the tiff in the app. It looked somewhat different even on my sRGB monitor. The yellows looked brighter and there was more green and yellow highlights. I'm going to take that file to the lab tomorrow and have it printed. I have a problem converting this file to sRGB to save it as a jpeg. I've tried exporting the file, setting the colorspace back to sRGB and opening the file again to no avail. Does anyone know how to do this?

I probably don’t understand, but you want to convert to sRGB and save it as a jpeg?? Stupid question, but why?

For web viewing

Indeed stupid question from my side. Sorry. I thought you wanted to take that (converted to sRGB file) file for printing.

I still can't figure out how to do it

For the purpose you indicated, don't.

It's counter-productive.

In a color-managed system the purpose of profiles is to inform the system how to deal with files so that they will properly display. Color management was developed for just this reason - to remove the need for individual file preparation depending on the use.

Web browsers have color management incorporated and "know how" to display your image as long as its profile conforms to standards.

--
Rich
"That's like, just your opinion, man." ;-)

 

[Jess B]

First and foremost, mujana's suggestion to look at Andrew Rodney's website is a very good one. He's about as much of a color management expert as I know of, and he has many free articles and videos, a bunch of which I found quite helpful years ago.

These are the things I think I know:

• sRGB was standardized around the largest color palette color films could produce

No, it was standardized around the palettes / gamuts of typical CRT color monitors as they existed many years ago.

• Adobe RGB and Profoto RGB can display the same number of colors 16.7m but both extend the gamut at the cost of precision

No, any RGB working space will be divided up into 16.8 million gradations if you use 8-bit precision for each of the RGB components (i.e., 2^(3x8) = 16,777,216). If you use 16-bit precision, then it's 2^(3x16) = 281 trillion. That does not mean that each gradation is a distinct color.

"Color" is one of those terms with different meanings in different contexts, but if we use it to mean distinct to 'normal' human perception, then the volume of sRGB's gamut is about 900,000 cubic color units (CCU) and the volume of Adobe RGB's gamut is about 1.3 million CCU, all of which are colors. ProPhoto RGB has a sort-of gamut volume of about 2.9 million CCU, but some of those are not colors as defined above.

But no, this does not mean that 8-bit precision is always plenty and 16-bit precision offers little or no advantage!

• Higher end monitors and printers are able to reproduce the Adobe RGB gamut.

Although many high-end monitors can produce all or nearly all of the Adobe RGB gamut, no printer (at least of the photo printers we usually discuss) can produce anything close to all of sRGB, much less Adobe RGB. Conversely, many quite modest inkjet printers on quite ordinary photo papers can produce colors that are outside of Adobe RGB's gamut (typically mostly in yellows, but often enough somewhat in other colors).

• There is no hardware that I'm aware of that can reproduce Profoto RGB

If you mean 100% of ProPhoto RGB, then same here. But there are monitors and printers that can produce some colors that are outside of Adobe RGB but within ProPhoto RGB.

These are my questions:

• Would it be correct to assume that for reproduction work, sRGB should be used given that it's the most accurate color profile?

No, not at all. It's not more accurate. In fact, if the source image contains colors that are outside of sRGB, then it's less accurate; colors have to be clipped or compressed to fit within it. IMO the best workflow is raw file -> ProPhoto RGB -> some other working space or profile only for final output.

• I have seen people recommend editing in Profoto RGB. This makes no sense to me given that nothing can output the Profoto gamut and you're in effect editing blindly. Additionally, the output would still have to be converted to another color space for printing

See above. Keep all the colors you can unless and until you have to discard some of them. Just use enough precision to avoid banding etc. Keeping 16 bits per RGB channel at least until final output is best.

• What's the point of Profoto RGB?

To be able to describe and allow processing of colors that you can see, your camera can capture, your monitor can display, and your printer can print, that are outside of not only sRGB but also Adobe RGB.

• Doesnit make sense to edit in Adobe RGB if you have an sRGB monitor, but the lab can print in Adobe RGB? Aren't you in effect editing blindly, how can you tell what your results are going to be like?

Again, labs can't 'print in' Adobe RGB. A color-managed workflow means converting from the color working space to the ICC printing profile for the applicable specific combination of printer + inks + driver settings + paper, using the rendering intent that is (subjectively) best under the circumstances. Go read up on Andrew Rodney's website. And if you want to read further on printing, then get Jeff Schewe's book The Digital Print (Amazon).

Thank you for taking a crack at a comprehensive summary. Alas, I still feel a bit lost in the color space discussion. But I keep reading these and hope to refine my own process.

A question for those more expert than I... Am I incorrect in believing that Lightroom's default color space is ProPhoto RGB? I know I have read that, but perhaps my source was incorrect. If it's not, how would one tell Lightroom to import photos in that space? (Please explain slowly! I shoot and edit digitally these days, but a big part of my brain still is back holding a completely manual film camera 40+ years ago and finds much of the digital world to be perplexing!) Thanks!

 

[JimKasson]

A question for those more expert than I... Am I incorrect in believing that Lightroom's default color space is ProPhoto RGB?

Sort of. The internal representation uses the PPRGB primaries, but not the PPRGB tone curve. The internal representation uses a linear tone curve (gamma = 1).

I know I have read that, but perhaps my source was incorrect. If it's not, how would one tell Lightroom to import photos in that space?

You don't get that choice.

(Please explain slowly! I shoot and edit digitally these days, but a big part of my brain still is back holding a completely manual film camera 40+ years ago and finds much of the digital world to be perplexing!) Thanks!

 

[NAwlins Contrarian]

A question for those more expert than I... Am I incorrect in believing that Lightroom's default color space is ProPhoto RGB? I know I have read that, but perhaps my source was incorrect. If it's not, how would one tell Lightroom to import photos in that space? (Please explain slowly! I shoot and edit digitally these days, but a big part of my brain still is back holding a completely manual film camera 40+ years ago and finds much of the digital world to be perplexing!) Thanks!

Any file you import into Lightroom remains itself. If you import a raw file, then it remains a raw file. If you import a JPEG in Adobe RGB, then it remains a JPEG in Adobe RGB. But if you process any file in the Develop module, then that processing is done in Melissa RGB, which (as Jim said) has the same primaries (i.e., what 'red', 'green', and 'blue' actually mean) as ProPhoto RGB, but has a gamma of 1.0 instead of (IIRC) 1.8 in ProPhoto RGB. That processing does not generate a new image file; it's stored as a set of instructions in the Lightroom catalog. If you export the file, then those instructions are rendered--again, the operation is performed in Melissa RGB--and then converted to whatever working space or profile you selected in the Export dialog box and then that's saved as a new TIFF or JPEG or whatever.

I've heard that other parts of Lightroom actually use other color working spaces. I don't know whether that's true, and if it's true, why it's done that way.

 

[JimKasson]

I've heard that other parts of Lightroom actually use other color working spaces. I don't know whether that's true, and if it's true, why it's done that way.

When I'm writing image processing software, the color space I use depends on what I'm trying to do. Lab or Luv is a good space for changing chromatic intensity. PPRGB is fast to transform into and out of. Linear tone curves are good for adding light values or for some types of compositing. I usually start out using double precision floating point, but if speed is a concern, I may do some calculations in single precision floating point or 16-bit unsigned integer.

 

[Rand 47]

I've heard that other parts of Lightroom actually use other color working spaces. I don't know whether that's true, and if it's true, why it's done that way.

When I'm writing image processing software, the color space I use depends on what I'm trying to do. Lab or Luv is a good space for changing chromatic intensity. PPRGB is fast to transform into and out of. Linear tone curves are good for adding light values or for some types of compositing. I usually start out using double precision floating point, but if speed is a concern, I may do some calculations in single precision floating point or 16-bit unsigned integer.

"Show off!" LOL

Rand

 

[JimKasson]

I've heard that other parts of Lightroom actually use other color working spaces. I don't know whether that's true, and if it's true, why it's done that way.

When I'm writing image processing software, the color space I use depends on what I'm trying to do. Lab or Luv is a good space for changing chromatic intensity. PPRGB is fast to transform into and out of. Linear tone curves are good for adding light values or for some types of compositing. I usually start out using double precision floating point, but if speed is a concern, I may do some calculations in single precision floating point or 16-bit unsigned integer.

"Show off!" LOL

Rand

I resemble that remark.

 

[Macro guy]

Ok, I've taken one of my images (this one) because it's got the blues, the yellows and the greens and ran it through the paces.

I ran it through the RAW converter, setting the color space to wide gamut and I converted it to a 16bit tiff. I set the working color space in my image processor (I use Corel paintshop pro) to wide gamut and I opened the tiff in the app. It looked somewhat different even on my sRGB monitor. The yellows looked brighter and there was more green and yellow highlights. I'm going to take that file to the lab tomorrow and have it printed. I have a problem converting this file to sRGB to save it as a jpeg. I've tried exporting the file, setting the colorspace back to sRGB and opening the file again to no avail. Does anyone know how to do this?

I have made a wide gamut jpg,and given that browsers are color management aware, I'm posting the sRGB and the Wide gamut RGB version of this shot for your viewing pleasure.






This is the sRGB version








This is the wide gamut version

Note that both images have been processed the same way.



Thanks to all of you for your help

 

[mplkn]

I posted what follows in another forum in which an extended discussion broke out about about color and fidelity, about reality and representation. The quoted section has been lightly edited.

The discussion taking place here is far more technical and frankly, far more likely to result in practical understanding.

That said, (post begins)

**********

In thinking about all of of this, I'm finding myself comparing it to something I have a pretty good understanding of: audio capture, recording, and reproduction.

There are measurable frequencies emanating from sound sources in the studio or concert hall. There are measurable characteristics to the microphones being used to capture the sounds and convert them to electrical signals. There are measurable characteristics to the circuitry in the recording system, resulting in measurable characteristics to the captured frequencies found in the recording.

Then, when we reproduce that recording, there are measurable characteristics to the reproduction circuitry, and measurable characteristics to the reproduction transducer (headphones, or speakers) that convert electrical signals back into audible sound.

If using speakers, we then run smack into the influence of measurable characteristics of the listening room, its acoustic hardness or softness, absorbtion or reflectivity, at different frequencies.

And finally the reproduced sound arrives at my seventy-year-old Mark I calibrated eardrums.

So, then we have what's called Equalization, to dial in compensation at any of these points to deal with accentuation or attenuation resulting from the cumulative characteristics of the waypoints along the chain.

The whole thing is fraught.

And, the clients have eardrums too, or at least, what's left of their eardrums.

So.

If you manage to achieve and maintain what's called "flat response" throughout the chain, and then play the recording back, perfectly flat, to the 19-year-olds who've just banged out their "song" (and whose ears are probably ringing), you'll be thinking, "How can they expect to hear *any* high frequencies at all, with those tattered, torn up eardrums of theirs??"

But then: somebody, I absolutely guarantee it, is going to say, "Needs more bass, dude!"

This is why producers wash down antacids with their espresso

********

 

[Jess B]

I don't print myself. Otherwise, I could try this by printing seeing the results or edit by hard proofing.

This process sounds to me like the film days where you really didn't know what the color of your images actually was if you were shooting negative film and unlessnyou had some well established point of reference within your image, it was difficult to argue with the lab. That's why I shot slide film, so I would at least have a point of reference and even then, there was a lot of back and forth until the print looked right.

I remember those days. I also always shot slide film. But back then I never printed (or let them printed).

Digital was supposed to solve that and to a great extent, it did. However, if we're to edit blindly, we're back to this back and forth with the lab until the print looks right.

Well, no experiences with labs in general (only with “pro” labs that accepted AdobeRGB). Printing myself is another story. AdobeRGB works fine. Just use a calibrated monitor that can be used for AdobeRGB (ofcourse preferably almost 100%). You then can see what you’re doing. Seems simple enough (?).
No, it’s not like you have work in complete blindness. Especially not when you get to know files/monitor/printer matching. That can need some time, but once I knew/tried again and again, now printing regularly is almost as easy as shooting an image.
If you don’t want to print yourself, then I would choose a proper/ well known lab that does accept the best quality files….and stick with that lab

Apart from beginner hassles, I certainly and very much recommend printing yourself!

It makes sense to get an Adobe RGB monitor and work with the Adobe RGB color space. I don't understand how one can edit in PPRGB without constantly doing hard proofing.

Printing by myself makes no financial sense to me. I don't print often enough to warrant the hassle and the expense. I use a custom lab and they're very good. However, I'm in Thailand, so there is a language barrier to a certain extent. Their English isn't good enough to understand the complexities and nuances and my Thai isn't good enough either. So, the simpler the process can be made, the better the results are going to be.

BTW, the prints I get look fantastic, but it seems that I'm missing out.

Nice country to be. In your case, except for language I guess.

Bottom line is very simple:

If you like your prints that much ("fantastic"), you're not missing out on anything. Keep working as you do, and enjoy your prints!

I would have if I hadn't started this thread LOL Ignorance is bliss!

This forum certainly is humbling. While I like to think I'm a lifelong learner, there are times when it's intimidating to read one of these threads and think, "Wait, what? What the heck is that about?"

Well, thanks for the ongoing education, folks!

 

[NAwlins Contrarian]

In thinking about all of of this, I'm finding myself comparing it to something I have a pretty good understanding of: audio capture, recording, and reproduction....

If you manage to achieve and maintain what's called "flat response" throughout the chain, and then play the recording back, perfectly flat, to the 19-year-olds who've just banged out their "song" (and whose ears are probably ringing), you'll be thinking, "How can they expect to hear *any* high frequencies at all, with those tattered, torn up eardrums of theirs??"

But then: somebody, I absolutely guarantee it, is going to say, "Needs more bass, dude!"

I'm sure you know this, but to lay it out for the sake of the larger discussion:

Whether it's color reproduction in a photo or frequency reproduction in a song, there's often a difference, and sometimes a large difference, between what is accurate and what subjectively looks good or sounds good. Our senses and sensibilities often prefer departures from the measurable reality. Accurate color many not look good, and flat frequency response may not sound good.

But I don't think that takes away from the usefulness of developing a system that achieves accurate color or flat frequency response. Such a system provides a better starting point and a better tool for the artist to decide what and how much deviation from accurate is aesthetically desirable, and then apply it. The alternative is typically a far less precise and less controllable 'two wrongs might make a right' approach.

Recently I shot some portraits. I started with couple of shots with a calibrated white balance tile. The raw converter said that white balance tile indicated neutral would be about 8750 K and -5 on tint. But 'accurate' was not (in my personal opinion) flattering for the portraits, so I manually dialed in more like 8950K and +5, i.e., overall I shifted the color a little away from cyan and toward red. I find that dialing in what I want is usually much easier starting from a calibrated neutral than it is from some 'it is what it is' system.

 

[JimKasson]

In thinking about all of of this, I'm finding myself comparing it to something I have a pretty good understanding of: audio capture, recording, and reproduction....

If you manage to achieve and maintain what's called "flat response" throughout the chain, and then play the recording back, perfectly flat, to the 19-year-olds who've just banged out their "song" (and whose ears are probably ringing), you'll be thinking, "How can they expect to hear *any* high frequencies at all, with those tattered, torn up eardrums of theirs??"

But then: somebody, I absolutely guarantee it, is going to say, "Needs more bass, dude!"

I'm sure you know this, but to lay it out for the sake of the larger discussion:

Whether it's color reproduction in a photo or frequency reproduction in a song, there's often a difference, and sometimes a large difference, between what is accurate and what subjectively looks good or sounds good. Our senses and sensibilities often prefer departures from the measurable reality. Accurate color many not look good, and flat frequency response may not sound good.

But I don't think that takes away from the usefulness of developing a system that achieves accurate color or flat frequency response. Such a system provides a better starting point and a better tool for the artist to decide what and how much deviation from accurate is aesthetically desirable, and then apply it. The alternative is typically a far less precise and less controllable 'two wrongs might make a right' approach.

Recently I shot some portraits. I started with couple of shots with a calibrated white balance tile. The raw converter said that white balance tile indicated neutral would be about 8750 K and -5 on tint. But 'accurate' was not (in my personal opinion) flattering for the portraits, so I manually dialed in more like 8950K and +5, i.e., overall I shifted the color a little away from cyan and toward red. I find that dialing in what I want is usually much easier starting from a calibrated neutral than it is from some 'it is what it is' system.

I don't think the accurate vs pleasing color discussion has much to do with which color space you use for editing.

 

[mplkn]

These are good points, and, right, I wasn't attempting to draw the distinction between what's accurate vs. what's pleasing, but rather, between what's objective and measured in relation to a standard vs. what's judged subjectively.

In other words, I was cheering, from the peanut gallery, for that which is measured in relation to a known standard, vs. that which is picked by "feel".

So even if making a choice to create "pleasing" output vs. "accurate", I would want my measurement of whatever I'm doing to be as accurate, and also, as precise as possible, if only to make repeatability possible.

I don't mean to open the can of worms entailed in the ideas of accuracy vs precision, or at least, not to leave it standing open for too long.

But - it's like this: you'd want your thermometer to be accurate, but even more, you'd want it to be precise. That is, you'd want it to read the same repeatedly when placed in conditions of a particular temperature.

If your thermometer is precise, after comparing its readings to those provided by the reference standard thermometer, you could repeatedly use your thermometer to calculate accurate measurements, by knowing how far off the reference standard its readings are.

Anyway, the peanut gallery cheers the discussion!

Colorspaces of known characteristics! Way cool!