ipf8400 gamut

[digitaldog]

Robert, i am lost as to what you try to prove. Would like to see some serious proof of all these theories of colorshifts. I have not such experiences.

That would be useful (proof of concept) with outlined methodology we can try on our end. At least that was what I attempted to do with the ProPhoto vs. sRGB to print demo. As such, there's no question in my mind, with the raw images with large chroma I used, out to my 3880, ProPhoto RGB is the right answer for an encoding color space.

[JRSmit]

That would be useful (proof of concept) with outlined methodology we can try on our end. At least that was what I attempted to do with the ProPhoto vs. sRGB to print demo. As such, there's no question in my mind, with the raw images with large chroma I used, out to my 3880, ProPhoto RGB is the right answer for an encoding color space.

So do i andrew. I am getting tired of these posts on how big prophotorgb is, for me i t means the person has no clue as to what a color space is. But i want to learn, so a reproducable concept yes. P.S. still  need to digest your video.

[Robert Ardill]

Robert, i am lost as to what you try to prove. Would like to see some serious proof of all these theories of colorshifts. I have not such experiences.

Hi JR,

I'm not trying to prove anything ... I just don't agree with the notion that, effectively, the larger the working space gamut the better, and that using a smaller gamut will necessarily cause banding, clipping or shifting of colors, especially low luminosity colors.  I'm happy with how I do things and if you're also happy with how you do things then we're all happy.

But if you want to follow through on the question of color shifts in perceptual mappings, have a look here: http://www.argyllcms.com/doc/iccgamutmapping.html.  And if you want to get a more thorough explanation ask Graham Gill who is the author of ArgyllCMS.  If he doesn't know what he's talking about then we're all in trouble!

Essentially it works like this: when a profile developer creates a perceptual mapping table he/she does not know what colors will be in the source space (say in ProPhoto RGB) because these colors are only known for individual images.  With some images all of the colors may be within the destination space, with other images some of the colors may be at the edges of the source space.  So the profile developer can either a) assume the worse case that the general case is that image colors will be on the source boundary, or b) assume that the image colors will be in a smaller space than the source space.  

If a) is assumed then for a (perceptual) mapping, the mapping table will need to compress the whole of the source space (ProPhoto, say) into the destination space.  The table will attempt to minimize the color shifts that lie within the destination space, but the whole idea behind perceptual mappings is to preserve the relationship between colors: in order to do that the mapping will have to shift the colors that lie within the destination space.

If b) is assumed then the profile will be a mix of a relative and perceptual mapping.  The same will happen as in a), but for the smaller source space; and the colors outside of the smaller source space will effectively be mapped using a relative-type mapping. In other words the colors will be clipped to the smaller source space boundary before being shifted.  This will result in banding for colors that are outside of the smaller source space, but it will give less of an overall color shift.

The only way around this problem is for the mapping table to be constructed immediately prior to the conversion from source to destination.  Unfortunately this is a very compute-intensive task.  However, if you want to do this then ArgyllCMS does provide a mechanism to do so - and this will ensure that the sort of color shifts I'm talking about will be minimized (but only that: the color shifts will still occur if there are colors that are outside of the destination space).  What you do is to create a profile for each image.  This sounds complicated, and it is a bit, but not so much.  What you need to do is to run the profile mapping against the image using the tables you've already generated from your spectrophotometer and use this image-specific profile to convert to the destination space.  

Most of us aren't such perfectionists that we'll do this: a half-way-house solution is to use relative mappings (which result in clipping but not in the same color shifts) or to use smaller working spaces in order to minimize the shifting.

At any rate, as I said, you don't need to take my word for this: Graham Gill is a really nice guy and he will answer your questions if you post them here: argyllcms@freelists.org.

Robert

[digitaldog]

Hi JR,

I'm not trying to prove anything ... I just don't agree with the notion that, effectively, the larger the working space gamut the better, and that using a smaller gamut will necessarily cause banding, clipping or shifting of colors, especially low luminosity colors.  

The proof is in the output. Tests can be preformed to educate each person as to the most effective workflow. There is zero question in my mind that with raw data, using the converter I use, (which processes in ProPhoto gamut), with the images I selected (very saturated), ProPhoto is the right answer going out to my printer. The gamut maps are interesting but don't prove anything over what the print shows me! The gamut maps actually show me why IMHO the larger gamut produces a superior print (which has nothing to do with clipping alone). If I were using an image with a much lower gamut, and there are areas of this in my Gamut Test File, I see nothing the larger space provides that is better than the smaller gamut space. But nothing worse either!

I can attempt to view each image and decide what encoding color space they best fit into, or I can pick a really big one seeing that bigger IS better upon output and not worse than smaller on the same output. I suppose if I only had to process a few images, I could pick and choose but with high bit, wide gamut data in a raw processing using a wide gamut high bit processing color space, I don't see any benefit to doing this. Maybe you can provide some data to suggest otherwise. That's all we're asking.

[Robert Ardill]

That would be useful (proof of concept) with outlined methodology we can try on our end. At least that was what I attempted to do with the ProPhoto vs. sRGB to print demo. As such, there's no question in my mind, with the raw images with large chroma I used, out to my 3880, ProPhoto RGB is the right answer for an encoding color space.

Well, Andrew, if the image you start off with has colors that are well outside of sRGB, as is the case with your ProPhoto test image, then when you convert to sRGB you are immediately clipping all of the OOG colors to the sRGB space using a relative intent.  You don't need to print to see that - you will see the effect on your monitor since your monitor has a near aRGB gamut.  Since your printer gamut will mostly be wider than the sRGB gamut, little or no further damage will be done when you map to the print profile: the damage has already been done.

It's like taking a pint glass full of water and pouring it into a half-pint glass ... and then saying: "Jeez! I only have a half-pint left!  The action of converting to the half-pint glass has clearly resulted in a massive loss of water!".

The fact is that your printer is a 3/4 pint glass so don't convert to a 1/2 pint glass before filling it.

Robert

[digitaldog]

Well, Andrew, if the image you start off with has colors that are well outside of sRGB, as is the case with your ProPhoto test image, then when you convert to sRGB you are immediately clipping all of the OOG colors to the sRGB space using a relative intent.

So what? It's like saying if you start with a 21 megapixel file and sample for the web, you lose pixels. Yes you do, so what? If you have an sRGB output need (which I would suggest is only to the internet or non color managed devices), you have to use sRGB and converting clips the colors and they look fine.

It's like taking a pint glass full of water and pouring it into a half-pint glass ... and then saying: "Jeez! I only have a half-pint left!

IF you have a pint of water and the recipe calls for a half pint, you follow the recipe, so what? If the 2nd recipe calls for more water, you've got it in the first place. Clipping is a fact of life, there's nothing wrong with it. It has to be done. The question is, do you start with less when you may often need more or start with more when you need it and end up with less when you need it? I makes no sense to me, others I suspect and clearly the people who built the Adobe raw engine to start off with less.
You've as yet provided nothing in terms of a testing methodology that suggests it is better to start with less than start with more (color gamut). That's where I suggest you focus the efforts.

[Robert Ardill]

If I were using an image with a much lower gamut, and there are areas of this in my Gamut Test File, I see nothing the larger space provides that is better than the smaller gamut space. But nothing worse either!

If you are using a relative mapping then the colors that are both within your source profile and destination profile won't be affected, so you gain / lose nothing.  If you're using a perceptual mapping there will be larger color shifts with the larger source space - but you may not notice these if you have a good profile that succeeds in maintaining the relationship between the colors well (not an easy task as I understand it!).

I can attempt to view each image and decide what encoding color space they best fit into, or I can pick a really big one seeing that bigger IS better upon output and not worse than smaller on the same output. I suppose if I only had to process a few images, I could pick and choose but with high bit, wide gamut data in a raw processing using a wide gamut high bit processing color space, I don't see any benefit to doing this. Maybe you can provide some data to suggest otherwise. That's all we're asking.

That's absolutely the whole point of the discussion.  What we're trying to do is to squeeze the most out of our images, which is why we buy really expensive cameras, lenses, software, monitors and printers.  If we don't care all that much or process so many images that we simply don't have the time then the quality will suffer: we won't sharpen as well, compose the shot as well, develop as well, print as well.  Then it makes sense to have as simple and as efficient a workflow as possible ... and accept the probable quality-hit.

If, on the other hand, we want the best that we can achieve from each of our images ... well then we will take the shot with a tripod and mirror lock-up, make sure the exposure is right ... etc., etc., ... to include the best working space for this image and the best rendering intent using the best profile and best paper.

My only real issue with what you say is that you are making the claim that 'ProPhoto is best'.  It is not.  It's a compromise like all the other workspaces.  The ICC should not have defined a triangular working space when the output devices are non-linear ... but it has and that is what we are currently stuck with.  Hopefully we will be moving towards on-the-fly profile mapping which will improve things quite a lot, but right now you are better off using a working space that fits your image.

Robert

[digitaldog]

If you are using a relative mapping then the colors that are both within your source profile and destination profile won't be affected, so you gain / lose nothing.

Great, losing nothing is terrific when I can have the best of both worlds!

 If you're using a perceptual mapping there will be larger color shifts with the larger source space - but you may not notice these if you have a good profile that succeeds in maintaining the relationship between the colors well (not an easy task as I understand it!).

You simply can't make blanket statements about perceptual rendering because they all differ. It is probably why the canned Epson profile for Glossy did a worse job than my custom profile on the Blue Balls (referenced in the forum post about color management misconceptions). A perceptual mapping my be better or worse based on the profile and how it was built and the image. That is why we soft proof before we make such decisions. Profiles only understand a solid color going from one color space to the other, it has zero idea about colors in context.

That's absolutely the whole point of the discussion.  What we're trying to do is to squeeze the most out of our images, which is why we buy really expensive cameras, lenses, software, monitors and printers.  If we don't care all that much or process so many images that we simply don't have the time then the quality will suffer: we won't sharpen as well, compose the shot as well, develop as well, print as well.  Then it makes sense to have as simple and as efficient a workflow as possible ... and accept the probable quality-hit.

IF that's the whole point of the discussion, then ProPhoto RGB, certainly with Adobe raw converters is the right answer for the encoding color space with the images I used. You are welcome to test other images in that converter and show us where encoding in a smaller working space is better.

If, on the other hand, we want the best that we can achieve from each of our images ... well then we will take the shot with a tripod and mirror lock-up, make sure the exposure is right ... etc., etc., ... to include the best working space for this image and the best rendering intent using the best profile and best paper.

That's fine too. That's why people should test differing workflows, to decide what it is they are willing to do for presumed or expected image quality.

My only real issue with what you say is that you are making the claim that 'ProPhoto is best'.  It is not.

I never said that. With the raw images and converter, with the printer and methodology I used, it's the best (better) than sRGB by a long shot. Hugely better. You've got the same file and instructions for the methodology I used. You can add your own image. You can try a different raw converter (Aperture uses Adobe RGB gamut for processing, so YMMV). You can use a different printer or profile. When you do this, show us when and where sRGB produces a better print than ProPhoto RGB.

[Robert Ardill]

So what? It's like saying if you start with a 21 megapixel file and sample for the web, you lose pixels. Yes you do, so what? If you have an sRGB output need (which I would suggest is only to the internet or non color managed devices), you have to use sRGB and converting clips the colors and they look fine. IF you have a pint of water and the recipe calls for a half pint, you follow the recipe, so what? If the 2nd recipe calls for more water, you've got it in the first place. Clipping is a fact of life, there's nothing wrong with it. It has to be done. The question is, do you start with less when you may often need more or start with more when you need it and end up with less when you need it? I makes no sense to me, others I suspect and clearly the people who built the Adobe raw engine to start off with less.
You've as yet provided nothing in terms of a testing methodology that suggests it is better to start with less than start with more (color gamut). That's where I suggest you focus the efforts.

You know Andrew, I have the greatest respect for you and your teaching is of real benefit to us photographers and printers.  You clearly know a lot about the subject and have given it a lot of thought.  But I think you have nailed your flag to the ProPhoto mast and cannot unnail it.

I have not suggested, at any stage, that it is best to start with less.  What I have suggested is that it is better to start off with the right sized container for your image.  If you only need a teacup then don't use a boat.  If you need a boat, use a boat.

The best way I know of testing this question is to look at the profiles because they tell you exactly what is happening to the image as it is being converted from one color space to another.  Doing it visually is very difficult because it is subjective and depends on the image, the profile, the printer, the paper ... and on our eyes and the viewing conditions.  It's easy to demonstrate clipping from ProPhoto to sRGB for colors that are outside of sRGB: you've already done that very effectively.  It is much more difficult to visually demonstrate the shifting of colors that can occur when compressing a large gamut into a smaller one.  But by looking at the profiles it is quite easy to see, at least in general terms, what is going to happen.

If I did spend a lot of time devising a test to demonstrate the shifting of colors you would almost certainly disagree with my conclusions.  But if I was bothered about it (which I'm not because I don't need any convincing), what I would do is to pick a spectrum of color spots going from the edges of ProPhoto to grayscale, note the Lab value of each of the color spots, print to your 3880 with your profile and then, using my i1Pro2, I would measure the Lab value of each of the printed spots.  I would then enter all of this data into an Excel spreadsheet and plot the shift in colors.  I would then repeat the test with sRGB and compare its plot to the ProPhoto plot.  Providing your printer gamut is larger than sRGB I would expect to see a curve for the ProPhoto plot and a straight line for the sRGB plot.

But it is a lot easier just to inspect the mapping!

Robert

[digitaldog]

But I think you have nailed your flag to the ProPhoto mast and cannot unnail it.

I've done the testing which I believe is based on sound methodology of which you've as yet haven't disproved. If and when you can and do, I'll re-examine things. On the other hand, you've done nothing to suggest that with the workflow I'm using, ProPhoto isn't the best answer. So the ball is in your court.

What I have suggested is that it is better to start off with the right sized container for your image.

Define right, tell us how a raw image falls into that container especially when the processing color space is a fixed, wide gamut based on ProPhoto.

The best way I know of testing this question is to look at the profiles because they tell you exactly what is happening to the image as it is being converted from one color space to another.

 That's kind of absurd considering that the images don't have a gamut until processed from raw and encoded (at which time, it's all water under the bridge).

Doing it visually is very difficult because it is subjective and depends on the image, the profile, the printer, the paper ... and on our eyes and the viewing conditions.

 
Yes it is subjective and useful when the final result IS THE PRINT itself!

It's easy to demonstrate clipping from ProPhoto to sRGB for colors that are outside of sRGB: you've already done that very effectively.  It is much more difficult to visually demonstrate the shifting of colors that can occur when compressing a large gamut into a smaller one.  

It is actually very easy; look at the prints side by side. You're making this far more complex and difficult than it needs to be.

But by looking at the profiles it is quite easy to see, at least in general terms, what is going to happen.

Looking at the print tells me exactly what's happening!

If I did spend a lot of time devising a test to demonstrate the shifting of colors you would almost certainly disagree with my conclusions.

OK, that's not useful.

 But if I was bothered about it (which I'm not because I don't need any convincing), what I would do is to pick a spectrum of color spots going from the edges of ProPhoto to grayscale, note the Lab value of each of the color spots, print to your 3880 with your profile and then, using my i1Pro2, I would measure the Lab value of each of the printed spots.

 Why would one need to measure anything? Just view the print (in differing illuminants if you so desire). The proof is in the print. That's the final process for many of us. It's pointless to measure anything, especially when at this point, the working space is a far removed step in the process to make the print.
You do realize that by the time you've gone from RGB working space to output profile, there's a step in between that has zero idea what the working space was. By the time the profile connection space comes into play, the RGB working space is history and gone from the scene.

But it is a lot easier just to inspect the mapping!

It is even easier just to look at the damn print!

[Robert Ardill]

You simply can't make blanket statements about perceptual rendering because they all differ. It is probably why the canned Epson profile for Glossy did a worse job than my custom profile on the Blue Balls (referenced in the forum post about color management misconceptions). A perceptual mapping my be better or worse based on the profile and how it was built and the image. That is why we soft proof before we make such decisions. Profiles only understand a solid color going from one color space to the other, it has zero idea about colors in context.

Absolutely correct - and I made this very point in an earlier post.  The mapping table is dumb (well pretty dumb): it has no knowledge of the image.  However the profile does 'know' the context: it is source to destination. The skill in creating a really good perceptual profile is to preserve the relationship between the colors as closely as possible while minimizing the color shifts.  And so it is possible to make a statement about perceptual mappings ... to the extent that they conform to the ICC recommendations (pretty vague).  The fact is that you cannot squeeze a peach into a grape without compressing the fruit.

[/quote]

IF that's the whole point of the discussion, then ProPhoto RGB, certainly with Adobe raw converters is the right answer for the encoding color space with the images I used.

Totally - you started off with an image with a very wide gamut so you should use a wide working space.  No argument there.

Robert

[digitaldog]

The fact is that you cannot squeeze a peach into a grape without compressing the fruit.

And there is nothing wrong with that if you need a peach the size of a grape. If you don't, stick with the larger peach size.

[Robert Ardill]

Define right, tell us how a raw image falls into that container especially when the processing color space is a fixed, wide gamut based on ProPhoto.  That's kind of absurd considering that the images don't have a gamut until processed from raw and encoded (at which time, it's all water under the bridge).  

It's very easy to define the best working space post Lightroom.  You just soft-proof with sRGB ... OOG.  Then aRGB ... OOG.  Then Beta RGB ... OK.  So you either pick Beta RGB (or ProPhoto if you must) or adjust your image so that it comes within aRGB.  If the image is not OOG in sRGB then pick that.  If, while editing in Photoshop, you find that you are going OOG in sRGB then move up to aRGB or adjust your edit.

Yes it is subjective and useful when the final result IS THE PRINT itself!  It is actually very easy; look at the prints side by side. You're making this far more complex and difficult than it needs to be. Looking at the print tells me exactly what's happening!  
OK, that's not useful.   Why would one need to measure anything? Just view the print (in differing illuminants if you so desire). The proof is in the print. That's the final process for many of us. It's pointless to measure anything, especially when at this point, the working space is a far removed step in the process to make the print.
You do realize that by the time you've gone from RGB working space to output profile, there's a step in between that has zero idea what the working space was. By the time the profile connection space comes into play, the RGB working space is history and gone from the scene. It is even easier just to look at the damn print!

I don't think I'm making it complex Andrew.  I'm entirely convinced that your image is better left in ProPhoto than being converted to sRGB.  I absolutely do not need to print to know that: it's self-evident because the image has very saturated colors well outside of sRGB, and your printer gamut is considerably larger than sRGB.  Converting it to sRGB is doing it a serious injustice.

I also do not need to print to know that an image that is within sRGB will in no way benefit from being converted to ProPhoto before printing.  If you really want to prove the point by printing (I don't) then take an image that is saturated in  sRGB, convert it to ProPhoto and then print both using a perceptual mapping (WITHOUT reconverting the image to sRGB before printing).

This reminds me that I had another issue with your video: you converted the wide-gamut image to a small color space before printing.  To compare like with like from a printing point of view, you should have converted the ProPhoto image to sRGB, then converted it back to ProPhoto and printed that image together with the sRGB image.  Otherwise what you are doing is altering the image BEFORE printing and comparing this to the print of the unaltered image.

That's almost as bad as applying a gaussian blur to a copy of an image, printing it along with the original and saying: "Look ... the blurred image is blurred!".

Robert

[Robert Ardill]

And there is nothing wrong with that if you need a peach the size of a grape. If you don't, stick with the larger peach size.

Then we're in agreement

Robert

[digitaldog]

It's very easy to define the best working space post Lightroom.  You just soft-proof with sRGB ... OOG.  Then aRGB ... OOG.  Then Beta RGB ... OK.

You'll never see OOG with ProPhoto so based on the above, that's always the right answer. Why go smaller, that's what you've failed to provide.

This reminds me that I had another issue with your video: you converted the wide-gamut image to a small color space before printing.  To compare like with like from a printing point of view, you should have converted the ProPhoto image to sRGB, then converted it back to ProPhoto and printed that image together with the sRGB image.

Why on earth would I or anyone do that? It buys you nothing! I take the pint container and pour half of it out into the half pint container. Then I pour that half pint back to the larger container for what?

 Otherwise what you are doing is altering the image BEFORE printing and comparing this to the print of the unaltered image.

  Yup, and that's proper color management! The idea is to show what you gain and lose by picking either ProPhoto OR sRGB for a working space before sending it thought the output profile.

That's almost as bad as applying a gaussian blur to a copy of an image, printing it along with the original and saying: "Look ... the blurred image is blurred!".

But if the original was in need of blurring and the blurred print looked better, then the blur was the right answer. The proof again is in the print. Using sRGB produced an inferior print compared to using Prophoto RGB. Simple. I think you've missed the point entirely based on your 'convert to sRGB and back to ProPhoto' comment.

[Robert Ardill]

Actually I was about to stop before we all go completely gaga, but there is one point that keeps coming up that I would like to address.  And that is, to paraphrase: “Adobe picked ProPhoto for Lightroom, they didn’t do it for a silly reason, so this means that ProPhoto is the best working space”.

If I was picking a working space for Lightroom I also would probably pick ProPhoto – simply because there is no way to predict what the gamut of every image will be.  But it is quite easy to predict that the union of all of the gamuts of all the images that have been and will be processed will be very large.  So the developer of a raw converter needs to pick a working space with a large gamut.

That does not mean that a large gamut is the best gamut for each individual image.

If the gamut of the image can be contained in a smaller working space then there is no loss of quality in converting it to the smaller working space.

So using a large working space as the (initial) container for the raw image is correct. (I’m sure Adobe will be relieved to hear me say that  .

Robert

[digitaldog]

That does not mean that a large gamut is the best gamut for each individual image.
If the gamut of the image can be contained in a smaller working space then there is no loss of quality in converting it to the smaller working space.

Why do this? Produce a test whereby the final is output to a print where you can prove that after using the wide gamut processing within ACR or LR, encoding into something smaller is better. That's all you need to do. At that point, you've validated that one should if possible, view the image and compare it to the encoding space from raw. Otherwise you haven't.

I've got to go raw to some RGB working space after which I plan to work on it so I need a master image and working space that I can move forward with to ALL output needs. What encoding working space should I use?

[Robert Ardill]

Why on earth would I or anyone do that? It buys you nothing! I take the pint container and pour half of it out into the half pint container. Then I pour that half pint back to the larger container for what?   ck to ProPhoto' comment.

If the prints are identical then you will have proved me wrong ... that is what it will prove (or disprove).

Using sRGB produced an inferior print compared to using Prophoto RGB.

Quite correct - I have already agreed and are happy to do so again: you damage an image by squeezing it into a smaller space before outputting it to a larger space.  Of course the sRGB print will look worse: you've chopped the legs off it.

Robert

[Robert Ardill]

Why do this? Produce a test whereby the final is output to a print where you can prove that after using the wide gamut processing within ACR or LR, encoding into something smaller is better. That's all you need to do. At that point, you've validated that one should if possible, view the image and compare it to the encoding space from raw. Otherwise you haven't.

Andrew ... this has been fun but I think we've hacked this one to death.  I really have answered your question several times already on this very page.  Answering it again is going to drive us loopy

Robert

[digitaldog]

I really have answered your question several times already on this very page.  Answering it again is going to drive us loopy

I know you think you did, anyone else here agree?