The Red Feather Duster

[Colin Southern]

Hi Chris,

I edit my images in the ProPhoto colour space (it has the widest colour range and so any graduations in tones should be smoother).

Um no, just the opposite unfortunately. Think of it as being like a staircase with a set number of steps (256 for 8 bit, and 32768 for "16 bit") (it's called 16 bit because 2 bytes are used to represent it, but in reality Photoshop only uses 15 bits, and most camera analog to digital converters are ony 12 or 14 bit). So - the bigger the distance that the stairs need to cover (from the bottom to the top), the greater the distance between them for a given number of steps - and thus the greater the chance of graduations between tones being visible.

That's the theory anyway - in practice - the human eye can barely discern 200 levels - so you're unlikely to see any graduations with even 8 bit representation. BUT (and as you can see, it's a big BUT) - things change when tonal ranges get compressed and stretched as part of normal post-processing - and that's when other factors come into play. As a rule though - one is unlikely to have problems with graduations with either a 8 or 16 bit representation, so long as the image doesn't require large corrections.

So what IS Prophoto good for? To be honest, not a lot. It does have a very wide gamut - but - unfortunately - we don't currently have any device capable of reproducing some of the colours it is capable of capturing. Monitors have traditionally been sRGB only gamut devices (with a trend towards Adobe RGB gamut), and many photo printers are oriented around an Adobe RGB gamut (with the ironic exception of the printers at most photo "labs"). To put that another way, ...

- If you capture RAW and process to display on the web, you have to reduce the gamut to sRGB

- If you capture RAW and process for prints at your local print shop, again you have to reduce the gamut to sRGB

- If you capture RAW and process for prints you do yourself, you end up effectively reducing to an Adove RGB gamut

Some purists still like to work with a Prophoto gamut so that they're not "throwing away colours", but (a) they're not colours that the untrained eye misses (because close colours are substituted), and (b) there is a very real danger of manipulating the image into areas that can't be displayed or printed - BUT - (here's the kicker) - the monitor and printer drivers WILL STILL RENDER THESE COLOURS AS BEST THEY CAN - but they'll almost certainly be rendered differently between monitor and printer; so you end up playing with things like bright reds (that the monitor can reproduce well because it has a red channel), but the printer can't print because it doesn't have red ink, and you get a gamut (colour) mismatch that will drive you nuts.

So in a way - often - most people are better off using sRGB for web / shop printing, or Adobe RGB if they're doing their own printing; Prophoto is fine in theory, but (a) you have to know what you're doing with it ("with great power comes great opportunity to stuff things up"!), and (b) generally, we have no way to display or print the extra colours anyway -- so why not just start with the biggest gamut that one can actually use, and avoid any potential pitfalls?

After completing my editing I save the image as a 16 bit, unsharpened .tiff file.

Since ALL images need capture sharpening, you may as well apply that sharpening phase. Also, if you're working in Photoshop, you may just as well save as a PSD - it's more compatible with Photoshop, and generally smaller.

FOR PRINT
When I need the image for print I open the .tiff file, apply 1 level of sharpening, resize, apply 2nd sharpening and assign the Adobe RGB profile before printing.

FOR WEB
I do exactly as 'For Print' except apply the sRGB profile before saving as .jpg.

We need to be careful about terminology here - the only time we should be ASSIGNING a profile is when the image doesn't already have one ("mystery meat" as we call it) - if it already has one then it's VITAL that we CONVERT to the new profile, not ASSIGN it - they're very different operations.

Hope this helps

[CBImages]

Thank you Colin for helping to brush away the mists of confusion, although I still don't fully understand your illustration of the stairs.
Surely a graduation i.e. from black to white would be smoother if it 32,768 steps (15 bit) than if it had 256 (8 bit) irrespective if all those exact steps or graduations could be displayed on a monitor?
It would seem that a problem may occur when that 15 bit image is reduced to 8 bit and the software has to try and assign 256 shades of grey to a graduation that contains 32,768 shades?
Ahh have I just answered my own question?
Its not so much the 32,768 steps that is the problem, its the squashing of them into 256 steps - which has to be done at some stage for output purposes, either printing or as a .jpg for the web.

I thought I was beginning to understand things, I listened to a chap give a talk on HDR photography earlier this year and he did blow my mind somewhat when he started talking about 32 bit floating decimal point calculations.
What I did come away understanding is that the human eye can discern an exposure range of about 22 stops, most consumer digital cameras can capture around 8 or 9 stops (this can be enhanced with post processing) yet a normal monitor can only reproduce about 9 stops and paper can only capture about a 7 stop range. So in effect we can not reproduce what we see, nor even what we capture with our cameras or for that matter what we process in programmes like Photoshop.

I can certainly now see a case for NOT processing my images using ProPhoto as it will almost certainly introduce colours that can't reproduced by my monitor and printer. Even though both my monitor and Epson 2880 are profiled they may not render the colours introduced by ProPhoto in the same manner when the image is 'converted' to AdobeRGB for printing.
I suppose the best solution would be to shoot in RAW, then process and convert to sRGB which would save lots of problems with printer/paper profiles etc?

The reason why I save an image 'unsharpened' is that I can then apply different levels of sharpening depending on how the image will be output, it also saves time should I want to undertake carry out some additional processing on an image in the future.

Thanks Colin

"The more I learn the less I realise I truly know"!

[Colin Southern]

Hi Chris,

You're kinda on the right track, but lets see if we can refine a few things ...

With regards to colourspaces, Prophoto is bigger than sRGB - but we only have the same (limited) range of numbers to describe "where things sit" in each. Perhaps we could think of sRGB as being like a fish tank that was 256 inches high (ignore the other dimensions for now), and Prophoto as being like a fish tank that was 512 inches high. With the sRGB (smaller) fish tank since we only have 256 levels to play with, each level would only represent a change of 1 inch in height, whereas with the Prophoto fish tank - we still only have 256 levels, but each one must now represent a change of 2 inches (or their won't be enough levels to get us to 512 inches). So the bigger the tank (colourspace), the bigger the size of each step (when the number of steps are limited) - and the bigger the size of the step, the more likely it is that it'll be noticed. Does that help?

With regards to dynamic ranges, most modern cameras capture around 12 stops - most monitors display around 6 stops, and most prints can display around 4 stops (also, the human eye CAN differentiate around 22 stops, but not all at once - so it's not comparing apples with apples to compare it to a camera having 12 stops). Regardless, the issue with cameras becomes not so much an issue in capturing most dynamic ranges (purely reflective scenes are only 4 stops), but it becomes more of an issue trying to display that range (in essence, one normally just can't do it unless one compresses that range into something we can either display on our screens, or print on a paper.

Surely a graduation i.e. from black to white would be smoother if it 32,768 steps (15 bit) than if it had 256 (8 bit) irrespective if all those exact steps or graduations could be displayed on a monitor?

Yes, but the steps will still be larger if one is using a larger colourspace, regardless of weather they're finer 15 bit ones, or coarser 8 bit ones. Also, in terms of final representation of an image, 8 bit is always sufficient as we can only differentiate - at best - 200 levels on a monitor, and 8 bit gives us 256.

It would seem that a problem may occur when that 15 bit image is reduced to 8 bit and the software has to try and assign 256 shades of grey to a graduation that contains 32,768 shades?

It's not really a problem per se - it's just that at that point a lot of "rounding up" and "rounding down" is performed, and a lot of information is discarded - but - if it's done at the appropriate part of the workflow (ie after all large tone adjustments have been performed) then it's visually lossless. The problems occurs when one does it too early in the workflow - and then tries to stretch a tonal range too far - then the graduations can become obvious. In really, conversion to 8 bit & sRGB is pretty much the last thing I do before saving to JPEG.

I can certainly now see a case for NOT processing my images using ProPhoto as it will almost certainly introduce colours that can't reproduced by my monitor and printer. Even though both my monitor and Epson 2880 are profiled they may not render the colours introduced by ProPhoto in the same manner when the image is 'converted' to AdobeRGB for printing.
I suppose the best solution would be to shoot in RAW, then process and convert to sRGB which would save lots of problems with printer/paper profiles etc?

Yes / no / maybe

It all depends on what you're doing. Case in point - if I shoot a studio session RAW - and process using Prophoto, it's unlikely to cause any problems because I'm aiming for accurate colour reproduction anyway. In any case I wouldn't want to restrict myself to sRGB because my 7800 printer can reproduce a wider gamut that that, so Adobe RGB or Prophoto would both give a good result. On the other hand - if I capture a landscape scene containing brightly coloured flowers - and start "playing" with those colours, then the "colour" of the flower as represented internally by Photoshop may well be different to what the monitor can display which can be different again to what the printer can reproduce. The classic example of this is when I photographed a couple of fishing boats with bright red hulls; the brighter the hulls looked, the better the photo looked - but when I tried to print it, I ended up with a hue shift towards orange, and very uneven & inconsistent colours on the print. It drove me nuts - I ended up reprofiling the printer - doing nozzle checks etc - until I did a gamut check / soft proof and "BINGO" - I was trying to print out of gamut.

With regards to sharpening, I agree - but capture sharpening is independent of output size or if additional processing is needed - it's done regardless, usually as one of the first steps in a workflow (although sometimes dust bunny removel is easier if done before capture sharpening).

Hope this helps

PS: Forgot to mention - out of gamut situations are handled by what are called Rendering Intents (Relative Colorimetric, Absolute Colourimetric, Saturation, and Perceptual) RC is probable the most common, and normally does a pretty good job of smoothly adjusting all colours so that they fit into the gamut available in a "visually inconspicuous" way. If you want to learn more about how the whole colour management thing works, grab a copy of Real world Colour Management 2nd edition, by Fraser, Murphy, and Bunting. It'll make your head hurt, but the knowledge gained is invaluable - especially if you're doing your own printing.

[FrankMi]

Thank you Chris and Colin for making this issue, and what can and should be done to resolve it, clearer.

I have discovered that if you start your post processing in Lightroom, it uses the ProPhoto RGB color space internally and will always embed a profile for this color space within images that are exported via the Edit in Adobe Photoshop command.

The quickest and easiest way to avoid the problem is to configure the Lightroom External Editor preference (Edit/Preferences/External Editing) so that images exported from Lightroom using the Edit in Adobe Photoshop command are already converted to the preferred color space. The options are File Format: PSD/TIFF, Color Space: ProPhoto/Adobe RGB/sRGB, Bit Depth: 8/16, and Resolution, which on my system is set to 240.

Alternatively, you can configure the Photoshop Color Settings dialog (Edit/Color Settings) to set your color working space for RGB to sRGB and the Color Management Policy to use the RGB workspace rather than to Preserve the Embedded Profile.

Hopefully, this well help others to avoid the pain I went through when I started exporting files from Lightroom to Photoshop.

[CBImages]

Colin
Thank you for all that information, I have read through it 4 or 5 times and am still digesting it but I now have so many more questions. Sadly the cat is now out of the bag, I thought I understood things but its obvious after reading your posts I need to go away, think some more and even buy that book you suggested.
I did buy a book at the same time I bought my Epson 2880 - Epson Complete Guide To Digital Printing by Rob Sheppard, sadly it was a complete waste of money, he spent the first 60 pages telling the reader 'why they needed the book'. It put me off buying other publications on the subject but I will take a look at the one you recommended as I am the kind of person who isn't happy just 'half understand something'.

I think one thing is clear - colour management COULD be made so much easier if all parties (Apple, Microsoft, Adobe, Epson, Canon, Nikon etc) got together, created a generic profile and stopped trying to 'empire build'.
Or is that too simplistic?

I doubt I will sleep much tonight with all those fish tanks, stairs, colour spaces floating around in my head.

I will let you know when the book arrives.

Thanks again

Chris

[Colin Southern]

@ Chris,

No worries Chris. The book is pretty much the industry standard reference text, and (the late) Bruce Fraser is legendary for his knowledge and ability to communicate (he was very much an Adobe "insider" - you'll even see him mentioned in the credits).



In terms of a standard profile - in essence - that's what sRGB was intended to be - kinda like "the lowest common denominator" (and it was a good one), but there's no way one could ever come up with a truely accurate standard profile for the likes of printers because it depends on the characteristics of the printer + paper + ink set + media settings - so there are just too many variables, but not to worry - that's what colorimeters and spectrophotometersmeters are for

Feel free to ask any other questions you have though.

[CBImages]

Good Morning Colin

Well it is morning over here and a fine one at that, I expect the sun is over the yardarm with you and its time to settle down for the evening with a nice glass of red.

I have searched eBay and Amazon for the 2nd Edition Real World Colour Management and there aren't any copies available in the UK, apart from those at stupid money £69.00 +. I will keep checking to see if one becomes available over the coming months, I have also asked some of my photographer friends if they have a copy in their libraries but sadly not.
I did see another book by Bruce Fraser - Real World Image Sharpening, so I ordered it and will have to satisfy my thirst for knowledge by soaking up the pearls of wisdom in this for the time being.

You did say I could ask questions, I have just one.

I have an iMac and have just profiled it using a Datacolor Spyder3Pro so I should imagine my colours are fairly accurate. It was accurate before so Friday's re-profile was just a double check.
My printer uses a CIS system by a company called Fotospeed, who's paper I also use and my printer/paper has been profiled too.

Ok - I open an image in Photoshop, process it, convert it to AdobeRGB and it is ready to print.
In Photoshop I usually go to VIEW - PROOF SET UP - CUSTOM and enter my details:
Custom Proof Conditions = 'Custom'
Device To Simulate = PF_Lusture2880-4K (which is my paper profile by Fotospeed)
Preserve RGB Numbers = 'unchecked'

Now when I come to Rendering Intent it has been suggested that I use Relative Colorimetric but when I select that in the drop down box, tick Black Point Compensation everything still looks fine. Now when I select Simulate Paper Color in the Display Options (On Screen) my images changes and looks washed out.
This occurs with all the different Rendering Intent options - the image changes and becomes washed out. If I un-tick Black Point Compensation the image then darkens slightly but not to the same level as the original.

Do you know what is going on and why the image changes so much?
Usually the printed image comes out of my printer much darker than the original 'edited' image that is displayed in Photoshop before the RI lightening is applied under the Simulate Paper Color.

This has driven me mad for months but usually I just lighten the finished image in Photoshop before printing, so it looks way too washed out knowing that it will come out of the printer darker and more like the finished 'edited' image before I applied the additional lightening. I know this is not the ideal solution to the problem as I would like to achieve a print that is identical or very, very close to what I see on my screen.

Sadly when I have tried to resolve the issue before Fotospeed tell me my profiles are correct, as are the settings I am using to print my images, so I don't know how to proceed.

Regards

Chris

[Colin Southern]

Good Morning Colin

Well it is morning over here and a fine one at that, I expect the sun is over the yardarm with you and its time to settle down for the evening with a nice glass of red.

Yep - dark here. Prefer a glass of flat budget diet lemonaide, but I'm strange like that!

I have searched eBay and Amazon for the 2nd Edition Real World Colour Management and there aren't any copies available in the UK, apart from those at stupid money £69.00 +. I will keep checking to see if one becomes available over the coming months, I have also asked some of my photographer friends if they have a copy in their libraries but sadly not.
I did see another book by Bruce Fraser - Real World Image Sharpening, so I ordered it and will have to satisfy my thirst for knowledge by soaking up the pearls of wisdom in this for the time being.

How about any of these 2nd hand?

http://www.amazon.co.uk/s/ref=nb_sb_...al+world+color

You did say I could ask questions, I have just one.

I have an iMac and have just profiled it using a Datacolor Spyder3Pro so I should imagine my colours are fairly accurate. It was accurate before so Friday's re-profile was just a double check.
My printer uses a CIS system by a company called Fotospeed, who's paper I also use and my printer/paper has been profiled too.

Ok - I open an image in Photoshop, process it, convert it to AdobeRGB and it is ready to print.
In Photoshop I usually go to VIEW - PROOF SET UP - CUSTOM and enter my details:
Custom Proof Conditions = 'Custom'
Device To Simulate = PF_Lusture2880-4K (which is my paper profile by Fotospeed)
Preserve RGB Numbers = 'unchecked'

Now when I come to Rendering Intent it has been suggested that I use Relative Colorimetric but when I select that in the drop down box, tick Black Point Compensation everything still looks fine. Now when I select Simulate Paper Color in the Display Options (On Screen) my images changes and looks washed out.
This occurs with all the different Rendering Intent options - the image changes and becomes washed out. If I un-tick Black Point Compensation the image then darkens slightly but not to the same level as the original.

Do you know what is going on and why the image changes so much?
Usually the printed image comes out of my printer much darker than the original 'edited' image that is displayed in Photoshop before the RI lightening is applied under the Simulate Paper Color.

This has driven me mad for months but usually I just lighten the finished image in Photoshop before printing, so it looks way too washed out knowing that it will come out of the printer darker and more like the finished 'edited' image before I applied the additional lightening. I know this is not the ideal solution to the problem as I would like to achieve a print that is identical or very, very close to what I see on my screen.

Sadly when I have tried to resolve the issue before Fotospeed tell me my profiles are correct, as are the settings I am using to print my images, so I don't know how to proceed.

Regards

Chris

The short answer is "this is normal" (I get the same).

The long answer is: I really don't know why it's so far off, but I've just come to accept that it is. With the monitor only having a 6 stop dynamic range - and the paper only having around 4, I'm guessing that it's trying to simulate the loss of dynamic range to a degree. On top of that, you're also trying to simulate a device that uses subtractive colour on a device that only does additive colour, and the two just don't equate that well (in my opinion).

So personally, all I can suggest is limit the use of soft-proofing to the gamut check side of things (that's what I do) - other than that, I really don't use soft-proofing at all. Just a little thing to remember too - a printer profile is also null and void if you use different media settings (not just different media or ink) - so make sure you set it to the same paper characteristics as you used when you profiled it.

Hope this helps.

PS: Real World Image Sharpening is a great book too - just be sure to get the new edition co-written by Jeff Schewe, not the older editing written only by Bruce Fraser.

[CBImages]

Thanks Colin.
Yeah I looked at those 2nd hand books on Amazon but the cheaper ones are all copies of the 1st Edition. There is a Kinder version available of the 2nd edition and a copy in the USA for £43.58 + p&p, which I might end up going for if a used copy from the UK doesn't appear on eBay or Amazon in the next few weeks.

You got me worried about my new purchase so re-checked with my outstanding orders on Amazon - PHEW yes it is the 2nd Edition by Bruce Fraser and Jeff Schewe.

PHEW again - I'm glad I'm not going crazy and others get the same 'soft proof' issues.

Chris

[Colin Southern]

Bingo - I'd been waiting for it to appear on Kindle Thanks Chris (I'm now the proud owner of the electronic version too).

Why not just grab the Kindle version? - they're better and cheaper than a printed copy (that make 53 books on my PC / iPhone / iPad!).