ProPhoto Colour Space?

[Mark D Segal]

Arising out of the threads on Silverfast and 8 versus 16 bit mode for image editing is the question of whether there is real merit to using ProPhoto as one's default colour workspace for image editing. Personally, I'm not interested in theory - only the differences observable in prints (without a loupe) using ProPhoto rather than Adobe RGB98 colour working space.

My question: is there anyone out there who:

(1) has capture devices that can "see" ProPhoto Colour Space;
(2) has printers that can print a gamut wider than Adobe RGB98 in any colours;
(3) has made comparison prints of images captured and processed in both RGB98 and Prophoto, and
(4) is therefore in a position to advise what additional benefits and/or issues are attributable to using ProPhoto rather than ARGB98?

If you are one of those people, it would be most interesting to learn of your observations.

[Andrew Larkin]

Mark, I can tell you the story of why I switched to ProPhoto.  It was a "red" day.  I had shot a number of images on my Canon D30 that included large amounts of red subject material - shops painted red, tricycles painted red. fire hydrants, you know the thing.  The images were taken in RAW and I was converting them to TIFF/AdobeRGB using a number of tools including Photoshop Camera Raw and BreezeBrowser.  What I was seeing consistently was a fading in the reds towards orange and try as I could with my photoshop skills at the time, I just couldn't drag the these colours back to where I wanted them.

Out of desparation, I reconverted some of the original files to using Camera Raw to ProPhoto and suddenly was looking at an image much closer to what I wanted.

In addition to this, I compared the histograms of the same image converted to AdobeRGB and ProPhoto and noticed that the AdobeRGB histogram was a subset of the ProPhoto chart: ACR was being forced to clip the image to fit the colour space.

The combination of the clipping and hue shift (likely to be a symptom of the clipping!) was limiting my editing of the image and was taking away creative options for the image from MY decisive control.

Your first two qualifying points:
"(1) has capture devices that can "see" ProPhoto Colour Space;
 (2) has printers that can print a gamut wider than Adobe RGB98 in any colours;"

are essentially flawed.  It is not a question of the size/volume of the gamut of the input and output devices.  It is a question of the overlap in the device gamuts.

You can conceivable have a narrow gamut device that produces colours outside of most standard colour spaces.  An example would be an infrared camera.

My Canon D30 and 10D bodies capture colours that are outside the AdobeRGB colour space.  That is *not* the same as saying that they have a larger gamut than AdobeRGB, just a different one.

The same applies to printers.

Your choice of colour working space needs to be either large enough to cover the gamut of your input device and output device so that you can make the decision on how the colours will translate, or else you need to be completely happy with the automatic conversions/compression/clipping that would be applied using a smaller colour space and specifying the rendering intent as your sole control over this conversion.

Andrew

[digitaldog]

Your first two qualifying points:
"(1) has capture devices that can "see" ProPhoto Colour Space;
 (2) has printers that can print a gamut wider than Adobe RGB98 in any colours;"

are essentially flawed.  It is not a question of the size/volume of the gamut of the input and output devices.  It is a question of the overlap in the device gamuts.

Exactly. No, there are no devices that capture or see ProPhoto RGB. It’s a synthetic color space; build solely on math. The same could be said of sRGB, Adobe RGB (1998) or any other synthetic RGB working space. ProPhoto RGB has the potential to hold colors you can capture that you can’t hold in a smaller space. It’s pretty simple actually. You can see this by simply viewing the histogram in Adobe Camera RAW as you toggle the working space options.

[Mark D Segal]

Yes, I agree - the way I expressed (1) and (2) can be misleading taken independently - what I was trying to hone in on is that it makes sense for one to have tested for Working Space as long as one's capture and printing devices can handle colors (not necessarily the same ones for each device) that are beyond the ARGB98 gamut. Andrew, your story about the red is most interesting because it is precisely intense reds that have been giving me the biggest headache in ProPhoto (hugely over-saturated to the extent the printer couldn't deal with them), but rescanning the negatives in ARGB98 went a long way to taming the red to something believable the printer (Epson 4000) could handle much more successfully - so the reverse of the solution you found.

[Jonathan Wienke]

But that's because your scanner software isn't doing proper color management, and Adobe RGB happens to be a more believable match for the output than ProPhoto. The problem has nothing to do with the color space, it's the stupid behavior of the software not performing proper color management.

[Mark D Segal]

OK, from the Silverfast thread we came to understand the limitations of the scanner software in respect of handling colour negative materials for the reasons explained there. But, still and well, for whatever reason that is a situation where a narrower colour space somewhat rescued the images. If you're saying this situation doesn't make a generic case for always using a narrower colour space, I agree. Come to think of it, I should seek out the same kind of "Chinatown red" tones here in Toronto, photograph them with my 1Ds, convert them 16 bit ProPhoto in Camera RAW, softproof them in PSCS2, print them and see what happens compared with the scanned negs. If I do it, I'll report back the findings.

[Jonathan Wienke]

I should seek out the same kind of "Chinatown red" tones here in Toronto, photograph them with my 1Ds, convert them 16 bit ProPhoto in Camera RAW, softproof them in PSCS2, print them and see what happens compared with the scanned negs. If I do it, I'll report back the findings.

Let me possibly save you some trouble. Here are some 1Ds RAWs shot at a car show with reds that fall well outside Adobe RGB:

RAW #1
RAW #2
RAW #3

With the default ACR conversion settings, all of them will clip badly if you set the color space to anything but ProPhoto. A few things worth noting:

1. Some of the cars have prominent blue sky reflections that make some red areas appear somewhat magenta. This is not a glitch, it is consistent with their appearance in real life.

2. The reds are probably beyound the bounds of your monitor gamut in places. Some areas will appear to be a flat color and devoid of detail as a result, even thouugh moving the cursor around in some of these areas shows unclipped and varying RGB values.

3. These RAW images are posted for personal educational and learning purposes only, and I am not giving up copyright or granting permission to anyone to use them for commercial purposes.

If you want to test the efficacy of using ProPhoto vs smaller-gamut color spaces for getting the most true-to-life prints possible when dealing with difficult colors, these RAWS are excellent choices.

[Mark D Segal]

Jonathan, thanks for doing that. So far I have failed to be able to open those images (don't know why - perhaps file size), but I understand what you are getting at, and yes, since you have had the experience, most likely no need to replicate it. You have indeed saved time and trouble.

[Jonathan Wienke]

Try right-clicking on the link, Save Target As...

[Andrew Larkin]

Andrew, your story about the red is most interesting because it is precisely intense reds that have been giving me the biggest headache in ProPhoto (hugely over-saturated to the extent the printer couldn't deal with them), but rescanning the negatives in ARGB98 went a long way to taming the red to something believable the printer (Epson 4000) could handle much more successfully - so the reverse of the solution you found.

The workflow I use is to keep the image in ProPhoto.

Then, setup your softproof settings for your printer and turn on Gamut warning.

Next, create a hue/saturation adjustment layer for the image.

Start by selecting a colour channel associated with your main problem colour.  You can adjust the sliders on the spectrum at the bottom to change the boundaries of the channel.

Then, adjust the saturation/lightness of the channel to bring the problem area under control.  You might also find a slight shift in hue might work.

By doing this with each of the problem channels, you can tame your colours into the gamut of your printer.

Andrew

[32BT]

I think that Bruce Fraser has hinted on several occasions that the colors you are looking for are to be found in the dark saturated colors. One of the reasons for the rather saturated primaries of Prophoto RGB is not so much saturated bright colors, but more so saturated dark ones.

I believe Mark is colorsavvy enough if I recall correctly, but even so: to actually "see" the difference between using Prophoto vs ARGB, you really need to make a final print and have your colormanagement set up correctly. Your monitor will just exhibit the same hue shifts using either colorspace. And if your printer profiles are remotely like the Adobe press profiles, you will find the same problems occur, even if you use softproofing. For more information on those types of hue shifts and the corresponding problems, you may find this link interesting reading.

[Ben Rubinstein]

I print my wedding/event work on a frontier whose output gamut seems to be slightly less than sRGB. I've worked in sRGB until now so that I don't get any suprises, all files need to be converted to sRGB and 8 bit for print on the frontier.

I tried two prints, one converted from ProPhoto one from sRGB both in 16 bit. They were then both converted to 8 bit and sRGB. The scene was of a bride standing in the middle of a flower garden, lot's of green/red/orange.

The one converted from ProPhoto looked anything but natural, the colours were way OTT and lurid. The sRGB one was similar to how I saw the scene and far more natural.

Therefore: For Me, with my workflow and output, ProPhoto does not work. For landscape work (printed on a Kodak LED printer on Ilford Hi-Gloss paper), I've not had the chance to try yet, getting a neutral print with that combo is hard enough.

[32BT]

I tried two prints, one converted from ProPhoto one from sRGB both in 16 bit. They were then both converted to 8 bit and sRGB. The scene was of a bride standing in the middle of a flower garden, lot's of green/red/orange.

If it works, it works. But do be careful. If you say that you convert *from* ProPhoto and *from* sRGB, what do you mean?

The problem is: you convert the RAW data TO either ProPhoto or sRGB. If you use ACR, then converting to ProPhoto will ensure that all data/colors are retained as ACR thinks they have been captured. If you convert directly to sRGB with ACR, then it will use channel clipping which may lead to undesirable results (as in unrecoverable colorshifts).

The workflow that should be used is the one described earlier:
Convert RAW to ProPhoto and use softproofing to tune the image to sRGB or to the Frontier profile if it is available.

[Mark D Segal]

Jonathan - yes, "SAVE AS" sometimes SAVES US when that happens, so I'll go back there and try again. Thanks.

Andrew, yes, been there-done that. It helps. Also, either alone or in combination, with soft-proofing on I have  added a Selective Colour adjustment layer, opened the Red channel, reduced the magenta till just before the reds just start to show more yellow than one wants and then boosted Black somewhat. All that said and done, these adjustments either alone or in combination were much more effective at taming and managing red for these particular images when implementing them on an ARGB98 scan rather than on the ProPhoto scan - I suspect because of the colour management software issue with negative materials Jonathan and I have been discussing.

opgr, thanks - yes, I am decently colour-managed. I've achieved generally reliable WYSIWYG between soft-proofing and my Epson 4000. So if I see garbage on the monitor I can be quite confident of getting similar garbage on paper. Helps save ink and paper.    Thanks for the referral to that article. Looks colourful enough, which is a-propos. I shall read it with interest. I agree with your advice to POM. That is why from the time I got tuned-in to the ProPhoto beat I have been going from RAW to ProPhoto - preserve as much data as possible. Worked well till I ran into these blazing (or as Andrew L suggests - dark) reds.

[bobrobert]

Then, adjust the saturation/lightness of the channel to bring the problem area under control.  You might also find a slight shift in hue might work.

By doing this with each of the problem channels, you can tame your colours into the gamut of your printer.

Could someone explain the difference between doing this and just letting the printer clip the colours? I read somewhere that it you may as well let the printer do it rather than spend time doing it yourself Am I being too simplistic in beleiving this? TIA

[Mark D Segal]

By doing what Andrew Larkin and I have been discussing here, we retain control over what will come out of the printer. If you just let the printer do it, you have no control over what is happening and therefore you end-up with the kind of mess that these adjustments are designed to mitigate.

[framah]

I have been reading this posting and have a question which to you all might seem dumb but... here goes.  Where do you find Prophoto RGB? I hadn't heard of it until this post and would like to check it out.  

 Thanks for the help.

[Mark D Segal]

ProPhoto is one of the colour working spaces provided with Photoshop since about 2003. I don't know what version of Photoshop you are using, but you will find it in the appropriate drop-down menus where you set your colour management preferences in Photoshop.

[PeterLange]

Andrew, your story about the red is most interesting because it is precisely intense reds that have been giving me the biggest headache in ProPhoto (hugely over-saturated to the extent the printer couldn't deal with them), but rescanning the negatives in ARGB98 went a long way to taming the red to something believable the printer (Epson 4000) could handle much more successfully - so the reverse of the solution you found.

The workflow I use is to keep the image in ProPhoto.

Then, setup your softproof settings for your printer and turn on Gamut warning.

Next, create a hue/saturation adjustment layer for the image.

Start by selecting a colour channel associated with your main problem colour.  You can adjust the sliders on the spectrum at the bottom to change the boundaries of the channel.

Then, adjust the saturation/lightness of the channel to bring the problem area under control.  You might also find a slight shift in hue might work.

By doing this with each of the problem channels, you can tame your colours into the gamut of your printer.

Andrew

Mark,

Of course I won’t say: “As if I haven’t said”.  Hope the sense does not get lost due to translation.  & Peace!

Just for background info, I recently took the opportunity to bring this issue to Bruce Fraser’s attention:
http://forums.robgalbraith.com/showfla....t358022


Now thinking about possible solutions, it’s easily obvious that a best practice for controlled de-saturation is required to move such an excess of out-of-gamut colors from ProPhotoRGB towards the printer space; more precisely into the expectance-sphere of the Lut.  In addition to Andrew_Larkin’s post, some further techniques are as follows:

1.)  Just convert to a slightly smaller space such as Bruce Lindbloom’s Beta-RGB prior print.

2.)  Reduce the contrast on both channels a & b in Lab mode (keep an eye on possible hue shifts).

3.)  A Hue&Sat-layer set to Saturation blend mode, combined with either a Color Range selection, or preferably with a selection of out-of-gamut colors.  Alternatively, instead of the Hue&Sat.-saturation slider the Channel Mixer can be used.

4.)  Assignment of a ProPhotoRGB Chroma Variant as introduced recently:
http://forums.robgalbraith.com/showfla....#358826

… and combinations thereof. In any case, SoftProof to the printer profile will contribute for prediction.


Hope this is of help!

Peter

P.S.: Sometimes theory is not so bad, too...
--

[framah]

Found it. Thanks. I have PSCS2 I ust didn't know about it nor where to look.