A Workflow with Beta RGB

[Robert Ardill]

I apologize in advance if this topic has already been discussed, as it no doubt has.

I would prefer to use Beta RGB as a working space than ProPhoto or Adobe RGB because ProPhoto is just too big IMO (colors not visible to the eye, more likely to have quantization errors than in a smaller working space) and Adobe RGB is just a bit small (so that there is the possibility of clipping colors that are printable, and there is little elbow-room). Unfortunately Lightroom doesn't allow the selection of the Lightroom working space nor does it allow opening the image into Photoshop in anything except sRGB, AdobeRGB or ProPhoto RGB.  

So I have to work in Lightroom's version of ProPhoto, open into Photoshop as ProPhoto ... and then convert to Beta RGB: this is my normal workflow.  As long as there are no colors that are outside the Beta RGB space there is no damage to the image.

One thing I didn't realize, and which is very nice I think, is that if the image is opened as a Smart Object and immediately converted into Beta RGB (or the workspace of choice), then editing the Smart Object in ACR is constrained by the Photoshop working space, not the ACR working space.  So if I edit the Smart Object from the Beta RGB workspace, all the edits in ACR are constrained by Beta RGB.

Since I always use a Lightroom -> Photoshop -> Lightroom -> Print/Web workflow (in other words, all my images go through Photoshop), I can delay (some of) the raw processing until I am in Photoshop, doing it in ACR rather than Lightroom, all safely in Beta RGB.

Of course there is the disadvantage of much bigger file sizes because of the embedded Smart Object.  

It's a workaround that I think will work for me for most of my images ... but clearly this wouldn't be of interest to anyone who primarily works within Lightroom.

I would be interested in your feedback on this workflow and especially if you see reasons why this is not a good way to do things.

Robert

[kirkt]

In the CC version of ACR (v8.5) you can convert into any color space for which your computer has a profile.  For some reason LR5 does not permit this.

kirk

[jwlimages]

I have been using Beta RGB as my working space for years, quite successfully, for all my color work from digital captures. I decided on it for pretty similar reasons as you. (for monochrome or B&W output, 16-bit Beta RGB is obviously overkill, so those are 8-bit sRGB)

However, I rarely use Edit in Photoshop straight out of LR, because as you note, the choice of working spaces is limited (and yes, ACR takes a different route here). Instead, I Export > Tiff or .psd, which allows exporting directly to Beta RGB, 16-bit. Anything that needs a visit to Photoshop gets this treatment.

My workflow is probably a bit different, due to the fact that it started before LR appeared, so I have a substantial body of legacy .psd's & .tif's, printed through either Photoshop or RIP software. This influenced me to eventually develop two LR catalogs - one is exclusively raw capture files, the other is "prints", the images that merited further work in PS, and were finalized for output (along with legacy files).

I must admit to being a bit reluctant to go the smart object route - I use them during editing, but rasterize at conclusion, wanting to avoid any possible nasty surprise years hence…

my 2 cents, YMMV

John

JWL Images
Emeryville

[Schewe]

I would prefer to use Beta RGB as a working space than ProPhoto or Adobe RGB because ProPhoto is just too big IMO (colors not visible to the eye, more likely to have quantization errors than in a smaller working space) and Adobe RGB is just a bit small (so that there is the possibility of clipping colors that are printable, and there is little elbow-room).

So, you know for a fact that ProPhoto RGB is "too big"? You've actually seen "quantization errors" using ProPhoto RGB? Or is what you "believe" based upon your internet reading?

I ask because I've never seen any quantization errors when working in a ProPhoto RGB color managed workflow...

If you were on Bruce Lindbloom's web site reading about it, you should note that page was last updated on Sun, 20 Apr 2003 09:06:14 GMT which was about the time that Adobe released Camera Raw v1. There is a reason why Thomas Knoll chose ProPhoto RGB and linear gamma as the internal working space for ACR and then LR.

[Robert Ardill]

I must admit to being a bit reluctant to go the smart object route - I use them during editing, but rasterize at conclusion, wanting to avoid any possible nasty surprise years hence…

Hmm - interesting point, which ties in to CC and backward compatibility.  I can see myself having to do a batch flatten to tiff when I can no longer afford CC .

Robert

[Robert Ardill]

So, you know for a fact that ProPhoto RGB is "too big"? You've actually seen "quantization errors" using ProPhoto RGB? Or is what you "believe" based upon your internet reading?

I ask because I've never seen any quantization errors when working in a ProPhoto RGB color managed workflow...

If you were on Bruce Lindbloom's web site reading about it, you should note that page was last updated on Sun, 20 Apr 2003 09:06:14 GMT which was about the time that Adobe released Camera Raw v1. There is a reason why Thomas Knoll chose ProPhoto RGB and linear gamma as the internal working space for ACR and then LR.

Well Jeff, as it happens I'm an engineer with post-graduate degrees in computer science, so I don't have to entirely rely on the internet to inform me of these things, fortunately

To be honest, I'm not really concerned about quantization errors with ProPhoto (although it doesn't take advanced maths to figure out that a resolution of 1 in 65535 is better than 1 in 55,000, so the smaller your working space the better use you make of the bits available to you and the more you manipulate the image, the truer this is).

What is of concern to me is that it is too easy in ProPhoto to end up with colors that are completely out of gamut for viewing or printing purposes.  To have to turn on soft-proofing with gamut warning during editing is a nuisance: for me it's better to use a working-space that is more limited.  Of course, even though Beta RGB is smaller than ProPhoto, it's still possible to end up with OOG colors for the destination, so even though it's better than ProPhoto from an OOG point-of-view, it isn't perfect.

I've no doubt Thomas Knoll had good reasons for choosing ProPhoto with a gamma of 1.0 for Lightroom.  But he could have allowed export to Photoshop using any icc profile (as does ACR, presumably with Thomas' blessing), and he could have implemented a user-selectable gamut constraint within Lightroom.  What I mean by that is that it would be very useful to be able to select, say, Adobe RGB as the limiting gamut in Lightroom: what goes on under the hood isn't my concern ... but as a user I want to be able to work safely in a more constrained setting. I very much doubt I'm the only person who feels that way.

But I think you missed the point of my post: there is a work-around to this problem, thanks no doubt to Thomas Knoll: and that is to open the raw image as a smart object into Photoshop, convert to whatever working-space takes your fancy, and then perfectly safely edit the raw object in ACR

Robert

[Robert Ardill]

In the CC version of ACR (v8.5) you can convert into any color space for which your computer has a profile.  For some reason LR5 does not permit this.

kirk

Very good point ... it would be nice if Adobe could keep Lightroom and ACR in sync!  It beats me why ACR isn't just the develop module of Lightroom ... but maybe Adobe likes to duplicate work.

BTW ... another excellent improvement with ACR 8.5 as a result of allowing any profile, is the ability to soft-proof to any working space or to any destination profile (while editing a raw smart object).  Great! 

Now, if they could add a button Limit To Profile Gamut I would be in heaven.

Robert

[digitaldog]

But he could have allowed export to Photoshop using any icc profile (as does ACR, presumably with Thomas' blessing), and he could have implemented a user-selectable gamut constraint within Lightroom.

You can select any profile using the Export command but not Edit In (which might be nice).

[Schewe]

What is of concern to me is that it is too easy in ProPhoto to end up with colors that are completely out of gamut for viewing or printing purposes.  To have to turn on soft-proofing with gamut warning during editing is a nuisance: for me it's better to use a working-space that is more limited.  Of course, even though Beta RGB is smaller than ProPhoto, it's still possible to end up with OOG colors for the destination, so even though it's better than ProPhoto from an OOG point-of-view, it isn't perfect.

I think you are far too enamored with OOG issues...any digital camera will be capturing colors that can not be seen on prints and displays. What's important is to use soft proofing to see what the colors will look like if they are OOG and do something about it.

So, you admit that Beta RGB is still too wide for displays and prints and it's just as easy to drive colors OOG as it is in ProPhoto RGB? So, you are still clinging to the ProPhoto RGB is too large and inefficient why? Is your argument theoretical or practical? Can you show how ProPhoto RGB is too big on a practical basis? I don't care about the math...I care about images. Show me where an image is better with Beta RGB vs ProPhoto RGB.

[Robert Ardill]

I think you are far too enamored with OOG issues...any digital camera will be capturing colors that can not be seen on prints and displays. What's important is to use soft proofing to see what the colors will look like if they are OOG and do something about it.

So, you admit that Beta RGB is still too wide for displays and prints and it's just as easy to drive colors OOG as it is in ProPhoto RGB? So, you are still clinging to the ProPhoto RGB is too large and inefficient why? Is your argument theoretical or practical? Can you show how ProPhoto RGB is too big on a practical basis? I don't care about the math...I care about images. Show me where an image is better with Beta RGB vs ProPhoto RGB.

Well Jeff,

I'm glad you didn't ask me to do anything too complicated .

So, here's a simple example, and you can replicate it yourself easily.

To make it easier to see, take an image that is quite saturated (so it will have OOG colors for the destination).  Duplicate it. Then convert the original to the destination profile using Perceptual.  To make the test more obvious, use a destination profile that has a small gamut, like an Epson Enhanced Matte profile for example.  Then convert the copy to Beta RGB (it will be a Relative conversion of course).  Convert that to the Epson Enhanced Matte profile (using Perceptual).  Then compare the two images.

Here is an example for you, to save you the trouble:



The top image is the one that suffered a direct ProPhoto to destination hit.  The bottom one is the one that went through Beta RGB.  The reason why the ProPhoto to Destination image looks so bad (OK, I admit, they both look bad! but the top one has lost most of its (very oversaturated) colors) is that the Perceptual shift squeezes the whole ProPhoto gamut into the Epson Enhanced Matte gamut, resulting in very desaturated colors.  The same damage cannot be done with Beta RGB because it's much smaller than ProPhoto (not to say that damage isn't possible, it's just likely to be less severe).

To be honest ... I think it would have been much better to have all gamuts limited to triangles, just as the working-spaces are, so we wouldn't be continuously trying to fit triangular pegs into oval-or-worse shaped holes.  Then I wouldn't have been offered the opportunity of  wasting people's time whingeing about what to do with OOG colors!

Robert

[digitaldog]

To be honest ... I think it would have been much better to have all gamuts limited to triangles, just as the working-spaces are, so we wouldn't be continuously trying to fit triangular pegs into oval-or-worse shaped holes.  

It's a problem but there isn't a thing we can do about it. Those simple triangular shapes are due to the color spaces being simple and theoretical. Output (and capture) color spaces are far different in shape and that mushing of shapes is just a fact of life. Toggle a couple rendering intents, pick the one that looks best, maybe do some minor tweaks (there's only so much you can do while soft proofing the destination color space). Move on.

[Tim Lookingbill]

I would be interested in your feedback on this workflow and especially if you see reasons why this is not a good way to do things.

Robert

For what reason are you interested?

So you can formulate arguing points toward folks already quite knowledgeable and experienced in both post processing and background history of the technology?

Why add more unnecessary workarounds to your processes? As time consuming post processing already is I don't think you're going to find a lot of adopters.

[Tim Lookingbill]

What we need here for the engineering minded to understand about digital imaging technology is the equivalent for someone to provide a perspective on its limitations, best practices and expectations as is demonstrated in the YouTube video below on the myths about digital audio in this instance dithering and bit depth (issues shared by digital imaging sensor recordings)...

https://www.youtube.com/watch?feature=player_detailpage&v=BYTlN6wjcvQ#t=2075

It's another hobby of mine that acts as a respite from digital imaging but find has similar online enthusiasts just as obsessive about the details and issues of the technology whether they exist or not in the course of creating content.

It helps to listen with headphones. Jump to the section about recording at certain bit depths. At what bit depth do you hear a difference.

[Vladimirovich]

I don't care about the math...

not only you, there is one Ph.D with "6 stops of DR advantage" 

[Schewe]

not only you, there is one Ph.D with "6 stops of DR advantage" 

Yeah, well I never agreed with him about that either...(and his Ph.d was in some sort of nuclear stuff not imaging science)

[fdisilvestro]

it doesn't take advanced maths to figure out that a resolution of 1 in 65535 is better than 1 in 55,000, so the smaller your working space the better use you make of the bits available to you and the more you manipulate the image, the truer this is

Tell me the noise level and the maximum signal to noise ratio (SNR) and I will tell you which one is better. Don't be fooled by the digits, otherwise the solution to improving quality would be just to add bits or digits.

BTW, Photoshop uses signed integers so in practice you use 15 bits, not 16

[Simon Garrett]

I don't follow the OP's logic.

If you first open the image in Lightroom, then you've already converted to ProPhoto RGB.  Any quantisation errors (noise) resulting from this large colour space will already have occured.  Converting to Beta RGB can't undo those errors.  However, by doing another colour space converstion (to Beta RGB) simply introduces another small source of noise. 

As FranciscoDisilvestro hints, all this is quite likely to be well below the SNR in the image anyway (working at 16 bits encoding), so is of little consequence. 

But I can't see how converting to Beta RGB improves things. 

[Robert Ardill]

For what reason are you interested?

So you can formulate arguing points toward folks already quite knowledgeable and experienced in both post processing and background history of the technology?

Why add more unnecessary workarounds to your processes? As time consuming post processing already is I don't think you're going to find a lot of adopters.

Actually, the reason for my post was that I found, inadvertently, that when I edited a raw Smart Object, that ACR retained the color space from Photoshop.  I had assumed that it would not, and would use ProPhoto, so I was concerned by the possibility that my edits might throw the image OOG in relation to my working space.  I was really delighted to find that this was not the case, and I thought that I might share this with you ... as perhaps some people reading this post may not have realized this (as I did not until I stumbled upon it).

If many of you know this already, as I have no doubt you do, ... then what harm has been done? If you know a reason why doing what I am doing is a bad idea ... well then you can tell me; if you think it's a good idea, great!

I'm not in the least interested in arguing about this or anything else with you or 'folks knowledgeable and experienced in both post processing and background history of the technology'. My interest is to learn, and because my experience of color management and post processing is quite limited, there is a lot that I still have to learn ... and already this forum has helped me a lot, for which I'm very thankful.

If I am wasting your time then I'm truly sorry.

Robert

[Simon Garrett]

If I am wasting your time then I'm truly sorry.

You're not wasting our time, that's what we're all here for - to debate and learn.

My thoughts are that you might be wasting your time. 

If the image goes through Lightroom, it gets converted to ProPhoto RGB on the way.  This conversion - any conversion - introduces a small amount of quantisation noise arising from the quantisation levels in the destination colour space, and simply the action of changing from one colour space to another.  However, the image starts in 12 or 14 bits raw and already has other sources of noise probably at least as large as the 12 or 14 bit quantisation noise.  The quantisation noise introduced by conversion to 16-bit ProPhoto RGB, even though greater than that of Beta RGB, is likely to be significantly lower than the SNR already in the image data, and so is unlikely to be perceptually discernable. 

Converting then from ProPhoto RGB to Beta RGB is another conversion, introducing another processing step, and another (small) source of noise.  Again, not likely to be significant, but the conversion can't undo any (small) quantisation noise introduced by the conversion to ProPhoto RGB introduced by Lightroom. 

Further noise introduced by processing in Photoshop might be a bit lower if the processing is done in Beta RGB as compared to ProPhoto RGB, and I'd be interested to see any tests that show if this can be discerned in the image.  Personally I doubt it's significant, but I'm open to being shown otherwise. 

As others have said, I can't follow the logic of any benefit of Beta RGB with regard to colours out-of-gamut for real devices. 

[Robert Ardill]

I don't follow the OP's logic.

If you first open the image in Lightroom, then you've already converted to ProPhoto RGB.  Any quantisation errors (noise) resulting from this large colour space will already have occured.  Converting to Beta RGB can't undo those errors.  However, by doing another colour space converstion (to Beta RGB) simply introduces another small source of noise. 

As FranciscoDisilvestro hints, all this is quite likely to be well below the SNR in the image anyway (working at 16 bits encoding), so is of little consequence. 

But I can't see how converting to Beta RGB improves things. 

Hi Simon,

The conversion to Beta RGB or Adobe RGB or any smaller space isn't going to improve things in itself, of course (and as you say, any conversion will most likely introduce some small errors) so it would be a pointless thing to do unless there was some benefit down the line.

The question is, do you think, like Jeff Schewe, that it's totally fine to work in ProPhoto and that I am over-concerned about OOG colors? ... or do you think, as I do, that there is a real risk of damage to the image in the final conversion to the destination, and that the larger your working-space the greater the risk?  That's your call of course ... and what you decide certainly makes no difference to me.

As far as I'm concerned it's like this: experts like Jeff Schewe can work perfectly safely in ProPhoto because they know what they are doing and are aware of the possible problems and don't fall into the kinds of traps that I have fallen into (for example changing to Lab mode and doing bonker things without any understanding of the consequences).  I'm getting closer to that point as I get a better understanding of what is under the hood and what to watch out for ... but I'm certainly not there yet.  Not too long ago, it was like giving a Lamborghini to a kid on speed!

Robert