Hue and saturation changes (again)

[e2b]

I'm new to this forum, but it didn't take long to see that the issue of changes in hue and saturation (as when using tone Curves with Photoshop) is a topic with a long history here. Following various threads led me to an article by Martin Evening comparing hue and tone shifts with Lightroom and Photoshop (http://lightroom-news.com/2007/10/04/lightroom-versus-photoshop-curves/). He points out that these reflect deliberate (but different) program design choices made because, according to Thomas Knoll, “mild saturation boost produces more pleasing, film-like results”.

I appreciate that Lightroom and Photoshop are designed to be useful tools to help make attractive images easily, and that the user has to learn work within the constraints imposed by the developers. However, considering the mathematics involved, it would seem easy enough to make a tone adjustment tool, similar to Photoshop Curves, that changed luminosity without changing either hue or saturation. It would also appear easy enough to make a similar tool to change saturation as a function of luminosity without changing either luminosity or hue. In combination, these two tools would give the user complete flexibility when adjusting tone.

Is this just theory getting in the way of practice, or could there be any practical need for such tools? Are there any programs that provide such tools already? Are they in Photoshop already, but not immediately obvious?

[Mark D Segal]

Are you familiar with *Blend Modes* in Photoshop? If not I recommend you read-up on them. Then select any photo, add a Curves Adjustment Layer. If you wish to affect brightness and contrast without affecting colour, set the Curves blend mode to *Luminosity*. If you want to affect colour but not luminance, set the Blend Mode to *Color*. Etc.

[gromit]

Is this just theory getting in the way of practice, or could there be any practical need for such tools? Are there any programs that provide such tools already? Are they in Photoshop already, but not immediately obvious?

Luminosity blend mode (by itself) doesn't achieve what you're asking, the luminosity component has to be extracted first. Check out Lobster:

http://www.freegamma.com/

[Mark D Segal]

The OP asked about "practical need". But each to his own. Lobster is an option.

[e2b]

Thanks Mark and gromit. Regarding "practical need", I suppose the changes I am making to tone curves are a bit stronger than those normally needed in digital photography- my need has arisen mainly in restoration work I have been doing on very old and faded color slides

I generally use luminosity blending mode, but as gromit says, it doesn't do quite what I am after. However, I hadn't used the other modes so, to clarify my understanding, I opened a synthetic IT8 target in Photoshop and played around with the blending modes on a Curves adjustment layer. Just for the record, this is what I found:

Saturation blending mode: H - no change; S - changed (least effect with mid-tones); B/L - a slight change (least at mid-tones)
Luminosity blending mode: H - no change; S - strong change over full range (greater than with S mode); B/L - very strong change (of course!)
Color bending mode: H - slight changes; S - weak changes across range; B/L - weak changes (least for mid-tones)
Hue blending mode: H - very slight changes; S - no changes; B/L  no changes

A couple of further thoughts:

1. I found it interesting that Martin Evening said he preferred the Lightroom effect over the effect of using Luminosity mode in Photoshop (which he recommends in preference to Normal mode). In that article he refers to the tendency of the Photoshop luminosity blending mode to "tendency to suck some of the saturation out of an image".

2. I also played around with the "Image Apply" operation, with white and black source images and using the complementary add/subtract and multiply/color-dodge modes to get independent changes to H, S and B/L, but this is rather complicated and I can't see any way of to make changes as a function of luminosity.

It looks as if "Lobster" might be an option - I'll investigate further.

[Mark D Segal]

If the main thing you're trying to achieve is restoration of faded colour from film media, there are a couple of other things you may wish to consider:

(1) SilverFast Ai has an "Adaptiove Color Restoration" (ACR) tool in its Selective Color Adjustment panel which does a pretty decent job of this at the scan stage. The one thing you want to be mindful of is not inducing excessive correction at the scan stage, because once the image is scanned the result is "baked-in", unless you also happen to use SilverFast HDR, which allows you to selectively change scan edits without rescanning the image.

(2) Make a duplicate of your image file, convert it to L*a*b*, open a curves adjustment layer, and gently shift the end-points of the a* and b* curves inward, making sure that they pass through the center point of the curves graph all the time. This will bring out latent colour effectively - if you're not careful, too effectively. You can also alter the slope of these curves to introduce corrections of colour casts. When finished, flatten it, convert it to RGB, re-open the original RGB version of the image and add the L*a*b* treated version as a layer onto the original image. That way you don't lose the integrity of your original image file, and you can play with the properties of the added L*a*b*-treated version.

I did at one time look over Lobster - interesting to see what it can do, but I haven't revisited it for quite a while, so I don't know its latest configuration.

I guess for me, and other people may have differing preferences of course, I like to maintain the backward compatibility, convenience and flexibility of an integrated workflow as much as possible, so I try to make the best use I can of the tools in Lightroom and Photoshop, as well as plugins that play nicely with these "foundation" applications in respect of these criteria. When it comes to scanning, as one must start with a scanning application, I also like to use what it offers that is safe and effective.

[DerW]

OOC, why use Lobster (which costs money as far as I've seen), instead of simply recreating the action on your own?

Personally I think that 99$ is a little too much money for something as easy as this.

Anyways I attached you an action that'll do merely the same (I don't own Lobster, but the layer stack and results look the same and it's 0% difference to the original).

Edit: Ah, small problems still, that's always the risk of doing an action too quickly.
Works on most images though.
I'll post an updated version once the issues are fixed.
It won't work on areas of >90% saturation, otherwise it's 100% accurate. I wonder if Lobster has a workaround?

Edit #2: It seems like Lobster uses a pretty similar method to mine (also lossy if saturation >90%). The loss is probably due to the maths behind the "Luminosity" mode that clips certain channels.
I could upload a second action that'll revert all the areas that are different from the original, but that'll also mean that you can't work on them and therefore it's pretty easy to create banding.
Since most pictures don't contain areas of >90% saturation, I'd stay with the original version :-).

[gromit]

OOC, why use Lobster (which costs money as far as I've seen), instead of simply recreating the action on your own?

Personally I think that 99$ is a little too much money for something as easy as this.

The real power of Lobster is not its ability to edit luminosity independently (the extraction of which is easily duplicated) but an image's chromaticity (not so trivial). Being able to target colour on a saturation axis (enabling, among other things, similar functionality to Mark's Lab a/b edit above but more elegantly) gives enormous control. Most sadly don't make it this far as it requires a complete rethink of how you edit colour.

[Mark D Segal]

The real power of Lobster is not its ability to edit luminosity independently (the extraction of which is easily duplicated) but an image's chromaticity (not so trivial). Being able to target colour on a saturation axis (enabling, among other things, similar functionality to Mark's Lab a/b edit above but more elegantly) gives enormous control. Most sadly don't make it this far as it requires a complete rethink of how you edit colour.

I mentioned the L*a*b* Curves edit as an option; frankly even that I hardly ever find necessary; but if needed, it's within Photoshop and therefore easily integrated into a Photoshop workflow. I really hardly ever come across the need for saturation edits that can't be handled very satisfactorily using the bespoke tools with RGB colour spaces in Lightroom and Photoshop. But I'm curious about your expression "more elegantly" - could you unpack this - tell us what you mean by it?

[gromit]

But I'm curious about your expression "more elegantly" - could you unpack this - tell us what you mean by it?

You do know that steepening the a/b curves is the same as bumping the saturation (with Hue/Saturation) in Lab mode? It's more elegant with Lobster as you stay in the same space throughout and get to retain the layer stack. Pretty well all the nifty things you can do with Lab mode can be done with Lobster. I haven't used Lab in years.

Other than running Lobster to do the luminosity/chromaticity separation, the environment is standard Photoshop and uses standard Photoshop tools so I don't understand your point about workflow.

Most people have ideological or conceptual issues that get in the way of grasping Lobster's power and approach. Or they get bogged down in an intellectual exercise of trying to duplicate its functionality. Clearly it isn't for everyone.

[Mark D Segal]

You do know that steepening the a/b curves is the same as bumping the saturation (with Hue/Saturation) in Lab mode? It's more elegant with Lobster as you stay in the same space throughout and get to retain the layer stack. Pretty well all the nifty things you can do with Lab mode can be done with Lobster. I haven't used Lab in years.

Other than running Lobster to do the luminosity/chromaticity separation, the environment is standard Photoshop and uses standard Photoshop tools so I don't understand your point about workflow.

Most people have ideological or conceptual issues that get in the way of grasping Lobster's power and approach. Or they get bogged down in an intellectual exercise of trying to duplicate its functionality. Clearly it isn't for everyone.

My point about workflow is that the file size gets multiplied by 3 just by virtue of using Lobster, and then will increase further with other edits that need to be made. They recommend using it on a duplicate image file which, if it had layers, gets flattened before this application is applied. It works for its own adjustments in a Lobster file, but exactly how well this file integrates with all the other functions in Photoshop I haven't tested. If you have done that and are satisfied that all Photoshop functions can work perfectly well in a Lobster file, that's good. Re your point on L*a*b, I resort to it very very infrequently as I said - in fact these days pretty well never, because I'm satisfied with the standard contemporary RGB tools in Lightroom and Photoshop and really haven't bumped into the need for either Lobster or Lab - but that's just me - I accept of course that other people have different perceptions and needs, possibly making sense of using these approaches.

As for the ideological or conceptual issues you say are in most peoples' minds, I don't know who you are or how you know that, but just for clarity - speaking for myself I could care less about ideology in digital imaging. Processes either work efficiently and produce good results in the hands of capable users, or they don't, and as far as I'm concerned that's all there is to it. I just look for the easiest and most efficient way of producing the output I require.

To conclude my side of the dialogue on Lobster - one of these days if I really feel incentivized and have the time to download a demo and give it a thorough tryout, I'd do that. Until then I'm not positioned to say any more about it. In this thread I'm simply pointing out other viable ways in which the OP's requirements *may* be met.

[e2b]

DerW, Mark, gromit, thanks for sharing your various views about Lobster (and DerW's PS action). I have been on a steep learning curve over the past couple of days coming to grips with the range of approaches that can be taken with color manipulations, and while I can't say that I am on top of it all, I have now come read enough to know that they tend to generate very lively discussions!

I had a look at the FreeGamma site and the explanatory documents, but I was left wondering if it is still active - I didn't see references to CS5, for example.

[gromit]

I had a look at the FreeGamma site and the explanatory documents, but I was left wondering if it is still active - I didn't see references to CS5, for example.

It may not say so, but the latest version works fine with CS5.

[e2b]

Mark, Thanks for sharing your thoughts on colour restoration:

If the main thing you're trying to achieve is restoration of faded colour from film media, there are a couple of other things you may wish to consider: ......

I appreciate your remarks about SilverFast, even though I am a "rusted-on" Vuescan user (since 2003). Your comments also prompted me to reconsider how my scanning workflow has changed over the years. I have always made a 16 bit RBGI Raw image for archiving, but years ago I used to do as much processing as I could within Vuescan, output a Tiff file and finish processing in PhotoShop. Since Lightroom now dominates my digital workflow, I now only use Vuescan to re-process the (archived) Raw file to perform IR cleaning, and generate a cleaned Raw dng file, which I take into Lightroom.

I reviewed a few of my old efforts today and confirmed that some rather difficult slides that had taken me at least 20 minutes to fix with Vuescan/PhotoShop workflow can now be done in less than 2 minutes. I'd reckon that this is quite satisfactory 95% of the time. It's only the really faded slides that need the further treatment that led to this post.