PDA

View Full Version : Increasing Saturation



Kirk Gittings
6-Sep-2006, 08:33
I know this is a huge subject and I really just want an overview if possible. Could someone discuss the advantages/problems with increasing saturation by various approaches, RGB vs. LAB and using Joseph Holmes' methods


I use it because I also have Joe's Chroma Variant profiles (both Chrome 100 and Ekta Space) and use them instead of normal saturation methods. The variants are assigned profiles (rather than converted to) and you can change them anytime in the workflow without and file degradation (since the don't affect the file itself).Doug Dolde

Thanks

Gordon Moat
6-Sep-2006, 09:52
Advantage: saturation of some tones can be increased without affecting other tones.

Disadvantage: colours that might pop on a monitor might be unprintable on some printing systems. Knowledge of using Pantone colours in a commercial press run might be necessary to achieve desired saturation appearance of certain tones. Not always as effective as increasing saturation while working in CMYK (that last one should cause an uproar).

The goal most people want with high saturation is to see that in the final prints. So the determining factor of what and how to adjust should be the limitations of that final printing method. It does no good to work up an image for really good paper and Pantone touch plates on an eight colour press, and then the next day sending the same file to be printed in your local newspaper without doing major changes to it. Anyway, if there was something more specific to your question, feel free to ask.

I get the feeling the majority here have some sort of Epson printer, and are just trying to get nice inkjet prints, with or without using a RIP to control inks. I do know that you, and a handful of others, want nicer commercially printed images. Those two goals do not always dictate the same workflows.

Ciao!

Gordon Moat
A G Studio (http://www.allgstudio.com)

Brian Ellis
6-Sep-2006, 10:33
I certainly can't answer the question but out of curiosity, apart from the potential gamut problems, what's wrong with using the hue/saturation slider assuming there isn't just one tone you wish to increase?

Kirk Gittings
6-Sep-2006, 10:45
Brian, I use those quite often Q&D on commercial images, but on my own work I am always interested in the most targeted, least destructive means of manipulating tone and color in an image. Some people have argued that working a curve in the A&B channels in lab is less destructive than using the hue/saturation slider in RGB. I would like to understand why better etc.

Greg Miller
6-Sep-2006, 11:06
I certainly can't answer the question but out of curiosity, apart from the potential gamut problems, what's wrong with using the hue/saturation slider assuming there isn't just one tone you wish to increase?

Increasing saturation using a hue/stauration layer can cause posterization as more and more tones get driven to a totally saturated tone. This is not really a problems with subtle use of this tool (i.e. increasing stauration +10%). An easy way to see the problem is to increase saturation to 100% - you see the destrcution of subtle tone shifts quite readily.

I personally rarely use hue/saturation - and usually I use it to decrease saturation slightly or to colorize. I think percevied lack of saturation is more often a lack of contrast. A little increase of contrast with a curve layer usually creates a more saturated looking image.

Another option is to duplicate the backgorund image and set the blending mode of the duplicate layer to "multiply". This creates a very strong effect of more density in the colors. Backing the opacity of that layer way down can create a more normal but saturated looking image.

Greg Miller
6-Sep-2006, 11:13
I really hate changing to LAB mode. I understand the power of LAB. But changing from RGB to LAB can cause a loss or shift of tones during the conversion. And changing again from LAB back to RGB can amplify the same problem. Additionally, you need to flatten your image when you make the conversion. So now you just lost all the poqwer of layers.

Personally I prefer to leave my image in RGB and add saturation with a normal curve layer.

chris jordan
6-Sep-2006, 11:26
Kirk, it's kind of like the zone system. Imagine you are working in B&W, and have an image with lots of detail in the zone 8 and 9 areas. If you darken the whole image by two zones, all of that detail gets turned into zone 10, so you will have areas of pure black where detail and tonal gradations have been lost. So ideally, instead of darkening the whole image equally, you would want to use a curves adjustment that leaves the zone 8 and 9 detail untouched, but darkens the zone 5 and 6, or whatever.

Saturation works the same way. In a given image there might be some high-saturation colors that are already be near the max that your printer will print (so you don't want to change those), and there are low saturation colors that you might want to "pop" by quite a bit. Unfortunately in Photoshop's standard RGB color tools there is no such thing as a saturation curve. So when you increase saturation using the standard tools, that increases saturation over the whole range equally, including the high-saturation colors that you don't want to touch. Photoshop lets you can divide it up by color at least, but if you saturate your greens for example, that will pop all of the greens, including the "zone 9" greens that are already very saturated and are at risk for being blown out. In one leaf, there might be 500 green pixels, all different shades of green, with 100 of those pixels already near the top of the green scale, and at risk for blowing out to pure green if saturation is increased. Popping the saturation of all of the greens equally will blow out those 100 pixels all to the same 0,255,0 value (pure blocked-up green, like pure black in the B&W analogy). So now instead of a nice tonally-complex leaf, you will have a leaf with some big green blobs. Elsewhere in the image, whole leaves might get blown that way (or areas of the sky if you're working with blue, etc.)

So there are a few things you can do to avoid blocking up your high-sat colors. In RGB mode you can make an adjustment layer that increases saturation (and put it in "color" mode), and then paint it only where you want it, carefully leaving out the high-sat areas. You can do this with either the Hue/Sat tool or the Selective Color tool. Sel Col is a bit more elegant and it creates fewer artifacts. This takes some time though, because in some cases you have to zoom in and work around the high-sat areas carefully.

Or you can use Joe's Chroma Variants, which are a more elegant way to do it because he has mapped the saturation increases so that the high-sat colors don't get blown. I rarely use them, however, because they are too broad-brush for my taste.

Or you can use LAB, which has a saturation curve tool. The a and b channels function as curves. The low saturation colors are at the middle, and the high saturation colors are at the top and bottom of the curve. So to increase saturation of only the low-sat colors, you can make an S-curve that will protect the high-sat colors. Using the eye-droppper tool, you can find the high-sat colors and lock them down, and then adjust the other colors. Green/magenta and yellow/blue are easy to do, and red/cyan are a bit more difficult because it takes two adjustments.

Depending on what I'm trying to accomplish, I do a combination of these things, but my most frequent method for increasing saturation (which I do pretty rarely; usually I'm reducing it) is to paint it only where I want it, on a Selective Color adjustment layer. Make sure the layer is in "color" mode so the brightness isn't affected, and you also can pull it out of the shadows and/or highlights using the advanced blending mode sliders.

Okay, more in person on Saturday...

Cheers,

~cj

chris jordan
6-Sep-2006, 11:28
Whoops, I got the zones backwards-- I was thinking of zones 8 and 9 as being dark, next to black. That will help with the above illustration...

Kirk Gittings
6-Sep-2006, 11:45
Thanks Chris,


I do a combination of these things, but my most frequent method for increasing saturation (which I do pretty rarely; usually I'm reducing it) is to paint it only where I want it, on a Selective Color adjustment layer. Make sure the layer is in "color" mode so the brightness isn't affected, and you also can pull it out of the shadows and/or highlights using the advanced blending mode sliders.

And this is still in RGB mode correct?


I really hate changing to LAB mode. I understand the power of LAB. But changing from RGB to LAB can cause a loss or shift of tones during the conversion. And changing again from LAB back to RGB can amplify the same problem. Additionally, you need to flatten your image when you make the conversion. So now you just lost all the poqwer of layers.

From another post it seems like you do not agree with this statement by Greg.

paulr
6-Sep-2006, 11:45
... but my most frequent method for increasing saturation (which I do pretty rarely; usually I'm reducing it)...

i was hoping someone would say that. i'm amazed by the universal quest for more saturation. even low contrast color neg films can look almost surreally saturated to me sometimes.

Kirk Gittings
6-Sep-2006, 11:53
Didn't Michael Smith comment "is the population growing colorblind?"

Greg Miller
6-Sep-2006, 13:41
From another post it seems like you do not agree with this statement by Greg.

To clarify my point, LAB any any RGB color space have different gamuts. And both are limited to a finite # of tones by the number of bits. Therefore during conversion from RGB to LAB, a specific tone in the source color space may not exist in LAB; that tone will be shifted to a nearby tone that LAB identifies (similar in some ways to converting from Prophoto RGB or Adobe RGB 1998 to Adobe sRGB; Adobe sRGB does not contain all the tones of the larger color spaces so those tones must be shifted if they exist in an image if it is converted to ADobe sRGB). When converting from LAB back to RGB, the same phenomenon can occur, and the tone can actually be shifted further away from the original tone. So tones can and do shift with every conversion.

dwhistance
6-Sep-2006, 14:33
I'm sure that I'm not as expert as the other people who have answered this post, however my understanding of the problem in using the Hue/Saturation control in RGB is somewhat different. From my reading about this subject, as well as practical experience, the RGB Hue/Saturation control affects the luminosity of the tones adjusted as well as their saturation. (My understanding is that this is because of the way the RGB colour space is defined but it may be a quirk of Photoshop). In LAB, saturation and luminosity can be adjusted seperately, albeit that you have to manage the A and B channels to adjust the saturation, something I find less than intuitive. Using Joe Holmes Chroma Varients allows the saturation to be adjusted in RGB without affecting luminosity as it is the image profile (i.e. its mapping within the colour space) that is changed rather than individual pixels within the image.

Having said all of that I must confess that I still use the Hue/Saturation control for the usual small, and often local, changes requred during processing of an image. I have used LAB to assist in the separation of similar tones in rocks and very occasionally have used Joe Holmes Chroma Varients when I have wanted to make a major change to saturation (generally incremental desaturation of Velvia 50 in my case!)

If you are interested in using LAB colour then I suggest reading: "Photoshop Lab Color: The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace" by Dan Margulis. It is a surprisingly good read and one of the most useful Photoshop books I have read, although I suspect this all shows I should get out more!

DrPablo
6-Sep-2006, 15:17
The advantage of using LAB is that by steepening the A and B curves you are increasing contrast along the magenta-neutral-green and yellow-neutral-blue axes. When used in moderation you therefore get separation of subtle tonal gradations.

The saturation slider, does nothing more than drive all colors towards their fully saturated prototype (red, green, blue, cyan, magenta, yellow, black, white). So you're more likely to see posterization and loss of color gradations with saturation, especially with saturated colors near the edge of their gamut.

When shooting RAW images with a digital camera, using the saturation slider in the RAW converter is probably the safest. Otherwise, my preference is LAB but for subtle tweaks it probably makes no difference.

Kirk Gittings
6-Sep-2006, 15:22
dw,

If you are interested in using LAB colour then I suggest reading: "Photoshop Lab Color: The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace" by Dan Margulis. It is a surprisingly good read and one of the most useful Photoshop books I have read, although I suspect this all shows I should get out more!

I just ordered it yesterday. Thanks

Greg Miller
6-Sep-2006, 16:36
Option 1:
1) Convert image to LAB (and flatten in the process (Bad))
2) Create curve layer
3) Select A channel
4) create S curve
5) Select B channel
6) Create S curve
7) Convert image to RGB (and flatten in the process (Bad))
8) result = increase in saturation; luminosity not affected; more separation in mid tones; potential shifting of tones due to conversion from RGB to LAB to RGB; no layers left

Option 2:
1) Create curve layer
2) Change blending mode to "Saturation"
3) create S curve
4) result = increase in saturation; luminosity not affected; more separation in mid tones; no shifting of tones due to conversion of RGP to LAB to RGB; All your layers left to tweak or rework if desired

Option 2 yields same result with fewer steps, all your layers intact, and no shifting of tones.

Mike Boden
6-Sep-2006, 16:58
If you are interested in using LAB colour then I suggest reading: "Photoshop Lab Color: The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace" by Dan Margulis. It is a surprisingly good read and one of the most useful Photoshop books I have read, although I suspect this all shows I should get out more!

Excellent book. I highly recommend this as well.

Doug Dolde
6-Sep-2006, 17:51
As far as Joe Holmes profiles go it's best to get it straight from the horse's mouth:

http://www.josephholmes.com/news.html

Ron Marshall
6-Sep-2006, 18:25
If you are interested in using LAB colour then I suggest reading: "Photoshop Lab Color: The Canyon Conundrum and Other Adventures in the Most Powerful Colorspace" by Dan Margulis. It is a surprisingly good read and one of the most useful Photoshop books I have read, although I suspect this all shows I should get out more!

A very worthwhile book for color work.

Kirk Gittings
6-Sep-2006, 20:24
Greg, option one I was familiar with. Option two looks like it is a very useful "go to" kind of tool with allot of control. I will have play with it some more.

Kirk Gittings
6-Sep-2006, 20:28
Chris is getting ready for his two shows I'm sure. In another thread he said this about working in LAB space:


There's another reason to use Ektspace: It converts seamlessly back and forth between RGB and Lab (Joe designed it that way on purpose). There are some things that can be done to control color in Lab that can't be done in RGB, so I jump back and forth between the two a lot. With colorspaces other than Ektaspace, there can be some image degradation when making the switch, but with Ektaspace it's flawless.


and


Rich, there are color adjustments that can be made in LAB that cannot be made in RGB, so I make at least one jump from RGB to LAB and back in every image I work on; sometimes several back-and-forths. In Ektaspace the histogram stays flawless; Joe Homes designed Ektaspace specifically with this in mind. Just for kicks I once tried converting back and forth 50 times using a Photoshop action, and the the before-and-after images were virtually identical when viewed at 800%. The histogram changed slightly, but not in a way that affected the image at all.

Tim Lookingbill
7-Sep-2006, 04:40
Do saturation edits require it be applied in 16bit? And is the hue/saturation induced posterization indicated in previous posts still induced in this mode? IOW, is saturation edits considered an extreme edit move inducing loss of RGB levels?

I like the saturation curve on a layer set to color blend mode tip, but I can't do this on 16bit images in PS 7.

Good to know the minimal image degradation claim converting back and forhtJoeRGB>Lab>JoeRGB because I've been applying it on some difficult images and was a bit concerned.

Something else I find useful for editing saturation to avoid RGB clipping for output to the sRGB space is to set Custom Proof Setup to sRGB/Relative Intent and have the info palette set to Proof Colors. It will indicate any edit induced clipping converting directly to this space. I find it very useful.

MJSfoto1956
7-Sep-2006, 05:38
I really hate changing to LAB mode. I understand the power of LAB. But changing from RGB to LAB can cause a loss or shift of tones during the conversion. And changing again from LAB back to RGB can amplify the same problem. Additionally, you need to flatten your image when you make the conversion. So now you just lost all the poqwer of layers.

A key thing to understand with LAB is that the curve only has 50% of the tonality of a similar curve in RGB (i.e. the curve tool in RGB @ 24bit represents a full 8-bits of data (256 shades of gray), whereas the curve tool in LAB @ 24bit represents 4-bits of data PER COLOR (128 bits of gray) as there are two colors represented per channel in LAB --one on the left and one on the right.)

So the important thing to remember with LAB is this: *ONLY* do edits in 48bit mode and never in 24bit mode. Once you convert back to RGB you can safely convert it back to 24bit if desired.

DrPablo
7-Sep-2006, 09:39
I understand the power of LAB. But changing from RGB to LAB can cause a loss or shift of tones during the conversion. And changing again from LAB back to RGB can amplify the same problem. Additionally, you need to flatten your image when you make the conversion. So now you just lost all the poqwer of layers

First, you only have to flatten adjustment layers. But regular layers do not have to be flattened. For sharpening I routinely preserve layers when switching between color spaces.

Next, in Dan Margulis' book he addresses at length the idea that LAB-RGB conversions are lossy. He shows examples of an image he's converted back and forth something like 25 times, and there's no detectable loss of image information or quality. Certainly for one or two conversions you're fine (converting to CMYK and back is a completely different story). But if you're worried about it just use a large gamut RGB space like ProPhoto RGB, and you'll find that there is very extensive overlap between it and LAB.

Greg Miller
7-Sep-2006, 13:43
First, you only have to flatten adjustment layers. But regular layers do not have to be flattened. For sharpening I routinely preserve layers when switching between color spaces.

I don't know about you but flattening adjustment layers is a big deal for me. I avoid this at all cost because it makes it much more difficult or impossible to tweak the image later.


Next, in Dan Margulis' book he addresses at length the idea that LAB-RGB conversions are lossy. He shows examples of an image he's converted back and forth something like 25 times, and there's no detectable loss of image information or quality. Certainly for one or two conversions you're fine (converting to CMYK and back is a completely different story). But if you're worried about it just use a large gamut RGB space like ProPhoto RGB, and you'll find that there is very extensive overlap between it and LAB.

My comment was not about loss of information. It was about shifting of colors (changing information). In the RGB to LAB to RGB process ther are 2 points where colors can change.