From: Bruce Fellman, INTERNET:bruce.fellman@yale.edu
Date: Fri, Oct 12, 2001, 11:30 AM
RE: [colortheory] A skin color question
Dear Friends: I want to start with a personal aside -- in terms of the photography in the Yale Alumni Magazine where I'm managing editor, buying Dan's book and learning to apply his principles to our workflow has been one of the best investments I've ever made. Would that I'd known about it earlier; it would have saved us the green-faced embarrassment of having a too-green portrait of George W. Bush and Hillary Clinton (both spoke at Commencement) on our cover. Oh well, at least no one could accuse me of playing favorites; my ignorance (you just do a mode change from RGB to CMYK, right?) of how to handle and color correct digital photos and why one shouldn't res them up past their 7.5 meg size extended across the political spectrum.The lastest edition of our magazine bears the fruits of my labors with curves, plate-blending, sharpening in the L channel of LAB, etc, and we're rather pleased. Thanks to all on the list for a timely and helpful addition to this naif's education.
I did recently run into a quandary that I hope the group can address: In the various numbers for the proper rendering of skin tones, I noticed that the mix only extends towards people of the Caucasian persuasion. How do you handle the skin tones of non-whites? Is there a ballpark formula that offers target ranges for people of other races?
Thanks, Bruce Fellman, Managing Editor
The Yale Alumni Magazine
203-432-0650
PO Box 1905
149 York Street
New Haven, CT 06509-1905
From: Dan Margulis, INTERNET:76270.1033@compuserve.com
Date: Fri, Oct 12, 2001, 7:49 PM
RE: [colortheory] A skin color question
Bruce Fellman writes,>>In the various numbers for the proper rendering of skin tones, I noticed that the mix only extends towards people of the Caucasian persuasion. How do you handle the skin tones of non-whites? Is there a ballpark formula that offers target ranges for people of other races?>>
Persons of Hispanic or Asian ancestry tend to share approximately the same range, which is roughly the same as the dark half of the Caucasian population. Persons in these ethnic groups always have significantly more yellow than magenta, normally 10 to 15 points. Cyan plus black tends to be 1/4 to 1/3 of the magenta value, occasionally higher in the case of unusually dark skins.
The ethnicity loosely known as black or African-American has a much wider range of possibilities than any other. Cyan is usually at least 1/3 of the magenta value but there is no upper limit and there may be significant black. In the case of someone with light ("coffee-colored") skin, the yellow is significantly higher than the magenta. However, unlike other ethnicities, as the skintone gets darker, the variation between magenta and yellow decreases, so that in the case of a very dark-skinned person, the values would be almost equal.
Dan Margulis
From: "Maris V. Lidaka, Sr.", INTERNET:mlidaka@ameritech.net
Date: Fri, Oct 12, 2001, 10:39 PM
RE: Re: [colortheory] A skin color question
Thank you - I have been struggling with this myself. Please consider including this info in Chapter 2 of your next edition of Professional Photoshop - I think it would be valuable.Maris Lidaka Sr
From: "Mike McNamee Compuserve", INTERNET:mikemcnamee@compuserve.com
Date: Sat, Oct 13, 2001, 6:27 AM
RE: Re: [colortheory] A skin color question
Good question and good answers guys and girls.You may care to note that McCamy, Davidson and Marcus discussed the topic in their landmark paper to the Journal of Applied Photographic Engineering in 1976 and that the results of their deliberations are contained in swatches 1 and 2 of the Macbeth chart. Munsell also make a set of "ethnicity" swatches intended for anthropological research. If you go to the Gretagmacbeth web site you can pick up the details. If you then use the freely downloadable CMC software you can extract the pages that contain the skin colours. The problem is that (thank goodness) we come in all sorts of shades! My own preference is to use the Hue angle in the info palette, ranging roughly from 10 degrees for Caucasian gradually and seamlessly moving toward 35 degrees for oriental. Does anybody else use this measure and how good do they find it on press (which is what matters after all!)? Given a set of swatches you could, if you have one available, bang the swatches into your spectro? I have had some success creating a whole gamut profile then doing the fine tuning by tweaking the flesh colours, leaving others to drift a little.
Plenty to go at here for next edition of your book Dan!
Mike McNamee
From: INTERNET:amerphoto@ix.netcom.com, INTERNET:amerphoto@ix.netcom.com
Date: Mon, Oct 8, 2001, 12:52 PM
RE: [colortheory] Smooth Skin?
What's the best way to smooth skin on photos? I've seen *those* web sites where the girls skin looks real smooth. How do you do that? Is it blurring AB channels?Thanks
Hector Davila
From: INTERNET:varis@varis.com, INTERNET:varis@varis.com
Date: Mon, Oct 8, 2001, 4:47 PM
RE: Re: [colortheory] Smooth Skin?
APR wrote:
> What's the best way to smooth skin on photos?
> I've seen *those* web sites where the girls skin looks real smooth.
> How do you do that? Is it bluring AB channels?There are many ways to smooth out complexions but if there are real physical skin features (not just noise) that need to be smoothed then blurring the AB channels is not going to do it. One of my favorite techniques is to make a duplicate layer and run the median or dust and scratches filter on the new layer with enough of a radius to smooth out any defects. Option-click on the layer mask icon at the bottom of the layers pallet to create a black layer mask thus hiding the newly blurred layer completely. Now you can paint into the layer mask with white using a soft airbrush, smoothing just those areas that require it.
You'll probably need to add some noise to the blur layer to match the underlying layer - a small amount in the blur image should be enough to kill any banding artifacts from the median filter as well. If you have areas of skin highlights that approach white you might want to add the noise in an overlay layer. Create a new layer and enable the overlay apply mode. Fill this layer with 50% gray and group it with the underlying blur layer. Now run the noise filter on this gray layer. Noise is applied to the underlying smooth skin but ramps off as the values approach white and black so you get cleaner highlights and shadow areas.
-- regards,
Lee Varis
varis@varis.com
888-964-0024
www.varis.com
From: "John Opitz", INTERNET:jas10286@earthlink.net
Date: Mon, Oct 8, 2001, 4:46 PM
RE: [colortheory] Smooth skin?
<: Smooth Skin?
>>What's the best way to smooth skin on photos?
Is it blurring AB channels?>>
The Gaussian Blur Filter. Don't make selections to target skin areas(doing that is like chasing rainbows amd setting traps for unicorns). Do it to the whole image. Then take a snapshot of this state. Next undo gaussian blur. Use the history brush in lighten and darken modes(5-15% opac.) to paint back the areas you want to smooth. Make sure you click on the snapshot(history state) to paint back from.
John Opitz
From: Dan Margulis, INTERNET:76270.1033@compuserve.com
Date: Mon, Oct 8, 2001, 11:31 PM
RE: [colortheory] Smooth Skin?
Hector Davila writes,>>What's the best way to smooth skin on photos? I've seen *those* web sites where the girls skin looks real smooth. How do you do that? Is it bluring AB channels?>>
A method that I illustrated at Seybold is similar to the ones mentioned previously by Lee Varis and John Opitz, but I think is a little better in avoiding the overly airbrushed look.
1) Make four identical layers. Layer 1 remains untouched.
2) On Layer 2, Gaussian blur, Radius 2.0 (for starters, may adjust).
3) Having blurred Layer 2, apply a curve that lightens the midtone very slightly.
4) Change layering mode on Layer 2 to Darken. This will disallow most of the layer because of Step 3, and will only fill in small white blotches.
5) On Layer 3, apply Gaussian blur at double the radius of Step 2. (Reason: darker spots on the skin tend to be larger than white spots).
6) Having blurred Layer 3, apply a curve that darkens the midtone very slightly.
7) Change layering mode on Layer three to Lighten. This will disallow most of the layer because of Step 6, and will only correct age spots and other dark blemishes.
8) Compare Layer 3 to Layer 4 (an untouched original) and merge the two as desired with a layer mask.
This, incidentally, is how one "unsharpens" a picture, when a client has been so helpful as to supply something that's been oversharpened, without giving us access to the original.
Dan Margulis
From: INTERNET:varis@varis.com, INTERNET:varis@varis.com
Date: Mon, Oct 8, 2001, 11:53 PM
RE: Re: [colortheory] Smooth Skin?
Dan wrote:> A method that I illustrated at Seybold is similar to the ones mentioned
> previously by Lee Varis and John Opitz, but I think is a little better in
> avoiding the overly airbrushed look.Bravo, Dan- as usual, your method is particularly elegant. I think the girls on *those* web sites do look overly airbrushed and this approach gives a more realistic look (if that's what you're after :) - I know I'll be using this one more often, thanks!
> This, incidentally, is how one "unsharpens" a picture, when a client has > been so helpful as to supply something that's been oversharpened, without > giving us access to the original.
Something I've been wondering how to do for a while now and I can see how this would work with just the right tweaks to the curves in steps 3 and 6 - again thanks so much!
-- regards,
Lee Varis
varis@varis.com
888-964-0024
www.varis.com
From: "John Opitz", INTERNET:jas10286@earthlink.net
Date: Tue, Oct 9, 2001, 1:29 AM
RE: Re: Smooth Skin?
--- In colortheory@y..., Dan Margulis wrote:
> A method that I illustrated at Seybold is similar to the ones mentioned
> previously by Lee Varis and John Opitz, but I think is a little better in
> avoiding the overly airbrushed look.>Ect...Ect...
In reference to this. I'm going to use this method. I think this is a good time Mr. Margulis to write a book on your method you posted and others that you have on retouching alone. Because retouchers have a tendency to do that. Overly airbrush look. This will apeal to retouchers and photographers as well as people who like to restore pictures.I'm not talking here about run of the mill retouching. I'm talking about "other" methods(as you just stated) that target speed and accuracy of a stiletto(no pun intended). Example: I surfed this well known photographers web site who is an old time wedding photographer and he's just,I guess just starting to learn to use photoshop. And he was showing(and going to teach classes)how to transpose an image of a bride from one background to another. Well he was saying how he used the pen tool to do it with. Well I believe it took him about an hour to do it. Well I copied the image(just to prove a point) and just used one of the channels to make a mask. And it only took me less than 5 minutes to do it. That's what I'm saying. Your like E.F. Hutton(that old retro commercial). When you talk everybody listens.
John Opitz
From: INTERNET:amerphoto@ix.netcom.com, INTERNET:amerphoto@ix.netcom.com
Date: Tue, Oct 9, 2001, 2:14 PM
RE: Re: [colortheory] Smooth Skin?
Thanks everyone for the "Smooth Skin" techniques, I'm now an expert!Hector
From: Gordon Pritchard, INTERNET:gordon_pritchard@creoscitex.com
Date: Tue, Oct 9, 2001, 2:48 PM
RE: RE: [colortheory] Smooth Skin?
RE: Hector wrote:> Thanks everyone for the "Smooth SKin" techniques, I'm now an expert!
One item that I did not see covered in the thread is the impact of halftone screening on smooth tones. You may want to investigate round dot (round dots simply grow larger from highlight to shadow) or even FM screening (esp if presswork is from CTP). Halftone have an "optical bump" when the dots first touch in the tone scale. With Euclidean (round - square - round) this happens at the 50% dot. With elliptical it's split between 40 and 60%. Both screens have the bump in critical tone areas. Round dot places the optical bump at 75%. Well into the shadows where it is much less visible. Some printers also try to use exotic screening strategies to avoid rosettes, which can also appear as "noisy" tones in critical skin color areas.
thx, gordo
Gordon Pritchard
Commercial Print Specialist
CreoScitex
Vancouver Canada
T: 604.451.2700 ext 2870
C: 604.351.2437
gordon_pritchard@creoscitex.com
http://www.creoscitex.com
From: "M.Z. Beg", INTERNET:mzbeg@web-epoch.com
Date: Wed, Dec 5, 2001, 4:09 PM
RE: [colortheory] Smooth Skin?
Dear Sir,You wrote on October 9th :
"8) Compare Layer 3 to Layer 4 (an untouched original) and merge the two as desired with a layer mask."
Could you please be more elaborate for a novice like me, how do you do that, and what happens to Layer 2 and 1, do you merge the two before or after step
8.Many thanks.
From: Dan Margulis, INTERNET:76270.1033@compuserve.com
Date: Thu, Dec 6, 2001, 6:22 PM
RE: [colortheory] Smooth Skin?
M.Z. Beg writes,>>"8) Compare Layer 3 to Layer 4 (an untouched original) and merge the two as desired with a layer mask." Could you please be more elaborate for a novice like me, how do you do that, and what happens to Layer 2 and 1, do you merge the two before or after step 8.>>
Layer masks are the most powerful tool in professional retouching, so it would be well to review the manual as to how they work and some of their uses.
Briefly: as far as the bottom three layers go, you could merge them or not, it makes no immediate difference. The layer mask is applied to the top layer by choosing Layer: Add Layer Mask while the layer is active. The choice "Reveal All" gives a white layer mask and "Hide All" gives a black layer mask.
In areas where the layer mask is white one will see the top layer. In areas where it is black one sees the bottom layer(s). In areas where it is gray one sees a combination of the two layers. The lighter the gray, the more the top layer is favored and the darker the gray the more the bottom layer is favored.
In this case we would have uncorrected skin on the top layer and smoother skin on the bottom layer. Some people would create an actual dark mask for the face in the layer mask channel so as to expose the smoother skin on the bottom. Others would be inclined to start with a white layer mask, and then use the airbrush tool set to paint black at a low pressure to paint into it. The more the face was airbrushed in this fashion in the layer mask channel the more of the smoother skin will show through.
Dan Margulis
From: INTERNET:amerphoto@ix.netcom.com, INTERNET:amerphoto@ix.netcom.com
Date: Fri, Dec 28, 2001, 2:38 PM
RE: Re: [colortheory] Smooth Skin?
What would be a faster and better method to accomplish the same end result as instructed on this web page?
http://w3.nai.net/~tomh/sa_howto_harli.htmThanks,
Hector Davila
From: INTERNET:varis@varis.com, INTERNET:varis@varis.com
Date: Fri, Dec 28, 2001, 5:19 PM
RE: Re: [colortheory] Smooth Skin?
Hector Davila wrote:> What would be a faster and better method to accomplish the same end
> result as instructed on this web page?
> http://w3.nai.net/~tomh/sa_howto_harli.htmCopy background layer. Run median filter (filter->noise->median) at about 3 pixel radius on this copy. Run gaussian blur at about .5 pixels. Create a black layer mask by holding down option and clicking on new mask icon at the bottom of the layer pallet (this will hide the blurred layer). Now using a white airbrush paint into the active layer mask you just created. Start with about 30% opacity and paint over areas of skin that you want to smooth out. Avoid painting into areas with detail you want to preserve. If you make a mistake just paint back into the mask with black. You might want to add noise to the blurred image to match film grain- just make sure you select the image layer before you run the noise filter. After you are done you can adjust the overall opacity of the blur layer to reveal a little more skin texture if desired.
There are more complicated techniques but this should suffice for pin up girlie pictures where you don't care if they look too perfect. You can get real smooth skin this way. By using the median filter you can blur without smearing major edges and avoid the "haloing" referred to in the article in your link. The slight gaussian blur is sometimes necessary to smooth over banding that can be a by product of the median filter. You may need different radius settings in the filters depending on the resolution/size of the file.
Have fun!
Regards,
Lee Varis
varis@varis.com
www.varis.com
888-964-0024
From: INTERNET:samarsh@ozemail.com.au, INTERNET:samarsh@ozemail.com.au
Date: Fri, Dec 28, 2001, 11:07 PM
RE: [colortheory] Re: Smooth Skin?
Hector writes on the on/off topic of skin smoothing, which is similar in concept to smoothing skies or reducing grain or overly sharpened images etc:> What would be a faster and better method to accomplish the same end
> result as instructed on this web page?
> http://w3.nai.net/~tomh/sa_howto_harli.htmHector, as Lee suggests, median and gauss blur are two filters which spring to mind (on a duped layer, alter opacity or blend). As to the layer masking/blending - this is where things get more complex. Some use an 'edgemask' technique to help mask/reveal key edge detail or image content, and the K plate of a UCR sep can often be handy for masking facial detail - although this is adding more complexity than 'painting' the correction using layer masks or history states/brush etc.
As an addition to median/gauss blur - also experiment with the smart blur filter. I have ignored this filter for some time, and when I rediscovered it, it's advantages became apparent. The APS help system does a good job of explaining the unintuitive features. It provides an effect similar to posterization - with variable control and inbuilt edge protection. It may not do the job by ifself (take note of gradations etc), but with lower amounts and fades combined with median/gauss blur - it might prove useful.
It might also be handy to play with duping the filtered/masked correction layer so that you have one in lighten and one in darken layer blend mode. Now you have split control of the application of the filtering, opacity, layer option blend if masking etc.
Hope this helps,
Stephen Marsh.
From: "jas10286", INTERNET:jas10286@earthlink.net
Date: Sat, Dec 29, 2001, 10:45 AM
RE: [colortheory] Re: Smooth Skin?
--- In colortheory@y..., Hector Davila wrote:
> What would be a faster and better method to accomplish the same end
> result as instructed on this web page?
> http://w3.nai.net/~tomh/sa_howto_harli.htmAs Mr. Varis and Mr. Marsh has stated, this might be of help to you also for layer masks.
http://ep.pennnet.com/Articles/Article_Display.cfm?
Section=Articles&Subsection=Display&ARTICLE_ID=130520
As well as the book "Professional Photoshop 6". Explains(the advantage) working with images in CMYK, instead of just RGB. The advantage of the black plate for images(like the one at your website)."In color correction the key is the K"..... Explains the use of channels for making masks(instead of selections). "Every file has ten channels".....From "Professional Photoshop 6".John Opitz
Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.