Removing variation in skin, is this the best way?

[Hendrik]

What technique do you prefer to even-out variations in skin tones? Every skin has some mild variation in color and lightness across the face or other body part.

After I’m done dodging and burning on a curves layer, mostly at 200-400% to remove skin blemishes, I make a final step with the Degrunge method. I don’t use it to blur the skin, but I use much higher settings and only target the variation in skin tones.

On a 10mp facial shot for example, I can use Gaussian blur of 8 and high-pass of 24. This keeps the skin detail intact.

After this step, I sometimes use some dodging and burning to bring some highlights and shadows back.

Are there better methods?

[DJSoulglo]

If you're D&B-ing already, why not just zoom out a little and D&B with a larger brush?

Or what I like to do is just draw around the uneven area's with the lasso tool, give a good bit of feathering and use a curves layer.

[mayday]

Its not good to Degrunge if you want professional results.
You could use the curve method with a layer mask to paint into.

With the curve take a sample of the good skin and bad skin from each channel then insert these values into the input & output.
Once this is done carefully paint into the layer mask.
This should even-out the skin

[Hendrik]

I'm talking about subtle variation. I will post an example when I'm back home.

Why is degrunge method not good. I understand is't not a good method to smoothen the skin, but with a high GB and HP, I only target the differences in skin tone and lightness, without touching the details ....

... or am I missing something?

[mayday]

Any degrunge is destructive to pixels to some degree.
But maybe you find this method works ok for you.
It depends on the size and quality of the file, on some med to low res
images you can probably get away with this it but you might not on a 40-80mb file.

[Markzebra]

Maydays method is the right one Hendrik. Best way to do subtle things is using pressure sensitivity on your wacom, although not all people do it this way. Take your curves, further than they need to go and then lightly brush in the tones.

[Hendrik]

Thank you for the responses, better techniques are always very welcome.

I have attached an example with and without the degrunge, please have a look. (GB 8 HP 24)

The curves method is a better method as you said, please can you give a little more details about this technique?

sample: http://www.xs4all.nl/~honey/fotograf...s/degrunge.jpg

[mayday]

Maydays method is the right one Hendrik. Best way to do subtle things is using pressure sensitivity on your wacom, although not all people do it this way. Take your curves, further than they need to go and then lightly brush in the tones. Markzebra


Yes a curve works better that maintains the pore detail

[Hendrik]

Originally Posted by mayday

Yes the curve looks better maintains the pore detail

Is this a comment on my sample? Both images are the same, only the one with degrunge has this extra. I didn't use any curves yet.

I haven't tried curves yet, because I'm not sure how I should use it with these subtle skin variations. How do you use curves?

Originally Posted by mayday

It depends on the size and quality of the file, on some med to low res
images you can probably get away with this it but you might not on a 40-80mb file.

Why not, can I not simply use a much higher value for GB and HP?

[JavierT]

Originally Posted by mayday

Any degrunge is destructive to pixels to some degree.
But maybe you find this method works ok for you.
It depends on the size and quality of the file, on some med to low res
images you can probably get away with this it but you might not on a 40-80mb file.

Hi Mayday

Very Very interesting in that technique. If you can explain it a little bit more? Or Maybe you, MarKzebra?

Thanks in advance

[mayday]

[QUOTE.
With the curve take a sample of the good skin and bad skin from each channel then insert these values into the input & output.
Once this is done carefully paint into the layer mask.
This should even-out the skin[/quote]



This is the best I can do to explain really Javier

[Hendrik]

I don't think the Curves method is a good method in this case, and the reason is simple, ... there is no good or bad skin, only slight variation. The face has not one specific tone, so taking a tone from a 'good' area will NOT be a good tone at another place on the face. The degrunge method is not using one specific tone, but simply removes variation. The degrunge method how I use it is not damaging the finer details.

... btw, I just saw that Godmother ALSO uses the same method.

[mayday]

Like I said if your happy with what your doing and its giving you good results
then stick with it. Just giving you another method of how I would deal with it.

[Hendrik]

… and I appreciate that Mayday, don’t get me wrong. Discussing about a technique is the best way of learning, but unfortunately on this board not very popular.

I tried your technique, but experienced a few problems.

The setup is easy, just copy the RGB values of the correct skin into the output values of the ‘bad’ skin in a curves adjustment layer. Using a soft brush at low opacity and gently stroke the RGB values of the bad skin to the RGB values of the good skin.

Yes, this is a good method if you want to remove a localized color or lightness difference and I agree this is the only correct method. But what I want to do is not removing one specific blotch or blotches of a specific hue all over the skin. Skin has subtle variation and I want to equalize it a bit. Using the curves method will bring everything closer to the sampled skin, but this is not what you normally want. retaining the normal global variation is important to keep depth and structure of the skin. (see also the sample I posted earlier)

You say the degrunge (using high values) is destructive, maybe it is, so I hope someone can tell me why!

[other experienced retouchers are also welcome to react ]

[edgework]

Originally Posted by Hendrik

I don't think the Curves method is a good method in this case, and the reason is simple, ... there is no good or bad skin, only slight variation. The face has not one specific tone, so taking a tone from a 'good' area will NOT be a good tone at another place on the face. The degrunge method is not using one specific tone, but simply removes variation. The degrunge method how I use it is not damaging the finer details.

... btw, I just saw that Godmother ALSO uses the same method.

You're overlooking how curves work. They don't target a specific color; the samples are simply meant to be from a midrange that's represntative.. Curves distribute the shifts smoothly across the whole range so you can use the same curve wherever you see unevenness. The tendency is almost always to see the "problem" spots as darker, so the same curve that lightens a blotch in the quartertone region will lighten a different blotch in the three-quarter tone region. Obviously, the farther away you are from the initial samples, the greater the chance that you will have a drift in tone, but nothing as severe as simple dodge and burn moves, where you are forced to apply a single color solution to the entire range.

And you don't have to do it all in one step. After your first pass, any further unevenness (which will be considerably less than you started with) can be addressed by a second curve layer.

It's true that the degrunge technique and all it's variants will blend imperfections in tone. But it simply creates a look that is generally not considered acceptable in the commercial world. Like Mayday said, if you like it, go with it. But if you're submitting your work for professional evaluation, it won't advertise you well.