One way is to seperate color and texture onto seperate layers.

Frequency Separation Retouching action set

Based on Andrey Zhuravlev Retouching a portrait based on the frequency decomposition of images web page and his Example PSD file.

Frequency Separation Retouch action set has two action made from the information found in that write up. The actions just set up the layers need to do Frequency Separation Retouching. No retouch is done however Yellow coded layer for doing retouching have been added. Also included in the package is a PSD file containing Andrey Zhuravlev 16 Bit Adobe RGB Portrait which the High Low Freq Retouch action was played and the Yellow layers edited to duplicate Andrey Zhuravlev retouch.

Download Warning 300MB because of the example PSD file.

If you download this please read the included text file if you need to change the texture for you need to change the blending mode before wotking on texture.

I don't see a lot of damage to the subject's texture in your example, but what I DO see is a fair bit of image noise.

If you want to keep 100% of the noise and texture, and still want full control of what's being fixed up, try the Healing Brush.  If you'd like a little automation to help you keep from having to choose good skin areas as the source for the "Healing", try the Spot Healing Brush, which employs Content Aware Fill.

Frankly, for this particular image the Spot Healing Brush does a bang up job...

-Noel

Juan,

first of all you'll need some simple models for color spaces:

http://www.fho-emden.de/~hoffmann/hlscone03052001.pdf

Here we have RGB, the double cone HLS and the single cone HSV=HSB.

H = Hue = angle

S = Saturation = radius (one may talk about Chroma as well)

L = Lightness = vertical axis = V = Value = B = Brightness

HSV is used in Photoshop, but the explanation is easier for HLS.

Hue is obviously the 'tone', e.g. a certain green at 130°. In Spanish 'matiz'.

Once the Hue is fixed, we can still modify the Lightness and the Saturation.

Blending Mode Hue works like this: The Hue is taken from the source,

Lightness and Saturation are taken from the destination.

Blending Mode Color (Photoshop nomenclature) takes Hue and Saturation

from the source and Lightness from the destination (I think so). 

It's perhaps interesting that such a threedimensional model was already

found by Munsell about 1907-1912 empirically, without any mathematics:

http://www.fho-emden.de/~hoffmann/munsell15052009.pdf

Each color page in this doc shows two hues, which differ by 180°.

Value=Lightness is represented by the vertical axis. Saturation increases

with the radius.

The appearance is similar to the CIELab color space (Lab in Photoshop),

which is based on human vision and mathematics.

http://www.fho-emden.de/~hoffmann/cielab03022003.pdf   

It's obvious that the surfaces of the CIELab space and the hypothetically

smoothed Munsell space are not regular shapes like spheres, ellipsoids,

cones or double cones.

Of course there are much older 3D-models. 

Best regards --Gernot Hoffmann