Frequency separation and it’s variants
Triple frequency separation (frequency separation into 3 frequencies) is one of the variants of classical frequency separation. Beginners are very fond of it because it seems more professional. I recommend try it, it may turn out that you will like the method. If you did not know how to do Basic Frequency Separation into two frequencies, then first start with it, and after understanding the principle of operation, learn to retouch with three frequencies.
For frequency separation into three frequencies, both 16-bit photos and 8-bit photos are laid out in the same way. BEFORE the frequency separation, all pixel cleaning is done (with the help of a healing brush, stamp), as during separation, one skin blemish will have to be removed not once, but already three times – from each layer. Therefore, we do not complicate our lives.
Frequency separation into 3 frequencies: Algorithm
Steps 1 to 4 consecutively
*Medium frequency AFTER high frequency
- + Layer blending and contrast (read below)
1 Gaussian blur
(a multiple of 3 value) e.g. 30
3 High Pass with the value for “Low” e.g. 30
4 Gaussian blur with a value for “High” e.g. 10
2 High Pass
(value = low frequency÷3) for example, 30÷3=10
Make three copies of the first layer. The bottom layer with the copy “Low”, the middle layer “Medium” and the upper “High”. Put the copies in a new folder and call it “Frequency Separation”.
Step 1: Select the “Low” layer and apply Gaussian blur (Filter – Blur – Gaussian Blur). Choose such a radius so that only the skin tone of the model remains in the photo and there are no irregularities and skin details visible. Select the blur radius so that the number without residue is divided by 3 (for example, 30) and remember it (we will need) as Y, for example, Y = 30.
Step 2: Select the “High” layer (top layer) and apply High Pass (Filter-Other-High Pass).
The value is conventionally called X, and 3 * X = Y.
In my case, if in Step 1 I have 3, then for the high frequency I choose 10 (because 30/3 = 10).
Simpler: the value for the high frequency should be three times less than the value for the low frequency
Steps 3 – 4: Go to “Medium”. We apply High Pass (Filter-Other-High Pass) with the value that we used for the “Low”, that is, Y. And then we apple Gaussian blur on the same layer (Filter – Blur – Gaussian Blur) with a value of X.
In other words:
I. On the “Medium” apply the High Pass with the value for the “Low”
II. On the “Medium” we apply the Gaussian blur with a value for “High”
In my example I choose 30 for High Pass and 10 for Gaussian blur.
Now you need to adjust these frequencies correctly by turning on the blending modes and reducing the contrast.
Layer blending modes and contrast
On the “High” and “Medium” we apply the Linear Light blending mode. We don’t touch the “Low “.
In the “High” and “Medium” layers we reduce the contrast by half. To do this, apply the Brightness/Contrast adjustment layer to the “High”. Put a value of -50% on the contrast (to do this duble click on the icon of this layer). Make a clipping mask on a layer. Let’s do the same with the “Medium”, simply duplicating the correction layer and putting it above the “Medium”, apply the clipping mask.
Note. To create a Clipping mask, place the Contrast layer above the layer you want. Click on the Contrast layer, and then click Alt (Windows) or Option (Mac) and move the cursor to the border between the layers, click on the arrow. When you do this, you see that the layer with contrast “nests” the bottom layer with the arrow icon. Second way: click right mouse button on the contrast correction layer, select “Create Clipping Mask”.
Let’s start with “Low”, achieving smoothness and beautiful tone. To do this, you can make a duplicate of the layer and work on it with the help of a mix brush, stamp, selection and blur, healing brush and patch. There are so many options that I can not recommend one thing. Someone prefers to select areas with a lasso and apply the Gaussian blur, others work on a layer with a mix brush. These are all standard Photoshop tools (on the toolbar) for this. There’s usually not a lot of retouching on the “Low”. The main work is carried out in the “Medium”.
On the “Medium”, we correct uneven skin, remove the “extra” texture.
Method: We create a mask on the layer, and, using a black brush on a white mask “erasing” imperfections. We literally erase the information from “Medium”, but on a mask, so that the information can be restored (using a white brush on the mask). To add a mask, click Create Mask at the bottom of the Layers panel (next to the Adjustment Layers button).
At “High”, we only correct the texture of the skin. The instruments are still the same familiar healing brush and stamp. From the “good areas” we take a sample and apply it to problem areas.
The main thing – you should not overretouch! Don’t try to work on the whole photo at once. Divide the image (face) into 6 segments and work on each segment in turn. In order to facilitate the task of working on the texture I create a correction layer (curves) and apply the contrast only on “High” with the Clipping mask.
The second approach: retouch from a position where the model’s face in the photo is no larger than your fist. Only after all visible problems have been removed, you can increase the scale of a photo.
A little trick: you can duplicate, for example, a “Low”, apply the Clipping Mask (Clipping Mask, putting a copy of the layer above the original), and work only on the copy mask layer to be able to “cancel” the changes. And if we work on “High”, then the copy of the layer and the clipping mask are no longer used with the “linear light” blending mode, but in the “normal” mode.
Conclusion (as well as caution)
On the one hand, yes, the frequency separation into 3 frequencies may seem simpler (compared to Dodge and burn), and the result is pleasing to the eye. On the other hand, the separation into three frequencies is something cumbersome and complicated.
Any method has pros and cons. So try it, as they say, one’s meat is another man’s poison.