A few years back I got a portrait job where the client handed me 200 RAW files from a corporate headshot session. Harsh fluorescent lighting, mixed skin tones, every subject photographed in under two minutes. The brief was simple: clean, consistent, professional. The deadline was 48 hours.

Before I knew frequency separation properly, that job would have wrecked me. I would have been painting and blending and clone stamping until my wrist gave out, and I still would have delivered soft, waxy skin that looked more like a video game character than a human being. Instead, I finished the whole batch in one day. Frequency separation is why.

What Your Image Actually Contains (Two Separate Things)

Every photograph holds two distinct types of information layered on top of each other. The first is tone and color, the broad shapes of light and shadow that define form. The second is texture, the fine surface detail like pores, hair, fabric weave, and grain.

When you retouch skin by painting or blurring directly on a pixel layer, you are hitting both of those things at once. Smooth out a blotchy patch and you flatten the pores underneath it. Clone stamp a blemish and you smear the surrounding texture. Frequency separation solves this by physically splitting those two types of information onto separate layers so you can edit one without touching the other.

The “frequency” in the name comes from audio and signal processing. Low frequencies are the broad, slow changes, your tones and colors. High frequencies are the rapid, fine changes, your texture and detail. Photoshop lets you separate them using a Gaussian Blur and a calculation that subtracts it back out.

Setting Up the Layers: The Exact Numbers That Work

Open your portrait and duplicate the background layer twice. Name the bottom copy “Low Frequency” and the top copy “High Frequency.”

Select the Low Frequency layer. Go to Filter, Blur, Gaussian Blur. For a portrait shot at typical resolution, somewhere between 200 and 300 PPI, I use a radius of 4 to 6 pixels. The goal is to blur just enough that skin texture disappears but color and tone shapes are still readable. You will see the exact threshold when you hit it.

Now select the High Frequency layer. Go to Image, Apply Image. In the dialog, set Layer to your Low Frequency layer, set Blending to Subtract, Scale to 2, and Offset to 128. Click OK. Then change the blend mode of the High Frequency layer to Linear Light.

What you are looking at now is a gray layer full of texture with no color information, sitting on top of a blurry layer full of color with no texture. They combine through Linear Light to reconstruct the original image perfectly. Nothing is destroyed. Everything is just sorted.

Working the Low Frequency Layer Without Wrecking Texture

Click on the Low Frequency layer. This is where you fix blotchiness, redness, uneven color, and tonal inconsistency. None of those fixes will touch a single pore on your subject.

I use the Lasso tool with a Feather of 20 to 40 pixels depending on image size, select the problem area loosely, and then run Filter, Blur, Gaussian Blur at 10 to 15 pixels. That blends the color smoothly into surrounding skin without smearing any surface detail. You can also use a soft brush set to around 30 percent opacity and paint directly with a sampled color. Either approach works. The key is that the High Frequency layer above is protecting all your texture the entire time.

For broader, uneven skin tones I sometimes drop a Curves adjustment layer clipped to the Low Frequency layer. That lets me push warmth or cool a highlight without it looking painted.

Working the High Frequency Layer Without Smearing Texture

Click on the High Frequency layer. This is where you remove specific blemishes, stray hairs, and hard-edged surface problems. Because this layer is gray and contains only texture, the Clone Stamp and Healing Brush behave very differently here than on a normal layer.

Set the Clone Stamp to Current Layer only. Sample from nearby clean skin texture and stamp directly over the blemish. What you are cloning is pure texture, pattern, with no color attached. The result is invisible in a way that is almost impossible to achieve by working on a merged layer.

One honest warning: be careful not to over-clone large areas of the High Frequency layer. I did that once on a swimwear campaign and the model’s shoulder looked like a repeating tile from a video game. The client noticed before I did. Sample from multiple nearby points and vary your source constantly.

The Mistake Most People Make When They First Try This

When I first taught frequency separation to a group of students in a live workshop, almost everyone set their Gaussian Blur radius too high on the Low Frequency layer. They would push it to 12 or 15 pixels, thinking more blur equals more separation. What actually happens is the color bleeds too far into neighboring areas, and when the layers recombine, you get a halo effect around high-contrast edges like eyelashes and hairlines.

Start at 4 pixels. Zoom in at 100 percent. If you can still see clear skin texture on the Low Frequency layer, add one pixel at a time until you cannot. That point, and not a pixel beyond it, is your correct radius. Write it down for each camera and lighting setup you work with regularly. After a while you will have a short list of go-to values and setup will take less than 90 seconds.

Frequency separation does not make retouching effortless. It makes the right kind of work possible. Once the layers are split, every edit you make is intentional, isolated, and reversible, and that precision is the difference between a retouch that holds up at full resolution and one that falls apart the moment anyone zooms in.