The first time a client told me my retouching looked “plastic,” I had no idea what she meant. I thought I’d done a beautiful job. The skin was smooth, the blemishes were gone, the whole image had this polished magazine quality I’d been chasing. She pulled up a reference image on her phone, slid it across the table, and said, “I want to look like that. Yours looks like a wax figure.”

She was right. I had been smoothing out everything, texture and tone together, as if skin were a painted wall instead of a living surface. What she wanted, and what I didn’t yet know how to deliver, was smooth skin that still looked like skin. That gap in my understanding sent me down a months-long rabbit hole, and frequency separation is what I found at the bottom of it.

What Frequency Separation Is Actually Doing to Your Image

The core idea is simple even if the execution takes some practice. Every photograph contains information at different “frequencies.” Low frequency information is your broad tonal shifts: the gentle shadow under a cheekbone, the warm flush across a nose, the overall color and lighting of the face. High frequency information is your fine detail: pores, hair strands, the micro-texture that makes skin look real.

When most beginners retouch skin, they’re working on both layers at once. They clone-stamp a blemish and accidentally blur the surrounding texture. They use Healing Brush and end up with that telltale smeared, waxy look. Frequency separation lets us split those two pieces of information onto separate layers so we can address tone and color without touching texture, and refine texture without disturbing the underlying light and shadow. The skin stays dimensional. It stays human.

The Exact Setup I Use on Every File

I run Photoshop, and I’ve built this into an action I call “The Godfather” because it does everything quietly and without fanfare.

Here’s the manual version. Duplicate your background layer twice. Name the bottom copy “Low Frequency” and the top copy “High Frequency.” On the Low Frequency layer, go to Filter, Blur, Gaussian Blur. For most portraits shot around 24-50mm with decent resolution (I’m usually working on files between 25 and 50 megapixels from a Sony A7R IV or similar), I start at a radius of 8-12 pixels. You want to blur until the texture completely disappears but the tonal shapes of the face are still intact.

Now the important part. Select the High Frequency layer. Go to Image, Apply Image. Set the layer to your Low Frequency layer, blending mode to Subtract, scale to 2, offset to 128. Click OK. Then change the blend mode of the High Frequency layer itself to Linear Light. You now have two separated layers. The Low Frequency layer holds all the color and tone. The High Frequency layer holds all the texture as a neutral gray field with detail floating in it.

The Subtract method (rather than the older Add method some tutorials still teach) preserves your high frequency detail more accurately, especially in shadows. It’s worth relearning if you picked up the Add workflow years ago.

Working Each Layer Without Wrecking the Other

On the Low Frequency layer, I use a large, very soft brush (hardness at 0%, flow around 3-6%) to paint with sampled skin tones, or I use the Mixer Brush tool to blend uneven color. This is where I fix redness, correct uneven foundation application, or even out the light when the photographer had inconsistent strobe output. Nothing I do here touches the texture, so the pores and fine lines remain completely undisturbed.

On the High Frequency layer, I use the Clone Stamp tool with hardness around 15% and opacity at 100% to remove specific blemishes, hairs, or textural anomalies. Because I’m only moving texture, not tone, the healed area inherits the correct color from the Low Frequency layer automatically. No more dark patches around a removed blemish. No more brightness halos from a careless healing stroke.

One thing I always do: add a Curves adjustment layer clipped to the Low Frequency layer and bring it up just slightly in the midtones, around 10-15 points at the center of the curve. Frequency separation can make skin look a tiny bit flat after you’ve smoothed the tonal transitions. That small lift brings the luminosity back.

The Mistake I See Even Experienced Retouchers Make

I teach workshops out of my studio here in Portland, and the most common error I see has nothing to do with the setup. It’s over-working the Low Frequency layer. Students get excited that they can blend tones freely and end up erasing every shadow, every subtle variation that gives a face its three-dimensionality. The natural variation in skin tone is not a flaw. A cheekbone looks like a cheekbone because the skin there catches light differently. When we paint that away, we’re back to the wax figure problem, just via a more sophisticated route.

My rule: if you can no longer tell where the natural light is coming from when you look at the Low Frequency layer in isolation, you’ve gone too far. Toggle it visible and invisible as a reality check. The shadows should still be readable. The face should still have structure.

Knowing When the Separation Radius Is Wrong

If your Gaussian Blur radius is too small, skin detail will bleed into the Low Frequency layer and you’ll see strange artifacts when you blend tones. If it’s too large, you lose accuracy on the High Frequency layer and fine corrections become muddy. The 8-12 pixel range works for high-resolution files, but for a 12-megapixel file (say, a social-media-focused shoot with a client’s iPhone or a lower-res mirrorless), I’ll drop to 4-6 pixels. When in doubt, zoom to 100% and check that the Low Frequency layer is completely textureless before you proceed.

Frequency separation is not a shortcut and it’s not magic. It’s a framework that gives you surgical precision, but your eye and your judgment still have to drive it. Learn to read what each layer is telling you, and you’ll start producing work that looks genuinely professional rather than merely processed.