A few years into my retouching career, a client sent me an email I still think about. She’d hired me to retouch headshots for her talent agency, and after I delivered the first round, she wrote back: “These look great, but… they don’t really look like people anymore.” She was being kind. What she meant was that I had smoothed every actress’s skin into something closer to a vinyl doll than a human face. I had erased pores, texture, the slight variations in tone that make skin look alive. I had been so focused on removing imperfections that I removed the person along with them.

That email sat in my inbox for two days before I could respond. Then I got to work figuring out where I’d gone wrong.

What “Realistic Skin” Actually Means to Photoshop

The core problem with most beginner retouching, and honestly with a lot of intermediate work too, is treating skin as one unified layer of information. It isn’t. Skin has two distinct components that behave completely differently under editing: texture and tone.

Texture is the high-frequency detail. That’s pores, fine lines, small hairs, the subtle surface variation that makes skin look three-dimensional and real. Tone is the low-frequency information. That’s the color, the shading, the gradients of light across a face. When we use a healing brush or a blur tool directly on a flattened image, we’re destroying both at once. We fix a red patch, and the pore pattern disappears with it. We smooth out a shadow, and suddenly the cheek looks like it was rendered in a video game.

Frequency separation works because it lets us edit those two layers independently. We can correct uneven skin tone on the low-frequency layer without touching the texture that lives above it. The result is skin that looks retouched in the best possible way: even, clean, and still recognizably human.

Setting Up Frequency Separation Without the Guesswork

I run this setup through a Photoshop action I named “Blade Runner” because it felt appropriately cinematic. Here’s the manual version so you understand what it’s actually doing.

Duplicate your base layer twice. Name the bottom copy “LF” (low frequency) and the top copy “HF” (high frequency). With the LF layer selected, run a Gaussian Blur. For beauty work at high resolution, typically a file around 50MB or larger shot at 100%, I use a radius between 4 and 6 pixels. For tighter crops or lower-res files, drop to 2 or 3 pixels. The goal is to blur just enough that individual pores disappear but facial features remain distinct.

Now select the HF layer. Go to Image, Apply Image. Set the layer to your LF layer, the blending mode to Subtract, the Scale to 2, and the Offset to 128. Click OK. Change the HF layer’s blend mode to Linear Light. You now have two layers that, combined, look exactly like your original, but can be edited separately.

Work on skin tone corrections exclusively on the LF layer using the Clone Stamp tool set to Current Layer, 10 to 15% opacity, with Aligned checked. Move color and tone across the skin slowly and in the direction of the natural light. Never touch the HF layer for tone work. When you need to address actual texture problems, like a single large blemish that has both a color component and a raised texture component, use the Healing Brush on the HF layer to address structure, and the LF layer to handle the lingering discoloration separately.

The Opacity Mistake Almost Everyone Makes

Here’s something that took me longer to learn than I’d like to admit: frequency separation does not replace knowing when to stop. I’ve seen retouchers run a perfect technical setup and still deliver plastic-looking skin because they over-corrected on the LF layer. Just because you can smooth out every tonal variation doesn’t mean you should.

A useful rule: after you’ve done your LF corrections, reduce the opacity of that entire layer to between 70 and 85%. Not 100%. This lets a ghost of the original tonal variation breathe back into the skin. It’s a small adjustment that reads as a large one. The difference between 100% and 78% opacity on a LF correction layer is often the difference between “edited” and “retouched by a professional.”

When Frequency Separation Isn’t the Right Tool

Frequency separation is not a universal solution. For acne, deep scarring, or anything with significant three-dimensional structure, you still need to begin with the Healing Brush or Clone Stamp on a merged layer before you separate frequencies at all. Trying to fix a raised blemish purely within a frequency separation setup often creates smearing artifacts on the HF layer that are more distracting than the original problem.

I also don’t use frequency separation for every portrait. For editorial work where texture and character are part of the image’s point, I might do nothing more than a few targeted healing corrections and some dodging and burning. Frequency separation is a precision instrument, not a default workflow. Reaching for it out of habit is how you end up with technically correct images that feel emotionally empty.

The Single Rule I Give Every Student in My Workshops

My one standing instruction for anyone learning this technique is to zoom out to 25% before you export. At 100% view, we make decisions based on pixel-level detail that no viewer will ever see. At 25%, you see what a client sees, what a social media follower sees, what a casting director sees when they glance at a contact sheet. If the skin looks smooth but alive at 25%, you’re done. If it looks smooth but wrong, something in the texture layer got away from you, and it’s worth going back.

The goal of skin retouching is not a perfect surface. It’s a convincing one.