The first time a client pushed back on my hair retouching, I was three years into freelancing and thought I was doing pretty well. She sent me a screenshot with a red circle around the subject’s hairline and the words: “It looks like she’s wearing a helmet.” She was right. I had smoothed the flyaways so aggressively and cloned the scalp so carelessly that the entire silhouette looked vacuum-sealed. The hair had lost every quality that makes hair look like hair: variation, transparency, the way individual strands catch light differently from one another.

That experience sent me down a months-long rabbit hole that completely changed how I approach every strand on a subject’s head, and it’s what I want to walk you through today.

What’s Actually Happening When Hair Goes Wrong

Hair is one of the most technically demanding elements in a portrait because it exists in two different visual registers at once. There’s the overall shape and color of the hair as a mass, and then there’s the fine-detail layer: individual strands, flyaways, the texture of the cuticle under studio light. When we retouch without keeping those two registers separate, we collapse them into one flat surface. That’s where the helmet effect comes from.

On top of that, hair sits against backgrounds with complex edges. A strand of hair is often partially transparent, blending with whatever is behind it. When we paint over those transitions too bluntly or clone-stamp across them, we destroy the optical information that tells the viewer’s brain “this is hair.” The brain is extremely good at noticing when that information is missing, even if the viewer can’t articulate what’s wrong.

The Two-Layer Foundation Before You Touch a Brush

Before any corrective work begins, I set up every hair retouch the same way. I duplicate the background layer twice. The bottom copy gets labeled “Color-Tone” and the top copy gets labeled “Detail.” These are the two frequency separation layers I’ll be working on, and I name the action that generates them “The Shining” because it’s a classic and because I’ve been naming actions after movies for so long at this point that I can’t stop.

For the blur radius on the low-frequency layer, I use a Gaussian Blur set to between 6 and 10 pixels for most beauty work shot at full resolution (typically 24-50 megapixels). For fine hair or editorial close-ups, I drop to 4 pixels. Apply Image on the high-frequency layer uses the standard formula: set the blending mode to Subtract, Scale to 2, Offset to 128. This gives you a neutral gray layer carrying only edge and texture information, completely separated from the underlying color and tone.

Working on the Color-Tone layer, I use a large, low-opacity brush (15-25% opacity, 200-400px depending on the area) to blend hot spots in the scalp or uneven color banding through sections of hair. I never touch the silhouette edge on this layer. I never touch flyaways. Those live on the Detail layer and they need to be handled with completely different logic.

Flyaways: The Part Everyone Gets Wrong

Most retouchers either remove all flyaways or leave all of them, and both choices are usually wrong. Flyaways that cross a clean background should generally stay unless they’re distractingly chaotic. Flyaways that cross the subject’s face or eyes almost always need to go. The question isn’t “is this a flyaway?” but “what is this flyaway doing to the composition?”

When I do remove them, I work on a blank layer above the stack with the Clone Stamp tool set to Sample: All Layers. This is non-destructive and lets me throw away any stroke that doesn’t work. Opacity at 80-90%, hardness at 0%. The key is cloning parallel to the direction of nearby hair, not perpendicular to the flyaway itself. You’re mimicking the existing flow, not painting over it.

For the Detail layer, healing is almost always preferable to cloning when smoothing texture inconsistencies at the scalp or correcting where hair sections have separated to reveal scalp in an unflattering way. The Healing Brush samples texture and blends tone simultaneously, which keeps the result looking continuous rather than patched.

When the Background Fights Back

Shooting on white or near-white? Lucky you, hair masking is your whole problem and luminosity masks will mostly save you. Shooting on a gradient or textured background? Now we’re dealing with something harder, and this is where I’ve seen experienced retouchers give up and over-smooth.

The answer is to resist the urge to refine the mask beyond what’s necessary for the final output size. A hair edge that reads as crisp and natural at 1200px wide for Instagram does not need to be perfect at 100% zoom in Photoshop. I make this mistake less often now than I used to, but it took a while to trust that the output context should drive the retouching decisions, not the other way around.

For stubborn background halos around fine hair, I use the Refine Edge brush inside Select and Mask with a Radius between 3 and 8 pixels, then output to a Layer Mask with Smart Radius checked. From there, small corrections on the mask with a low-opacity black brush (30-40%) at the problem edges are usually enough. I do not try to get this perfect in one pass.

The Texture Test That Catches Every Mistake

Before I deliver any hair retouch, I zoom to 100% and scroll through the entire hair area slowly while asking one question: does every section still look like it has texture? Not the same texture everywhere, not perfectly smooth, but some surface variation that reads as real hair responding to light.

If any section looks like a brushstroke or a gradient or a shape with no internal complexity, I’ve gone too far somewhere, and I need to find it before the client does.

The goal of hair retouching is never to make hair look perfect. It’s to make it look exactly like the best version of what was already there.