A few years back, a client sent me a revision note that stopped me cold. She’d hired me to retouch headshots for her acting portfolio, and her feedback on the eye work was: “They look like doll eyes. Can you make me look like a person again?”

She was right. I’d gone in hard with a Hue/Saturation layer, cranked the blue channel on her irises, and added a specular highlight so bright it looked like a flashlight was aimed directly at her face. The eyes were technically sharp and colorful. They were also completely dead. That note is the reason I now work through eye enhancement in four distinct layers, each one doing a specific job, none of them doing too much.

Why Eyes Read Differently From Everything Else in a Frame

The human brain is wired to process eyes before almost anything else in a photograph. Neurologically, we’re scanning for threat or safety, recognition or connection. What this means practically is that any enhancement you apply to the eyes gets scrutinized at a sensitivity level that nothing else in the image receives. A slight color shift on skin might go unnoticed. The same shift applied to irises will read as “off” to almost every viewer, even ones who couldn’t articulate why.

The other technical reality is that eyes contain more tonal variation per square inch than nearly any other element in a portrait. You have the specular highlight, the iris, the pupil, the limbal ring (the dark outer edge of the iris), the sclera, the tear duct, the wet line of the lower lid, and the catchlight, all within a region that might be 300 pixels wide in a properly shot portrait. Each of those zones behaves differently. Treating the eye as a single object to enhance, rather than a collection of micro-zones to adjust individually, is where most retouchers run into trouble.

The Four Layers, and What Each One Actually Does

I work in Photoshop, and I run this sequence on every portrait that needs eye work, whether it’s a quick editorial job or a full beauty composite. I name my actions after movies, so this group is saved as “All About Eve.” Fitting.

Layer 1: Clarity and structure. Duplicate your background, convert to Smart Object, and run a High Pass filter at 1.5-2.5 pixels radius, depending on the resolution of your file. Set the blend mode to Soft Light at 40-60% opacity, then mask to the iris only. This recovers detail in the texture of the eye without introducing sharpening halos. At full 42-megapixel files from a Sony A7R, I land at 2.0 pixels. For web-only portraits shot on a phone, 1.5 is usually enough.

Layer 2: Limbal ring deepening. Create a new layer set to Multiply. Using a soft brush at 10-15% opacity, paint a deep brown or desaturated navy around the outer edge of the iris. This is the single most underused technique in eye work, and it does more to make eyes look vivid than any saturation adjustment. The limbal ring naturally fades with age and fatigue. Restoring it with 3-4 careful strokes reads as health, not editing.

Layer 3: Targeted dodge and burn. New layer, blend mode Overlay, filled with 50% gray. Use a dodge brush (Highlights, 7-9% exposure) to lift the 2-3 catchlights that the original light source created. Use a burn brush (Shadows, 5-7% exposure) to slightly deepen the upper lid shadow that falls across the top of the iris. This creates dimensional depth, the sense that the eye has volume rather than sitting flat on the face.

Layer 4: Color refinement. Hue/Saturation adjustment layer, clipped to the iris region via a precise selection. Increase saturation by no more than 8-12 points. For blue or green eyes, shift hue by 1-3 degrees toward cyan rather than toward the primary color. It reads as more luminous and less obviously pushed. For brown eyes, pull saturation on the Yellow channel down 4-6 points, which removes the muddiness that makes brown irises look flat.

The Mistake That Taught Me to Work Smaller

After that doll-eye feedback, I did something that felt counterintuitive: I cut every adjustment in half. Whatever number I thought was subtle, I halved it again. Then I printed a test shot at 8x10 and held it at arm’s length.

The work held up in a way it hadn’t before. At a distance, the eyes looked bright and clear without looking treated. Up close, you could see the texture of the iris, the natural variation in the limbal ring, the catchlight that matched the light in the rest of the image. That experience permanently recalibrated what “subtle” means to me. I now export a 50% zoom JPEG of just the eye region and look at it on my phone before I call anything finished. If it looks retouched on my phone screen, it’s not done yet.

What to Do When the Sclera Reads Yellow or Red

Don’t reach for Hue/Saturation. Open a Selective Color adjustment layer, target the Reds and Yellows, and reduce Cyan, Magenta, and Yellow by small amounts (3-6 points each) while watching the result on a calibrated monitor. Blend this layer at 60-70% and mask carefully to the sclera only, avoiding the iris and the skin around the eye. If there’s redness from visible vessels, a separate step using the Healing Brush at 60-70% opacity on a new layer, sampling clean sclera, removes individual vessels without affecting the overall tone. Never use the Patch tool inside the eye. It smears, and the smearing will be visible at any reasonable print size.

Matching the Enhancement to the Light in the Image

Every adjustment you make to the eyes has to answer to the light that was present when the image was shot. If the key light is warm and directional from the upper left, your catchlights should be warm and positioned upper left. If the subject is lit with a large softbox producing even, cool illumination, a hard bright specular in the eye will fight the entire image. I look at the catchlight first and let it tell me what the eye needs. The original light is always the reference point, not the reference image on Pinterest.

The goal of eye enhancement is not brighter, not bluer, not more dramatic. The goal is presence. Get that right, and the viewer will feel it without ever knowing you touched it.