Masks in Photography: A Practical Guide to Selective Editing (2026)

Published October 16, 2026 · Daniel Tello · RevelRaw

You're editing a photo that almost works. The sky outside the window looks good, but the face is too dark. Push the exposure up and the sky blows out. Pull it down and the face disappears into shadow.

That's the point where masks in photography stop feeling optional and start feeling practical. A mask lets you tell the editor, “apply this change only here,” so one part of the frame can breathe without flattening the rest. That is the opposite of generative tools such as WearView's ai fashion photography, which renders a garment onto an AI model and produces a whole new image. A mask keeps the photograph you actually took and changes only the part that needs it.

Why Do Masks Matter in Everyday Edits?

A global slider is blunt. It affects the whole frame, which sounds efficient until you're trying to rescue a bright sky and a dim face in the same portrait. If you raise exposure across the image, the skin tones may look better but the highlights can burn out. If you darken the frame to protect the sky, the subject sinks further into shadow.

A mask solves that by separating the image into regions you can treat differently. Think of it as a note taped onto your editing controls: this adjustment goes here, not there. That's why the tool matters long before you care about which button opens it.

The first thing to picture

When a photographer says “mask,” they usually mean a selection with attitude. It isn't just a dotted outline, it's a decision about where light, color, contrast, or sharpness should change. Once you start thinking that way, the tool menu stops looking like a pile of features and starts looking like a set of ways to draw boundaries.

That boundary can be broad, like a sky. It can be soft, like a cheek fading into shadow. Or it can be tiny, like the catchlight in an eye. The point is always the same, local control without dragging the rest of the photo with it.

Practical rule: if a single adjustment makes one part of the photo better and another part worse, a mask is usually the cleaner fix.

The idea is simple enough that photographers use it in portrait work, scenes, product shots, and even fast-paced shoots where they need to keep attention where it belongs. In studio and fashion work, local edits become part of the lighting language, not just a cleanup step.

What Is a Mask in Photo Editing?

A digital mask is easiest to understand if you stop thinking of it as a hidden switch and start thinking of it as a grayscale map. White means full effect, black means no effect, and gray means partial effect. The image itself stays put, while the mask decides where the adjustment lands.

That basic idea has a long photographic history. In the nineteenth century, combination printers shielded parts of the paper so that several negatives could be printed onto one sheet, often a separate sky over a landscape that a single exposure could not hold. That practice fed into photomontage, and Graham Vickers' 1989 Computer Bulletin article, an Academic overview of photographic masking and photomontage history, called the computer the first major advance in image manipulation since that century. The same logic appears in unsharp masking, which began chiefly as a tonal-reproduction technique, including correct color borders in CMYK printing, with sharpening a by-product. Till Heilmann's 2025 history of unsharp masking and its reproduction roots traces how sharpening became its main purpose only once it went digital.

From physical shielding to digital paint

The old method was literal. A photographer blocked light from part of the paper, then let another part receive exposure. Today the editor does the same job with software, only the boundary is drawn on screen instead of with a card in the darkroom. Selection tools are just faster ways of painting that map than doing it all by hand.

That's why a mask feels like a layer sitting on top of the photo. The underlying image doesn't vanish, it's just being addressed in a particular region. Once you get that picture in your head, the rest of the tools become easier to sort.

A good mask doesn't change everything. It changes the right thing cleanly enough that nobody notices the boundary.

When Should You Use a Linear, Radial or Brush Mask?

Manual masks are the three basic shapes most editors give you, and each one exists because real photos tend to fall into familiar patterns. A horizon needs a straight transition. A face often needs a round or oval lift. A stray highlight on a jacket or glass needs a hand-drawn fix.

Linear for skies and flat transitions

Use a linear mask when the adjustment should fade across a broad plane, like a dim foreground or a sky that needs calming down. On a street scene, you'd drag the gradient from the top edge downward, so the sky gets the strongest effect and the buildings below stay mostly untouched. If the horizon is tilted, rotate the gradient so it follows the scene instead of fighting it.

The main control here is the feather, because a hard edge usually looks pasted on at print size. A soft blend gives the eye a natural place to travel. If the feather is left too tight, the transition can show up as a line you only notice after export.

Radial for faces and focal points

A radial mask works best when the subject is roughly oval, which is more common than it sounds. On a portrait, you can place it over the face and upper torso to lift exposure or warmth without brightening the whole frame. It also works well for guiding the eye toward a performer, a product, or a window-lit subject in a crowded scene.

The key controls are size, softness, and placement. If the mask feels visible, it's usually too small or too sharp. A gentle falloff keeps the edit present but quiet.

Brush for one stubborn detail

A brush mask is what you use when the job is too irregular for a gradient. Think a shiny edge on a sleeve, a single reflection in a lens, or a small patch of skin that caught extra light. You paint the area you want, and the mask follows your hand.

For brush work in editors that offer them, flow and density matter as much as size. Lower settings let you build the mask gradually, which is safer around hair, texture, and hard edges. If you rush this part and leave the transition abrupt, the edit can look obvious even when the adjustment itself is subtle.

• Linear mask: best for skies, horizons, and broad tonal moves.
• Radial mask: best for subjects, faces, and centered emphasis.
• Brush mask: best for odd shapes, tiny corrections, and selective dodge and burn.

If the part you want to protect is easier to outline than the part you want to edit, invert the mask and work from the safer side.

Where Do AI Mask Selections Fail?

A quick click on subject, sky, or background can save time, especially on clean portraits and simple scenes. The tool gives you a strong first pass. The trouble starts where the edge stops being obvious.

Clean edges are the easy case

A studio portrait against a plain background is the kind of frame AI selection tools handle well. The outline is clear, the subject stands apart, and the mask can look finished at a glance. The genuine test comes with bright windows behind a subject, loose hair, foliage, lace, veils, and glasses. Those details break the boundary into fragments.

A mask can still look fine on screen and cause halos, color contamination, or missing texture after sharpening or noise reduction. Fine edges need inspection before you trust them.

What the measurement means

In segmentation models, mask quality is often measured with intersection-over-union, or IoU. It compares how much the predicted mask overlaps the target region. That is useful during development, but a photographer still has to read the result in the file. A mask can score well and still miss wisps of hair or bite into a shoulder line.

Google's on-device segmenter for Snapseed fine-tuned its teacher model on roughly 30,000 high-quality masks across more than 350 object categories, predicts its mask at 768×768, and upsamples it to the image with edge-preserving filtering, capped at 4K, according to Google's write-up of on-device segmentation research and practical masking limits. Those limits matter because small detail errors can survive the selection stage and show up later, especially in high-resolution RAW files after sharpening or denoising.

When to trust it, when to fix it

AI masking works best when the subject boundary is clean and the edit is broad. It needs help when the edge is thin, transparent, backlit, or layered over similar tones. In those cases, subtract with a brush, or rebuild the mask by hand if the selection cuts too far into the subject.

For a closer look at how on-device AI selections behave in real editors, see this guide to AI photo editing software. Some editors, including RevelRaw, offer click-to-select faces, skies, and backgrounds alongside manual masking tools. In RevelRaw that needs macOS 27, and Apple's selection model downloads on first use and then runs on your Mac; linear and radial gradients and the brush work on macOS 26 and later. That setup keeps the edit responsive while you check the selection yourself.

How Do You Layer Masks and Refine Edges?

A single mask can do a lot, but layered masks give you more control over what changes first and what stays protected. A RAW file with a sky, a face, and a small highlight doesn't need one giant fix. It needs several smaller decisions stacked on top of each other so each region gets the treatment it deserves.

Stack the work instead of forcing one mask to do everything

A common sequence on a portrait is simple. Start with a linear mask on the sky, add a radial mask to lift the face, then paint a brushed dodge over the eyes or cheek catchlight. Each layer solves one problem, and the later masks can be tuned without disturbing the earlier ones.

That's where invert becomes useful. If the area you want to protect is easier to define than the area you want to adjust, invert the mask and work from the opposite direction. The shortcut saves time and usually keeps the edge cleaner.

Refinement is what makes the edit hold up

The part that separates a usable mask from a fragile one is edge refinement. In layer editors such as Photoshop's Select and Mask, feather radius, contrast, and shift edge control how the selection blends, how crisp it feels, and whether the boundary sits slightly inside or outside the subject. If the transition looks fine at screen scale but shows a hard ring at print size, the edge needs more work.

Practical rule: refine the edge before you push the adjustment. If the mask is wrong, stronger edits only make the mistake easier to see.

A recovery move helps in a layer-based editor when an AI selection is close but not quite right. Duplicate the layer, hide the copy, paint black into the original mask along the halo, then turn the duplicate back on to compare. That gives you a controlled way to repair the problem without starting from zero.

RAW editors usually offer a simpler set. RevelRaw gives each of its up to three mask layers Feather, Amount and Invert plus brush refinement, with its own exposure, contrast, highlights, shadows, warmth, tint, saturation, clarity and sharpening. Whatever the editor, read the Camera Raw sharpening guidance alongside this, because sharpening makes weak mask edges much easier to see.

How Do You Check a Mask Before Export?

Once the masks are in place, the job isn't done. The final pass is where you catch the small errors that only show up after resizing, sharpening, or denoising. That inspection matters most on high-resolution files, because a tiny boundary mistake can become visible once the image is exported or printed large.

A four-step check that works on any masked edit

First, view the mask overlay in grayscale at 100% so you can see the actual boundary, not just the pretty result. Second, cycle the before-and-after view at the output size you plan to deliver. Third, look for halos against high-contrast borders, especially hair against sky, skin against bright windows, and buildings against clouds. Fourth, check the file again after sharpening and noise reduction, since both can exaggerate edge errors. If you publish the result, a comparison slider shows the edit honestly; the guide to Repeatable before-after checking for masked edits shows how to build one.

A mask that looks fine at fit-to-screen can break apart when the file is sharpened or printed.

That's especially important for RAW files, where the edge may seem perfect until the surrounding texture gets cleaned up or the image is enlarged. Fix the boundary before final export, not after, because post-export correction is slower and usually less exact.

Privacy masking is a different job

This kind of editing is about tone, texture, and attention. Privacy masking is about hiding identity information, which is a different standard. The 2026 PrivateEdit paper masks the minimal inner facial region on the device before an image is sent to a third-party AI editor, then restores the face locally. It reports a tunable trade-off: larger masks conceal more but cost editability, while smaller masks preserve editability at the price of more biometric leakage (On-device privacy masking and its trade-offs).

For web delivery, SiteSelf's WordPress image guide covers resizing to the displayed width, WebP or AVIF, compression and lazy loading, but the mask itself should already be clean before it leaves your editor.

What Does a Full Masking Session Look Like?

A golden hour portrait usually starts with one annoying truth, the light is beautiful and the exposure is still uneven. I'd use an AI subject selection to protect the skin tones, then add a radial mask to lift the face just enough, and a linear mask to deepen the sky behind the subject. If the hairline or shoulder edge starts to glow, I'd refine the mask before export, not after.

For car work, the process feels different but the logic stays the same. Encore Film and Music's walkthrough of car studio photography behind the scenes treats a car as “a mirror with badges”: it stacks several locked-off exposures and masks in only the parts that improved, keeping retouching as refinement rather than rescue. On a cityscape, I'd use a linear mask to recover the sky, a brushed foreground mask to bring back clarity where the street falls flat, and a contrast tweak on the buildings so their edges stay crisp.

The sequence stays consistent. Select, stack, invert if needed, refine, inspect. If you keep that order in mind, masks stop feeling like a secret panel and start feeling like the same careful decision repeated in different shapes.

RevelRaw gives you linear gradients, radial gradients, and brush masks on macOS 26 and later, plus click-to-select faces, skies, and backgrounds on macOS 27, with up to three free mask layers per photo. Gradient and brush layers can be synced across a series; AI selections stay with their photo. Processing runs on your Mac, photos are not uploaded, and it requires an Apple M1 or later Mac. If you want local RAW editing and selective control in one place, visit RevelRaw and try the masking tools on a photo that needs one part fixed without changing the rest.

Frequently Asked Questions

What is a mask in photo editing?

A mask is a grayscale map that limits an adjustment to part of the image. White areas get the full effect, black areas get none, and gray areas get a partial effect. The photo itself is not changed; the mask only decides where an exposure, color or detail adjustment applies.

When should you use a linear, radial or brush mask?

Use a linear mask for skies, horizons and broad transitions across the frame. Use a radial mask for faces, subjects and focal points. Use a brush for irregular shapes and small details. If the area to protect is easier to select than the area to change, invert the mask.

Why do AI masks fail on hair and glass?

Thin, transparent or backlit edges break the boundary into fragments that segmentation models struggle to follow, and the mask is often predicted at a lower resolution and then upsampled. Check those edges at 100 percent, then paint the selection in or out with a brush where it misses.

How do you hide the edge of a mask?

Feather the boundary so the adjustment fades instead of stopping on a line, keep the adjustment moderate, and check the result at 100 percent and at the final output size. Look again after sharpening and noise reduction, because both make a hard or haloed edge more visible.

Does RevelRaw have masks?

Yes. RevelRaw has free selective edits with linear and radial gradients and a brush on macOS 26 and later, plus click-to-select faces, skies and backgrounds on macOS 27. Each photo can have up to three mask layers with their own light and color controls, feather, amount and invert. It requires an Apple M1 or later Mac and processes photos on the Mac.