Masking took three years until Select Subject clicked once
Masking, curves, colour grading and retouching sit behind almost every finished photograph. Three of them move fast. The fourth takes longer and rewards the time.
The one that changed fastest
Masking used to mean painting. You opened Lightroom, reached for a brush, and spent twenty minutes tracing around a sky, a face or a patch of shadow that needed different treatment from everything else. The brush radius never quite matched the edge. You painted over, painted back, reduced flow, increased flow, zoomed to 400 percent to clean up a fringe of sky that had leaked into a hairline. It was thorough, occasionally satisfying, and deeply slow.
Select Subject, which arrived in Lightroom Classic 10.3, changed the arithmetic. The feature uses Adobe's Sensei machine-learning stack to read the content of the image and draw a subject mask in roughly the time it takes to exhale. Select Sky had come slightly earlier and worked on the obvious tonal split between a bright sky and a darker foreground. Select Subject pushed further: it would find a person, a bird or a car sitting against a complicated background and return a mask that, in good light and reasonable contrast, required almost no refinement. Three years of hand-painting, compressed to a click, is not a metaphor. For portrait and wildlife work it literally reclaimed hours across a catalogue.
The masking panel in Lightroom Classic now stacks multiple masks with intersection and subtraction logic — Select Subject minus Select Sky gives a person extracted from a bright background in two operations. Capture One arrived at comparable AI masking on its own schedule, calling the equivalent feature Magic Brush and later AI Masking. GIMP, which is free and open-source, has no AI selection equivalent at anything like that quality; its Fuzzy Select and Scissors tool work on colour adjacency and require steady hands. The difference is real and worth naming plainly.
What masking cannot do is save a bad base edit. A mask only applies whatever adjustments sit underneath it, so a poorly exposed layer beneath a precise mask still looks poorly exposed in a precise outline.
The one that has not changed once
Curves is the oldest idea in digital image editing and almost certainly the most powerful. It was in early versions of Photoshop, and it will be in whatever comes after the programs that exist now. The concept is as close to a permanent piece of the craft as anything in software: a diagonal line representing the relationship between input brightness and output brightness, which you bend to alter that relationship wherever you choose.
A slight S-curve — the midpoint of the line lifted a fraction, the shadows held, the highlights held — adds contrast without touching the Contrast slider, because you decide exactly which tonal range gets the treatment. Lifting the curve's bottom-left anchor point opens the shadows without blowing the highlights. Placing a point at three-quarters of the way up the curve and pulling it down compresses the bright tones, which is how you bring texture back into an overlit sky before even reaching for a mask. None of this is new. Ansel Adams worked out the underlying logic by hand, exposing for shadow detail and developing for highlight density — the tone curve is the digital articulation of the same negotiation. His zone system ↗, which divided a scene into eleven steps from black to white, is essentially a tone curve described in the darkroom.
Per-channel curves — adjusting Red, Green and Blue curves independently — change colour. Lift the red curve in the shadows and the dark tones go warm. Pull the blue curve's highlights down and the bright tones shift toward yellow. This is not a separate tool or a more advanced mode; it is the same diagonal line, applied to one channel at a time. Learning where to grab the curve and how far to pull it is the real investment curves ask for. The reward is that, once the hand knows it, curves replaces brightness, contrast, fade and most of what the Levels dialogue ever offered. Curves are the one control worth that investment.
The one that sets the mood
Split toning — applying one colour cast to shadows and a different one to highlights — has been in darkroom practice since hand-tinting prints in the nineteenth century. In digital editing it arrived as a two-slider panel (Shadows hue and Highlights hue, with a Balance control to tip the mix) and then matured in Lightroom Classic 10 and Camera Raw 13 into the Colour Grading wheel, which gave a three-way arrangement: shadows, midtones, highlights, each with a full hue and saturation wheel rather than a single hue slider.
The upgrade mattered because midtones carry most of the information in a normally exposed image. Adding warmth or coolness specifically to midtones changes the mood of a photograph without touching the deepest shadows or the brightest specular highlights. The common commercial grade — warm shadows, cooler highlights — gives a look that reads as considered because it reverses the temperature logic of natural light, where shadows are typically cooler than open sky. Overdriving the saturation of either wheel produces the colour-contaminated look that marks an edit as effortful. Restraint, meaning wheel saturation values in single digits for most images, keeps the grade invisible.
Capture One's colour grading tools work on a similar three-way model and have long had a reputation for producing cleaner, less posterised results in the transition zones. Whether that difference is meaningful in practice depends on the file and the eye reading it. The relevant variable is never which programme's wheel has more pixels; it is the hand's decision about how far to push.
The one that requires a reason
Retouching — removing or diminishing something that is genuinely there in the scene — sits apart from the other three because it crosses from editing into altering. The distinction is not moral. Sensor dust on a sky is objectively not part of the photograph; removing it is correction, not deception. A wire crossing behind a subject's head that reads as growing from the subject is a composition problem; fixing it is sensible. These are clear cases.
The Healing Brush and Clone Stamp in Photoshop have been the standard tools for this work since the late 1990s, replaced for many tasks by the content-aware equivalents that arrived later. Lightroom's own Heal and Clone options within the masking panel are slower and less capable than Photoshop's, but they handle sensor dust and small distractions without opening a second application. For anything structural — removing a large object, rebuilding an area of background, filling behind a moved subject — Photoshop's Generative Fill (using the Firefly model, available from Photoshop 24.5 onwards) does the work faster than any manual patch. The quality of the result depends entirely on what surrounds the area being filled; a clean, regular background reconstructs well and a complicated one often does not.
The word "retouching" covers a large range. On this site's scope, it stops where the subject's form begins: light, colour, dust and distraction are fair targets. Where that line sits is worth thinking through before the tool is reached for.
These four controls — masking, curves, colour grading and retouching — account for most of what separates a finished photograph from its raw form. Three of them have been transformed by machine learning and new interfaces in the past five years. Curves was sitting there before any of that and will still be sitting there when the next transformation arrives.