Chapter 31: Color Correction

Overview

Shoot the same person, in the same chair, four times across one afternoon, and you will get four different colors. The first take, at noon, is cool and blue from the sky through the window. The second, an hour later, has warmed a stop as clouds thin. For the third you turned on a lamp for fill, and now half their face is orange. The fourth you shot on a second camera whose white balance was set one notch off. Cut those four takes together and the scene flickers — the wall changes color at every edit, the skin goes from ruddy to sickly and back, and a viewer who could never tell you why will feel that something is wrong within seconds. Nothing was wrong with the acting, the framing, or the light. The shots simply do not agree with one another, and disagreement reads, instantly, as amateur.

The fix is the subject of this chapter. Before anyone builds a mood, a "cinematic look," or an orange-and-teal grade, every shot has to be pulled back to a clean, honest, neutral starting point — real black, real white, skin that looks like skin, and a scene where shot two matches shot one so seamlessly that the cut disappears. That work is color correction, and it is the least glamorous, most important color skill you will learn. Grading gets the magazine covers. Correction is what separates footage that looks like footage from footage that looks like a film.

The single most important idea in the chapter is this: you do not do correction by eye. Your eye is a liar — a magnificent, adaptive, untrustworthy liar that will "fix" a color cast for you within about a minute of looking at it, then insist the shot is fine while the scope screams that it is blue. So you learn to read three simple instruments — the waveform, the RGB parade, and the vectorscope — and you let them tell you the truth your eye is hiding. This is the chapter where you stop trusting your monitor and start trusting your scopes.

In this chapter you will learn to:

• Separate color correction (making it neutral and consistent) from color grading (making it feel like something) — and see why correction must come first.
• Read the three scopes — the waveform, the RGB parade, and the vectorscope — well enough to diagnose any shot in seconds.
• Set white balance in post, killing a color cast by balancing the channels against something that should be neutral.
• Set a clean black point and white point, and shape contrast without destroying shadow or highlight detail.
• Practice matching shots so a whole scene lives in one consistent world.
• Run a repeatable primary correction pass on your projects — including the Café Scene, which you will finally pull to a neutral baseline here, ready to grade in the next chapter.

Learning Paths

Where your attention pays off most in this chapter. Everyone should learn to read the scopes (§31.2) — it is the foundation of everything that follows, in every editor, forever.

• 📱 Phone-first: §31.3 and §31.6 are your core. Phone footage often has an auto-white-balance cast and baked-in contrast; correction is how you rescue it. You do not need a color-grade monitor — you need the scopes, which are free.
• 🎥 Creator: §31.2 and §31.5. Matching your A-roll to your B-roll (and your face to last week's face) is what makes a channel look consistent and professional across episodes.
• 💼 Pro-track: all of it, but especially §31.5 (matching) and §31.6 (workflow). A client scene that does not match shot-to-shot is the fastest way to look like an amateur on a paid job.
• 🎓 Student: §31.1 (the correction-vs-grading distinction) and §31.4 (black point, white point, contrast) are the conceptual and technical spine. Learn the order of operations cold.

31.1 Correction vs grading: two different jobs, in the right order

Color work in post splits cleanly into two jobs, and confusing them is the single most common way beginners make their footage worse instead of better. Let us name both, precisely.

Color correction is the technical work of balancing an image to a neutral, accurate, consistent baseline: setting real black and real white, removing any color cast so that whites are white and skin looks natural, and making every shot in a scene match every other shot. Correction has a right answer. Two competent colorists, handed the same shot and told to correct it, will land in nearly the same place, because correction is measured against an objective target — neutral, natural, consistent — that the scopes can confirm. Correction asks: is this image honest and does it agree with its neighbors?

Color grading — the subject of the next chapter — is the creative work of building a look: pushing the corrected image deliberately away from neutral to create mood and feeling. A warm nostalgic amber, a cold thriller steel-blue, the bleached heat of a desert. Grading has no right answer, only better and worse choices in service of the story. Grading asks: what should this feel like?

The order is not negotiable, and it is worth stating as a rule you will follow for the rest of your working life: correct first, grade second. Here is why. A look is a set of deliberate shifts away from neutral. If you apply that look to an image that is already shifted — one that is too blue, too dark, and doesn't match the shot before it — you are stacking a deliberate shift on top of an accidental one, and you can no longer tell which is which. Your careful warm look lands differently on the blue shot than on the neutral shot, so your "consistent" grade comes out inconsistent. You will chase your tail for an hour. Correct every shot to the same clean baseline first, and a single look dropped on top of all of them will land identically on each. Correction is what makes grading possible.

FIGURE 31.1 — The finishing pipeline: you CORRECT before you GRADE

  flat / Log   →   PRIMARY CORRECTION    →   a matched, neutral   →   GRADE      →   export
  capture          • set black & white       baseline                 (the look,     & deliver
  (Ch.2–3)         • neutralize the cast      (every shot agrees)      the mood)      (Ch.36)
                   • match shot to shot       ──────────────────       Ch.32
                   THIS CHAPTER (Ch.31)       the handoff point

   Rule: never build a look on top of an unbalanced, mismatched image. You would be
   painting mood onto a crooked wall. Straighten the wall first — that is correction.

There is one more distinction to plant now, because it organizes both chapters. The corrections in this chapter are primary correction: adjustments that affect the whole frame at once — the entire image gets brighter, or warmer, or its blacks get set. That is different from secondary correction, which isolates and changes only part of the frame (just the sky, just the shirt, just the skin), and which — along with the node tree that organizes these operations — belongs to the grading toolkit in Chapter 32. In this chapter we stay global. We are balancing the whole image, not painting inside it. Whole-frame balance is nine-tenths of the battle and almost always the only color work a straightforward talking-head or documentary actually needs.

🚪 Threshold Concept: correct before you grade, and neutral before pretty. The move that separates people who color well from people who make their footage worse is discipline about order. Amateurs open the footage and immediately reach for a "cinematic" look — they grade a broken image. Professionals spend the first pass making every shot boringly, honestly neutral and consistent, then build the look on that foundation, once. Learn to love the neutral pass. It is not the exciting part, and it is the part that makes the exciting part work.

🔗 Connection. This chapter finishes what Chapter 3 started. Back there you chose a capture format — a standard profile or a flat Log profile — and we promised that the flat, washed-out look of Log existed so that you could correct and grade it here. If you shot Log, this is where it pays off: correction is the step that expands that flat image back into real contrast and color. If you shot a standard profile, correction is lighter but never skipped. Either way, Chapter 32 then builds the look on top.

31.2 Reading scopes: the waveform, the parade, and the vectorscope

Here is the uncomfortable truth that makes the rest of this chapter necessary: you cannot trust your eyes to judge color. Not because you have bad taste, but because human vision is built to do the exact thing that ruins color work. Your visual system constantly adapts — it decides what "white" is in a scene and silently corrects everything relative to it, which is wonderful for surviving in the world and catastrophic for correcting an image. Stare at a slightly blue shot for sixty seconds and your brain will quietly re-baseline, deciding that blue is white now, until the shot looks perfectly neutral to you while a scope reports it is a mile off. Walk away, get coffee, come back — and it looks blue again. Your monitor lies too: its brightness, its color, the color of the wall behind it, and the light in your room all shift what you see. You need an instrument that does not adapt, does not get tired, and does not care what color your walls are.

That instrument is a set of scopes — real-time graphs that measure the actual signal in your footage and display it as objective numbers, immune to your eye's adaptation and your monitor's quirks. Every serious editor has them; they read the same math the same way everywhere. There are three you must know. Learn to read these three and you can correct any shot in any software on any monitor, even a bad one.

The waveform: a map of brightness

The first scope you met already, back in Chapter 5, as an exposure tool on set. The waveform graphs the brightness (luminance) of your image: the vertical axis is how bright, from 0% (pure black) at the bottom to 100% (pure white) at the top, and the horizontal axis lines up with the left-to-right position in the frame, so a bright window on the right of the picture shows up as a bright cluster on the right of the scope. It is the same tool you used to nail exposure in the camera; now you use it to set and check tone in post.

FIGURE 31.2 — The luma waveform: a map of every tone in the frame
   (vertical = brightness, 0% black at the floor to 100% white at the ceiling;
    horizontal position matches left-to-right position in the picture)

  100% ┤· · · · · · · · · · · · · · ·   ← clipping ceiling (do not pile tone here)
   90% ┤               ▓▓▓▓              the window: bright, but held below 100
   75% ┤          ▓▓▓▓▓▓▓▓
   60% ┤   ▓▓   ▓▓▓▓▓▓▓▓▓▓▓▓            ← a lit face usually lands around 45–70%
   45% ┤ ▓▓▓▓ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓
   25% ┤▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓            the shadow side of the room
   10% ┤▓▓▓▓▓▓▓                         ← darkest corners, still above the floor
    0% ┤· · · · · · · · · · · · · · ·   ← black floor (do not pile tone here)
       └──────────────────────────────
        left             image             right

   Nothing is welded to the 0% or 100% line: you have real black, real white, and
   everything in between — a full, uncrushed, unclipped tonal range to work with.

Read a waveform for three things. First, where is black? The bottom of the trace should touch or nearly touch 0% somewhere in a normal shot — otherwise your blacks are milky and grey and the image looks flat and foggy. Second, where is white? The top of the trace should reach up near 100% if the shot contains anything genuinely white or bright — otherwise the image is dull. Third, and most important, is anything welded to a rail? A trace piled flat against 0% or against 100% is clipped — the tone has hit a wall and the detail that was there is simply gone.

FIGURE 31.3 — Crushed and blown: tone welded to the rails is gone for good
   TOO FAR (contrast cranked past the limits)          WHY IT MATTERS
  100%┤███████████· · · · · · · ·  ← BLOWN highlights   A rail is a wall. Once a tone
   90%┤███████████                    (the window is a    is clipped flat to 0% or to
   50%┤     ▓▓▓▓                       white hole now,    100%, the detail that lived
   10%┤▓▓▓                            no detail left)    there is deleted — and no
    0%┤███████████· · · · · · · ·  ← CRUSHED shadows     slider will bring it back.
      └────────────────────────       (the black jacket  So push contrast UP TO the
       clipped at BOTH ends            is a black void)   rails, never THROUGH them.

This is the number-one beginner mistake in color and it deserves a callout of its own below. The scope makes it impossible to commit by accident, which is precisely why you keep your eyes on the scope and not only on the pretty picture.

The RGB parade: a map of color balance

The waveform shows brightness. To see color, you split that same idea into three. The RGB parade shows three waveforms side by side — one for the Red channel, one for Green, one for Blue — so you can compare how much of each color lives in the shadows, midtones, and highlights. Its power comes from one simple fact: a neutral tone (pure white, any grey, pure black) is made of equal amounts of red, green, and blue. So on anything that should be neutral, the three traces should sit at the same height. When they don't, you have a color cast, and the parade shows you exactly which channel is misbehaving and at which brightness.

FIGURE 31.4 — The RGB parade catches a cast the eye has already "fixed" for itself
   (three brightness traces side by side — Red, Green, Blue. On a NEUTRAL image the
    three line up at the same heights, because grey/white/black are equal R, G, and B.)

   BEFORE — a cool cast (blue too high)          AFTER — balanced to neutral
      R      G      B                               R      G      B
  100│▓▓    ▓▓    ██   ← blue highs clip        100│▓▓    ▓▓    ▓▓   ← tops now level
     │▓▓    ▓▓    ██                               │▓▓    ▓▓    ▓▓
   50│▓▓    ▓▓    ▓█                             50│▓▓    ▓▓    ▓▓
     │██    ██    ██   ← blue floor lifted          │██    ██    ██   ← floors now level
    0│██    ██    ██      (blacks go blue)         0│██    ██    ██
      ‾‾     ‾‾    ‾‾                                ‾‾     ‾‾    ‾‾

   Blue rides higher than red and green at BOTH the top and the bottom, so the whole
   frame leans blue. The fix: pull blue down (or push red up) until all three tops
   align and all three bottoms align. Now white is white, grey is grey, black is black.

The parade is your primary instrument for correcting a cast, and §31.3 is built entirely on it. The habit to build: when you look at the parade, find the parts of the image that should be neutral — a white shirt, a grey wall, a black jacket — and check whether R, G, and B line up there. If blue is high at the bottom, your shadows are blue; if red is low at the top, your highlights are missing red and skewing cyan. The parade doesn't just tell you that there's a cast; it tells you where in the tonal range to fix it.

The vectorscope: a map of hue and saturation

The third scope ignores brightness entirely and shows only color itself — its hue and its intensity. The vectorscope is a circular graph where the direction a point sits from the center tells you its hue (which color), and the distance from the center tells you its saturation (how intense). Dead center means zero saturation — a perfect neutral grey. The further the trace spreads toward the edge, the more saturated the image; the direction it leans tells you toward which color.

FIGURE 31.5 — The vectorscope (schematic compass): hue = the direction a color points,
              saturation = how far from center it reaches

                        R  ──────  Mg
                       /    \    /    \
                     Yl      \ S/      B          S = the SKIN-TONE LINE
                       \      \/      /               (from center toward R / Yl,
                        \     ●      /                about the 11 o'clock direction)
                       G  \   |   /  Cy
                           \  |  /
                            \ | /               ● = center = neutral grey
                             \|/                    (zero saturation)

   Read it three ways:
     • Neutral shot  → the glowing trace huddles around the CENTER (little/no color).
     • Color cast    → the whole trace slides off-center toward the offending hue
                       (a blue cast drags everything toward B; a warm cast toward R/Yl).
     • Skin tone     → the little finger of trace thrown out by a face should land ON
                       the SKIN-TONE LINE (S) for EVERY complexion. Deeper skin reaches
                       farther out along it, lighter skin stays closer in — but the
                       ANGLE holds. That line is the single most useful landmark in color.

That skin-tone line deserves its own paragraph, because it is the most useful single fact in this entire chapter. There is a specific direction on the vectorscope — pointing up and to the left, between the red and yellow targets, at roughly the 11 o'clock position — along which healthy human skin of every complexion falls. This is not a target only for light skin. Deep brown skin, olive skin, pale skin, freckled skin — all of it lands along the same angle; what changes between complexions is how far out along the line the trace reaches (how saturated) and how bright it is on the waveform, not the direction. So when you have a person in frame, you have a built-in reference: the small arm of color their face throws should ride that line. If it leans off toward magenta, your skin is too purple; if it swings toward yellow, too jaundiced. Nudge it back onto the line and skin looks like skin. We will use this constantly.

💡 Why It Works: the scope doesn't adapt, and you do. Your eye's genius is chromatic adaptation — it decides what white is and silently recalibrates, so a white shirt looks white to you under warm tungsten and under cool daylight even though the actual light hitting it is wildly different colors. That is why you can read a book indoors or outdoors and the pages always look "white." It also means that after thirty seconds of looking at a blue-cast shot, your brain re-defines white as slightly-blue and the cast vanishes from your perception while remaining fully present in the file. The scope has no brain and no survival instinct. It reports the signal. This is the whole reason the scope exists: to see what your eye insists on hiding from you.

⚠️ Common Mistake: crushing and blowing while chasing "contrast." Beginners love the contrast slider because a quick shove of it makes almost any shot look punchier for the first two seconds — and then you notice the shadows have gone to featureless black and the highlights to featureless white. On the waveform you'd have watched the trace slam flat against 0% and 100%, welding tone to the rails. That detail is deleted, not hidden: no future slider recovers it, and if you shot 8-bit you may also see ugly banding appear in the sky. The fix is a discipline: keep one eye on the waveform, set black up to 0% and white up to 100%, and stop the instant the trace kisses a rail. Contrast is a cliff edge, and the scope is your guardrail.

🔄 Check Your Eye. Before reading on, answer from memory: 1. On the waveform, what does it mean when the trace is piled flat against the 100% line, and can you fix it later? 2. On the RGB parade, you look at a grey wall and see the Blue trace sitting clearly higher than Red and Green. What's wrong, and which way is the cast? 3. On the vectorscope, where does healthy skin sit, and does that change for darker or lighter skin?

Check yourself

1. The highlights are blown — clipped to pure white with no detail. It cannot be recovered; the information the sensor didn't record (or that got welded to the rail) is gone. Prevent it by watching the scope and stopping at 100%.
2. A grey wall should read equal R, G, B. Blue sitting higher means a blue (cool) cast across the image; pull blue down (or bring red/green up) until the three align.
3. Along the skin-tone line, roughly the 11 o'clock direction between red and yellow. The angle is the same for every complexion; darker or lighter skin sits farther out or closer in along that same line, not at a different angle.

31.3 White balance and neutralizing color casts

You met white balance as an in-camera control back in Chapter 12: telling the camera what color the light is (a warm 3200K tungsten bulb, a cool 5600K midday sky) so it can render white as white. When you nailed it on set, the footage arrives neutral and this section is a five-second check. When you didn't — you left the camera on auto and it guessed differently every take, or you set it to daylight and then a warm lamp crept in — the footage arrives with a cast, an overall wash of one color, and you fix it here. That post-production fix is white balance in post: neutralizing a color cast by rebalancing the red, green, and blue channels until something that should be neutral reads neutral again. It is the same goal as the in-camera control, achieved after the fact, with the parade as your guide.

The method is beautifully simple because you already have the tool. Find something in the frame that you know should be neutral — a white shirt, a grey concrete wall, the white of an eye, a black jacket, a coffee cup, a sheet of paper. This is your neutral reference. Put the parade up, look at where that neutral thing sits in the three traces, and adjust until R, G, and B line up there. That's it. That is white balance in post. A white wall that reads R=G=B is white; a white wall where blue is high is a blue room.

Most editors give you two roads to the same destination:

• The eyedropper (white/black balance picker). Point-and-click: you grab a tool, click on the thing that should be neutral, and the software calculates the correction to make that exact pixel neutral and applies it to the whole image. It is fast and often gets you 90% of the way in one click. It is only as good as the pixel you click, though — click a slightly-shadowed part of the shirt, or a specular highlight, and it will over- or under-correct, so always confirm on the parade afterward.
• Temperature and tint (manual balance). Two sliders that push along the two axes color casts actually travel: temperature slides the image warm (amber) to cool (blue), and tint slides it green to magenta. Between them they can neutralize almost any real-world cast, because almost every cast is some mix of "too warm/cool" and "too green/magenta." Fluorescent and LED lights are the classic green-cast culprits; mixed daylight-and-tungsten is the classic warm/cool problem. Nudge temperature and tint while watching the parade level out and the vectorscope trace pull back toward center.

Now the honest complication, and it is exactly the Café Scene's problem. What do you do when the shot has mixed light — two different colored sources at once, which you learned to recognize on set in Chapter 13? In our café, cool daylight pours through the big window (the room's motivated key, camera-right in the wide) while warm tungsten practicals hang over the counter. There is no single white balance that neutralizes both: correct for the window and the counter lamps go pumpkin-orange; correct for the lamps and the window goes arctic-blue. This is not a failure to fix — it is a decision to make. The professional move is to balance for the dominant, story-important source — here, the daylight falling on the person's face — so that skin and the main subject read neutral and natural, and let the practical lamps stay a little warm. And that residual warmth is not a bug; it is motivated, believable, the way a real café with real lamps actually looks. You are not trying to erase the warmth of the room. You are making sure the person is honest and letting the room keep its character — which, not coincidentally, is exactly the warmth Chapter 32 will lean into when it grades this scene into a cozy look.

🎞️ Read This Sequence. Here is the Café Scene's establishing shot, before and after a white-balance pass, described so you can picture the change precisely. Nothing about the framing, light, or performance moved — only the balance of the channels.

FIGURE 31.6 — "The Café Scene: neutralizing the mixed cast"        [constructed teaching example]
  THE FRAME    The establishing wide from Chapter 1: the person small in a warm, worn room, the big
               window glowing camera-right behind the counter, amber practicals hanging over it.
  BEFORE       The whole frame leans cool-blue: the camera was left on a daylight balance but the
               auto-exposure drifted, and the shadows have gone distinctly blue. On the PARADE, blue
               rides high at the bottom; on the VECTORSCOPE the trace sits off-center toward blue;
               skin leans slightly cyan, a hair off the skin-tone line. It reads faintly clinical, cold.
  THE MOVE     Pick the neutral reference: the white ceramic coffee cup on the counter. Eyedropper it,
               then fine-tune on temperature/tint while watching the parade.
  AFTER        Blue drops until the cup reads R=G=B — white is white again. The blue leaves the
               shadows; skin settles onto the skin-tone line and looks like a real person. The window
               reads a clean, honest daylight; the counter practicals keep a gentle, believable warmth.
  THE EFFECT   The room stops feeling like a fridge and starts feeling like a place — neutral, natural,
               and ready. It is not "warm and cozy" yet; that is the grade. It is *true* now.
  THE LESSON   White-balance the dominant, story-important light (the face), confirm on the scope, and
               let motivated practicals keep their warmth. Neutral does not mean sterile — it means honest.

♿ Accessibility & Inclusion: honesty for every complexion, and never rely on color alone. Two things belong together here. First, "neutral" and "natural" must mean true to the actual person. The skin-tone line works for every complexion, but the goal is never to push deeper skin toward a lighter target or lighter skin toward a tan — it is to render that person faithfully, with their real depth and warmth intact. Historically, film and video technology was calibrated around light skin and rendered darker skin poorly; a colorist's job today includes undoing that bias, not repeating it. Correct toward the truth of the subject. Second: roughly one in twelve men has some color-vision deficiency, so a clean, correct luminance range (real black, real white, readable contrast on the waveform) is part of accessibility, not just aesthetics — it means viewers who can't fully distinguish your hues can still read your image. And when meaning rides on color later (a red "record" dot, a green "go"), never let color be the only signal; back it with shape, text, or position. Good correction serves everyone who watches.

🔄 Check Your Eye. You're correcting an interview lit by an LED panel plus a window. The subject's white collar reads with green sitting above red and blue on the parade, and the vectorscope trace leans toward green. Which slider do you reach for first, and which way?

Check yourself

Tint, pushed toward magenta — green casts are a tint problem, classically from LED and fluorescent sources. Watch the parade: bring green down until the collar's R, G, B align, then check temperature for any remaining warm/cool lean. Confirm skin lands back on the skin-tone line.

31.4 Setting black point, white point, and contrast

Before you can trust any color judgment, the brightness of the shot has to be right, because color and tone are entangled — a too-dark image reads as muddy and desaturated, a too-bright one as washed out, and a cast is far easier to see once the tones sit where they belong. So a correction pass sets tone first: the black point (how dark the darkest part of the image is), the white point (how bright the brightest part is), and the contrast in between. You do all three on the waveform, by eye on the scope, not by feel on the picture.

The controls have standard names you will meet in every editor, usually as three color wheels each with a brightness slider underneath, splitting the tonal range into three overlapping zones:

• Lift controls the shadows — the dark end. Raising lift lifts the black point (shadows get brighter, milkier); lowering it drops the black point (shadows get deeper). You set your black point here.
• Gamma controls the midtones — the middle, where faces and most of the important stuff live. It brightens or darkens the image's body without moving the extremes much. Most of your perceived "exposure" adjustment happens here.
• Gain controls the highlights — the bright end. Raising gain lifts the white point (highlights get brighter); lowering it pulls them down. You set your white point here.

(There is a fourth, offset, which slides the entire range up or down together, and it is the ideal place to do a global white balance because it shifts all three channels' whole range at once. Some colorists do their primary balance on offset for exactly that reason.)

The target for a normal, natural shot is simple and you read it straight off the waveform:

FIGURE 31.7 — Setting the tonal range on the waveform (a normal, natural shot)

   BEFORE (flat Log-ish capture)          AFTER (contrast set to the rails, not through)
  100%┤· · · · · · · · · · · ·          100%┤· · · · · · ▓▓ · · · ·   ← whites reach up
   75%┤       ▓▓▓▓▓▓▓                     75%┤        ▓▓▓▓▓▓▓▓            near 100 (not past)
   50%┤     ▓▓▓▓▓▓▓▓▓▓▓                   50%┤   ▓▓  ▓▓▓▓▓▓▓▓▓▓        ← midtones/face sit
   35%┤   ▓▓▓▓▓▓▓▓▓▓▓▓▓                   35%┤ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓          in a natural band
   20%┤     ▓▓▓▓▓▓▓                       10%┤▓▓▓▓▓▓                   ← blacks reach down
    0%┤· · · · · · · · · · · ·             0%┤· ▓ · · · · · · · · ·      near 0 (not past)
      └───────────────────────               └─────────────────────
      Flat: nothing near 0 or 100 →          Corrected: full range, real black and white,
      grey, foggy, no punch.                 nothing crushed or blown. THIS is neutral.

Set the black point by lowering lift until the bottom of the waveform trace just reaches down to 0% — until the genuinely darkest thing in the frame is genuinely black — and then stop. Set the white point by raising gain until the top of the trace reaches up near 100% — until the genuinely brightest thing is bright — and stop before it clips. Then adjust gamma to place the midtones (the face) where they look right, usually with skin somewhere in the rough 45–70% zone depending on complexion and mood. That is contrast: the distance from your black point to your white point. A flat Log or foggy image has its trace bunched in the middle and needs the range expanded to the rails; an over-contrasty image has tone slammed against the rails and needs it pulled back off them.

Two cautions the scope enforces for you. First — the one from the last section, now with a name attached to the tool — do not crush and do not blow. Lift too far down welds shadows to 0% (crushed); gain too far up welds highlights to 100% (blown). The waveform shows you the exact moment tone hits a rail; stop there. Second, respect the footage's headroom for pushing. This is where Chapter 3's choices come back: if you shot 10-bit, you can stretch contrast and balance color hard before the image falls apart; if you shot 8-bit with heavy compression, aggressive moves reveal banding (visible steps in smooth gradients like skies) and blocky color, because there simply isn't enough data. Correct gently on thin footage, harder on rich footage — the scope shows you the tone, but the bit depth decides how far you can move it.

🔬 The Tech (optional): scope scales, Rec.709, and why Log looks broken. Skip this and nothing downstream breaks. Your scope's vertical axis can be labeled a few ways: as a percentage (0–100%), in the old broadcast unit IRE (also 0–100 for our purposes), or — the default in some color software — as 10-bit code values from 0 to 1023, where 0 is black and 1023 is white (so 64 ≈ 6%, 940 ≈ 100% in the common "video range" mapping). They all describe the same thing; pick the percentage view if the numbers confuse you. Rec.709 is the standard "recipe" that says how a normal HD video's brightness and color should map for a normal screen — it is the target your corrected image should live inside, which is why "keep it 0–100%" and "keep it Rec.709-legal" mean nearly the same thing. And Log footage looks washed-out and grey straight off the camera precisely because it isn't Rec.709 yet: Log packs a huge brightness range into a flat, low-contrast signal to protect highlight and shadow detail, on the explicit promise that you will expand it back to Rec.709 in correction. That expansion is the first thing you do to a Log shot — a technical starting transform (in the next chapter we'll use a conversion built for exactly this) — before any of the manual moves above.

✂️ In the Edit: the gray card is the cheapest correction tool you own. Here is where the shoot pays the edit back, and it is worth changing your on-set habits over. At the head of every setup — every new camera position, every lighting change — hold a grey card (or even a plain white sheet of paper) in the frame under the actual light and shoot two seconds of it. In correction, you eyedropper that card and white balance is solved in one click, perfectly, every time, because you gave yourself a guaranteed-neutral reference. The same habit fixes exposure: a grey card reads a known value on the waveform. Ten seconds on set saves ten minutes per shot in post and makes your shots match automatically, because they're all balanced to the same physical reference. And the deeper discipline: turn auto white balance OFF and lock it manually when you shoot a scene, so the camera doesn't re-guess the color between takes and hand you four different casts to reconcile. Correction is easy when the shoot was disciplined; it is a slog when it wasn't. You shoot for the color pass, too.

🔄 Check Your Eye. You expand a flat Log clip's contrast by dropping lift and raising gain until it "pops," and the sky develops visible stair-steps of banding. Two questions: what caused the banding, and what does it tell you about how hard you can push this clip?

Check yourself

The banding comes from too little data for the move — most likely an 8-bit, heavily compressed capture (Chapter 3): stretching the tonal range spreads too few brightness steps across a smooth gradient, so the steps become visible. It tells you this clip has little headroom; correct it gently, avoid extreme contrast and saturation pushes, and next time shoot 10-bit if you know you'll be pushing color.

31.5 Matching shot to shot

Everything so far corrects one shot. The real job — and the one that most visibly separates professional work from amateur — is matching shots: making every shot in a scene share one consistent color and tone, so that cutting between them is invisible and the scene feels like one continuous place and time rather than a collage of clips shot under drifting conditions. A single corrected shot is nice. A scene of eight shots that all agree is what a viewer reads as "this looks like a real film."

The reason matching is even necessary is that reality refuses to hold still. Across a shoot, the sun moves and clouds pass, so your exteriors drift warmer and cooler by the hour. You reposition for the reverse angle and now the window that was behind the camera is in front of it, filling the shot with different light — in our café, remember, the establishing wide sees the window camera-right, but the reverse onto the seated person naturally puts that same window camera-left, a correct reverse that nonetheless changes how the light falls. You cut to a second camera whose white balance was set a notch off. You grab a pickup insert a week later in totally different weather. None of these is a mistake; all of them produce shots that don't match, and matching is how you reconcile them.

The method is a discipline, not a talent:

1. Pick the hero shot. Choose the one shot in the scene that is already closest to right — usually the widest or most story-important, the one that best establishes how the scene should look. Correct it fully and carefully, to a clean neutral baseline. This is your reference; everything else will be matched to it, not to some abstract ideal.
2. Match black and white first. Put every other shot's waveform next to the hero's and match the black point and white point — get their shadows sitting at the same floor and their highlights at the same ceiling. Tone match before color match; a shot that's simply darker will look like it has a cast until you fix its brightness.
3. Match the color balance. Now bring up the parade on both and match where each channel sits — align the shots' casts so a white wall is the same white in both, a grey the same grey. The vectorscope confirms it: both shots' traces should lean the same direction by the same amount (or both sit neutral).
4. Match the skin. The final and most sensitive check, because faces are what viewers watch. On the vectorscope, both shots' skin should ride the same point on the skin-tone line — same angle, same distance out. If the reverse angle's face is a touch cooler or less saturated than the hero's, nudge it until the two faces are siblings. When skin matches, the cut vanishes; when it doesn't, the viewer feels the person subtly change between shots even if they can't say why.

The most common matching job you will ever do is the two-camera or shot-reverse-shot conversation — an A camera and a B camera, or the same camera turned around, that were set slightly differently — and Case Study 2 walks that exact task end to end with scopes before and after. But the principle scales to anything: the four recurring setups all live or die on it. A talking-head where the A-cam wide and B-cam tight don't match tells the viewer there are two cameras; matched, it reads as one seamless interview. A product on a table where the hero angle and the detail insert don't match makes the product look like two different colors — fatal when that color is the client's brand. A walk-and-talk where the shot in sun and the shot in shade weren't matched flickers with every cut. An event covered by three cameras with three white balances is a rainbow until you match them to one. Learn to match, and every one of these becomes one consistent world.

FIGURE 31.8 — Matching the Café Scene's coverage to one neutral baseline
   Four shots, one scene, shot across twenty minutes as the window light drifted.
   Before, they do not agree; after, they cut together invisibly.

   SHOT (café coverage)        BEFORE (the scopes report...)        AFTER (matched to the hero)
   ─────────────────────       ───────────────────────────────      ────────────────────────────
   A  establishing wide        blacks ~8%, whites ~92%, skin on     the reference HERO: corrected
      (window camera-right)     the line, gentle warmth from lamps    clean; everything matches to it
   B  medium at the counter    blacks ~2% (crushed), whites ~98%,   raise lift to ~8%, pull green
                               a touch green from a fridge light     out on tint, warm to match A
   C  CU reaction (reverse,    blacks ~15% (milky, flat), cooler —  drop lift to ~8%, add warmth,
      window camera-left)       the window fills this side           nudge skin back onto the line
   D  insert: the phone screen blue-white glow; the skin in its     tame the blue spill, match the
                               spill has gone cold                   skin warmth to A and C

   Cut A → B → C → D and the room stays ONE room, one afternoon, one place.
   The scene never changed — your command of it did.

🎬 On Set (in the edit): match a scene to death. Take any scene you've shot with at least three different shots — the Café coverage, a two-camera interview, or any conversation. Correct the hero shot to neutral, then match every other shot to it using the four-step discipline above, checking each on the waveform, parade, and vectorscope. Constraint: no looks, no "cinematic" moves — neutral and matched only. Then play the scene at full speed and watch the cuts. Self-review question: does any cut make the room, wall, or skin visibly change color or brightness? Every place it does is a shot you haven't finished matching. Fix those, and you've done the single highest-value hour of color work there is.

🔄 Check Your Eye. When matching two shots, why do you match the black point and white point before you match the color, and not the other way around?

Check yourself

Because brightness and color perception are entangled: a shot that is simply darker or lower-contrast than the hero will look like it has a different color cast even when it doesn't. Setting both shots to the same tonal range first removes that false signal, so the color you then see on the parade is the real difference to correct — not an illusion created by mismatched brightness.

31.6 A primary-correction workflow: the repeatable order of operations

Everything in this chapter assembles into one repeatable order you run on every shot and every scene. The order matters — do these steps out of sequence and you fight yourself — so learn it as a fixed routine until it's muscle memory. This is primary correction as a workflow: whole-frame, scope-driven, neutral-first, per shot, then matched across the scene.

For a single shot:

1. Normalize / expand contrast. If it's Log or flat, apply the technical starting transform (or set it manually) to get into a normal contrast range. Then, on the waveform, set the black point to 0% and the white point to 100% without crushing or blowing, and place the midtones with gamma.
2. Neutralize the cast. On the parade, find a neutral reference and balance R, G, B — eyedropper it, then refine with temperature and tint. Confirm on the vectorscope that the trace pulls back toward center.
3. Check and place skin. On the vectorscope, confirm any face rides the skin-tone line; nudge if not. On the waveform, confirm skin sits in a natural brightness band.
4. Sanity-check the whole thing. Whites white, blacks black, skin honest, nothing clipped, nothing banding. The shot is now a clean, neutral, truthful baseline.

Then, for the scene:

5. Pick the hero, match the rest. Correct the hero fully (steps 1–4), then match every other shot to it — tone, then color, then skin (the four steps of §31.5), each verified on the scopes.
6. Play it back at speed and hunt for any cut where color or brightness jumps. Fix those. Now the scene is correction-locked and ready to hand to the grade.

That is the whole workflow, and it is the same in every editor. In DaVinci Resolve — this book's free reference software, and genuinely the strongest free color tool there is — all of this happens on the Color page, where the primary wheels, the temperature and tint sliders, and the eyedropper live, and where you can pop up the waveform, parade, and vectorscope together in a scopes panel. (Each of these operations lives on a "node," which is Resolve's container for a correction; we deliberately leave the node tree — how you chain and branch multiple nodes — to Chapter 32, because that structure is where grading lives. For a clean primary correction, you can do everything above on a single node.) In other editors the words differ but the tools are identical: the same wheels, the same temperature/tint, the same three scopes, the same order of operations. Appendix E translates the button names across the common editors so you can run this exact workflow wherever you already work.

🎒 Gear Note: you need scopes, not a $2,000 monitor. The temptation in color is to believe you can't do it right without a pricey calibrated reference monitor. You will want one eventually if color becomes your profession — a display whose color you can trust matters when you're chasing the last 5%. But for the 95% that is correction, the thing that actually keeps you honest is free and already in your software: the scopes. A colorist on a perfectly calibrated monitor and a beginner on a cheap laptop screen, both reading the same waveform and parade, will set the same black point and neutralize the same cast, because the scope reports the file, not the screen. So the priority order for a beginner's budget is: (1) learn to read the scopes — free; (2) get a cheap grey card so your shoots hand you an easy correction — a few dollars; (3) work in a room where the light doesn't change and isn't colored — free; and only much later, (4) a calibrated monitor. Do not let the lack of a fancy display stop you. The instruments that matter cost nothing.

🔗 Connection. With every shot corrected to a matched, neutral baseline, you have built the foundation the next chapter stands on. Chapter 32 takes this clean, boring, honest scene and gives it a look — the warm, cozy amber the Café Scene has been waiting for — using nodes, LUTs, and secondaries, all of which land consistently precisely because you did the correction first. Correction is the wall; grading is the paint. You just finished the wall.

Production Checkpoint

Your task: color-correct Project 3 — and, while your eye is calibrated, your other two projects — balancing every shot to neutral and matching shots within each scene using the scopes.

Open Project 3, your 5-minute branded piece, on the color page of your editor. Do not grade it — no looks, no "cinematic" anything yet. Run the primary-correction workflow from §31.6 on every shot:

1. Set tone. On the waveform, set a real black point (down to 0%) and white point (up to 100%) on each shot, without crushing or blowing. Place the midtones naturally.
2. Neutralize casts. On the parade, balance R, G, B against a neutral reference in each shot so whites are white and skin is honest. Use temperature and tint for the stubborn ones.
3. Match every scene. For any scene with more than one shot, pick the hero, correct it, and match the rest to it — tone, color, then skin — verifying on all three scopes. Play each scene back and kill any cut where the color or brightness jumps.

Then do the same, faster, to Project 1 (your talking-head) and Project 2 (your documentary short) — they'll take minutes each now that you know the routine, and they'll look dramatically more finished.

Why this matters: correction is the step that makes everything you've shot look like it belongs to one professional piece rather than a pile of clips. It is also the mandatory foundation for the grade you'll build next chapter — a look applied to a corrected, matched project lands cleanly; a look applied to an uncorrected mess amplifies the mess. You are not making Project 3 pretty yet. You are making it true and consistent, which is the harder and more valuable half of color.

Summary

Reference-grade recap. Correction makes an image neutral and consistent; grading makes it feel like something. Correct first, always.

The two color jobs:

The three scopes — what each measures and reads:

The primary-correction workflow (memorize this order):

1. Normalize/expand contrast → set black point to 0%, white point to 100% (no crush, no blow), place midtones with gamma.
2. Neutralize the cast on the parade (eyedropper a neutral, then temperature/tint).
3. Check skin on the vectorscope (on the skin-tone line) and its brightness on the waveform.
4. Sanity-check: whites white, blacks black, skin honest, nothing clipped or banding.
5. Match the scene: correct the hero, match every other shot to it (tone → color → skin), verified on scopes.
6. Play at speed; fix any cut that jumps.

Decision rules:

The top mistakes and their fixes: crushing/blowing while chasing contrast → watch the waveform, stop at the rails; correcting by eye → your eye adapts, use scopes; grading before correcting → neutral-and-matched first; over-neutralizing a motivated warm room to sterile grey → balance the face, keep the room's character; pushing thin 8-bit footage too hard → banding, so correct gently.

Craft principles in play: motivate every choice (keep motivated warmth, don't erase it); you shoot for the edit (a grey card and locked manual white balance on set make correction trivial); story is the boss (correct toward the truth of the subject and the scene, for every complexion).

Spaced Review

Bringing back earlier chapters so they stick.

1. (Ch. 5) You're setting the white point on a shot in correction and the waveform highlights pile flat against 100%. Back on set, which scope told you the same thing about an over-exposed shot, and which two tools — one a striped overlay, one a color map — warn you about clipping while you shoot?
2. (Ch. 5) Why is it far better to fix exposure in the camera using the waveform than to "fix it" by raising gain in correction — especially for shadow detail?
3. (Ch. 2) Correction can set a black point and neutralize a cast, but it cannot invent detail the sensor never captured. Which two camera settings from Chapter 2 most determine how much real detail and how clean an image you hand the color page in the first place?
4. (Ch. 2 / Ch. 5) Your footage looks noisy in the shadows, and lowering the black point in correction makes the noise crawl. What on-set choice (name the setting and the scope) would have prevented most of that noise?

What's Next

Every shot is now honest: real black, real white, no cast, skin like skin, and every shot in a scene agreeing with its neighbors. The image is true — and, deliberately, a little boring. That is exactly the launchpad the next chapter needs. In Chapter 32 we take this neutral, matched baseline and give it a look — building mood and feeling by pushing away from neutral on purpose, with the node tree, LUTs, secondaries that isolate just the sky or just the skin, and the warm café amber this scene has been waiting fourteen chapters for. Correction gave us truth; grading gives us feeling. And because you did the truth first, the feeling will land the same way on every single shot.