Case Study 2: Two Cameras, One Face — Correcting and Matching a Testimonial
This is a from-scratch production walkthrough, the opposite of Case Study 1's film analysis. We are going to shoot a small, real, do-it-anywhere job — a two-camera testimonial in the talking-head setup, one of the book's four recurring scenarios — deliberately end up with two cameras that don't match, and then correct one to a clean neutral baseline and match the other to it using the scopes, with the waveform, parade, and vectorscope shown before and after. This is the single most common correction task you will ever face, and by the end you'll be able to run it in your sleep.
Everything here works with two phones, or one camera you reposition for a second take, or a phone plus a camera — the mismatch problem is identical regardless of gear, and so is the fix.
The brief and the constraints
A small bakery wants a 40-second testimonial: the owner, to camera, saying why she started the business, cut for the shop's website and socials. We want it to feel like a real, warm, professional piece — which means the two angles we shoot must cut together as one seamless conversation, not two obviously different cameras.
The constraints, all real:
- Two angles, one subject, shot simultaneously. A wider "hero" shot (Camera A) establishes her and the shop; a tighter shot (Camera B) catches the emotional lines. We roll both at once so her performance is a single take we can cut between.
- Two cameras that will not match out of the box. This is the whole point of the case study. Two different cameras — or even two of the same model — almost never render color and exposure identically. One will be a touch warmer, one a touch brighter, and if we don't fix it, every cut between them will flicker.
- Window key, one bounce. Free, soft, flattering light (Chapter 13's window as a natural key) plus a bounce for fill. No expensive lighting — this is a job a solo shooter does for a local business.
- A neutral reference, captured on purpose. We will shoot a grey card at the top so correction is a one-click start (the §31.4 discipline). This ten-second habit is what makes the whole correction fast.
Gear and settings
You need: two cameras (phones are fine), two small tripods or steady surfaces, a bounce (a white foam board or even a bedsheet), a lav or shotgun mic on the subject (sound is half the picture — see Chapters 14–15), and a grey card (a few dollars, or a white sheet of paper as a stand-in).
⚙️ Settings Box — two-camera testimonial (a starting point, adjust to taste)
Setting Camera A (hero / wide) Camera B (tight) Why Resolution / fps 4K / 24 fps 4K / 24 fps match both so the edit and any reframing are clean Shutter 1/50 s (180° at 24 fps) 1/50 s matched motion; Chapter 2 Aperture f/4 (more in focus, establishing) f/2.8 (softer background) the wide shows context; the tight isolates ISO native (e.g. 400), exposed on the waveform native, matched brightness clean shadows; Chapter 5 White balance manual, locked (e.g. 5600K) manual, locked to the SAME number the one setting people forget — lock both cameras to the same manual value so they start close Picture profile standard or a gentle flat same profile on both mixing a Log camera with a standard one is a matching nightmare — match profiles Grey card shoot 3 s at the top, under the light shoot 3 s, same card, same light a guaranteed-neutral reference for one-click balance The single most important row is white balance: set both cameras to the same manual number, not auto. Auto white balance will guess differently on each camera and drift between takes, handing you a much harder correction. Locking both to one value gets you 90% matched before you ever open the editor.
Here is the truth this case study exists to teach, though: even with everything above done right, the two cameras will still not match perfectly. Different sensors, different color science, tiny exposure differences — they always drift apart a little. That last 10% is the correction job, and it's unavoidable no matter how disciplined the shoot. So we shoot well and we correct. Let's set up.
The setup
FIGURE CS31.4 — Two-camera testimonial: window key + bounce (top-down view)
[ WINDOW / soft key ] ☀
|
▼ (soft key on her near cheek)
( S ) ──────► eyeline to interviewer (beside Cam A)
↗ ↖
○ white bounce (window is the key; the bounce
(camera-right, fills the far cheek ~1.5 stops down)
fills shadow)
| |
((• lav on her collar |
| |
[ CAM A ] wide, ~35mm [ CAM B ] tight, ~85mm
f/4, her on a third f/2.8, cutting past Cam A's edge
(the "hero") (the emotional close-up)
Both cameras share the window key and the bounce, so the LIGHT is consistent.
The mismatch we'll fix comes from the CAMERAS, not the lighting.
Camera A sits at the interviewer's side for a natural off-axis eyeline (Chapter 19), framing her wider with the shop behind. Camera B sits just past A's field of view, longer and tighter, catching the same eyeline for a clean cut. Crucially, both cameras see the same window key and the same bounce — the lighting is consistent. Any color or brightness difference between the two angles is therefore the cameras' doing, which is exactly the kind of mismatch correction is built to reconcile.
We frame her deliberately (Chapter 6): on a third, correct headroom, looking across the empty side of the frame toward the interviewer.
FIGURE CS31.5 — Framing the hero (Cam A), 16:9
┌─────────────────────────────────────────┐
│ · · · · · · · · O · · · · │ O = upper-third intersection
│ · · · ( head ) · · · · │ (her eyes land near here)
│ · · · ( S ) · · · · │ she looks screen-right,
│ · · · (subject ) · · · · → │ into the empty "lead room"
│ · · · · · · · · · · · · · │ toward the interviewer
└─────────────────────────────────────────┘
lead room on the right; shop softly out of focus behind
We roll both cameras, she talks, we get a strong take. We also — this is the ten-second habit that saves the whole correction — hold the grey card in front of her, in the same light, for three seconds at the head of each camera's recording. Now we go to the edit.
The correction pass
This is the heart of the case study. We ingest the footage, sync the two angles by their audio, and open the color page with all three scopes visible. Now we run the workflow from §31.6: correct Camera A (the hero) to a clean neutral baseline, then match Camera B to it.
Step 1 — Read Camera A honestly
Before touching anything, we read the scopes on Camera A's shot. Ignore how it looks on the monitor; read the instruments.
FIGURE CS31.6 — Camera A, BEFORE: the scopes tell the truth
WAVEFORM (brightness) PARADE (R G B on her white collar)
100%┤· · · · · · · · · · 100│▓▓ ▓▓ ██ ← blue high at top
80%┤ ▓▓▓▓▓ ← whites │▓▓ ▓▓ ██
60%┤ ▓▓▓▓▓▓▓▓▓ only reach 50│▓▓ ▓▓ ▓█
40%┤ ▓▓▓▓▓▓▓▓▓▓▓ ~80% │██ ██ ██ ← blue lifted at bottom
20%┤▓▓▓▓▓▓▓▓▓▓▓ 0│██ ██ ██
12%┤▓▓▓▓▓ ← blacks sit milky at ~12% ‾‾ ‾‾ ‾‾
0%┤· · · · · · · · · ·
Diagnosis: (1) blacks lifted/milky (~12%, not reaching 0%), (2) whites dull
(only ~80%, not reaching 100%), (3) a cool/blue cast — blue rides high on the
white collar at both ends of the parade. VECTORSCOPE: trace pulled slightly
toward blue; her skin sits a touch off the skin-tone line, leaning cool.
The monitor made it look "fine — maybe a little flat." The scopes are specific: milky blacks, dull whites, and a blue cast. Three concrete problems, three concrete fixes.
Step 2 — Correct Camera A to neutral
We run the single-shot workflow in order — tone, then cast, then skin.
Tone (waveform). Lower lift until the darkest part of the frame (the shadow behind her) drops from ~12% down to touch ~2–4% — real black, without crushing. Raise gain until the collar's white reaches up near 95–100% — real white, without blowing. The image already looks less foggy.
Cast (parade + eyedropper). We grab the eyedropper and click the grey card we shot at the head of the clip — a guaranteed neutral. In one click, the blue drops and R, G, B snap into alignment. We fine-tune on temperature (a hair warmer) while watching the parade.
Skin (vectorscope). We check her face: the skin arm now rides the skin-tone line at her real, honest warmth. Done.
FIGURE CS31.7 — Camera A, AFTER: corrected to a clean neutral baseline
WAVEFORM PARADE (R G B on her white collar)
100%┤· · · ▓▓ · · · · · ← whites 100│▓▓ ▓▓ ▓▓ ← three channels
80%┤ ▓▓▓▓▓▓▓ reach up │▓▓ ▓▓ ▓▓ LEVEL at the top
60%┤ ▓▓▓▓▓▓▓▓▓▓ 50│▓▓ ▓▓ ▓▓
40%┤ ▓▓▓▓▓▓▓▓▓▓▓▓ │██ ██ ██ ← LEVEL at the bottom
20%┤▓▓▓▓▓▓▓▓▓▓ 0│██ ██ ██
3%┤▓▓ ← blacks reach down to ~3% ‾‾ ‾‾ ‾‾
0%┤· · · · · · · · · ·
Now: real black (~3%), real white (~100%), no cast (parade level on the collar),
skin on the skin-tone line. VECTORSCOPE: trace centered, skin arm on the line.
This shot is now the HERO. Everything else matches to it.
Camera A is now a clean, neutral, truthful baseline. It is not "graded" — no warm cozy bakery look yet — it is simply honest. That honesty is what makes it a reliable reference to match against.
Step 3 — Read Camera B against the hero
Now the match. We put Camera B's shot up next to A and read both sets of scopes together. The mismatch is exactly what we expected from two different cameras.
FIGURE CS31.8 — Camera B, BEFORE (vs. the corrected hero A)
WAVEFORM: B vs A PARADE: B's white collar
100%┤· · · · · · · · · · 100│▓▓ ██ ▓▓ ← red & green high,
90%┤ ▓▓▓▓▓ ← B's whites │▓▓ ██ ▓▓ blue a little low
70%┤ ▓▓▓▓▓▓▓▓ push to ~90% 50│▓▓ ██ ▓▓ → a WARM cast
50%┤▓▓▓▓▓▓▓▓▓ (B is BRIGHTER │██ ▓▓ ▓▓
20%┤▓▓▓▓▓▓ than A) 0│██ ▓▓ ▓▓
6%┤▓▓ ← B's blacks at ~6% ‾‾ ‾‾ ‾‾
0%┤· · · · · · · · · ·
Diagnosis vs A: B is (1) a bit BRIGHTER (blacks at 6% vs A's 3%, whites hotter),
and (2) WARMER (red/green up, blue down on the collar — the opposite of A's old
cast). VECTORSCOPE: B's skin sits farther along the skin-tone line (warmer/more
saturated) than A's. Cut A→B right now and her face jumps warmer and brighter.
Two cameras, same light, same face — and they disagree: B is brighter and warmer. This is normal and unavoidable, and it is precisely why matching exists. If we cut A to B now, the viewer sees her skin warm up and brighten on the cut, and the piece reads as amateur.
Step 4 — Match Camera B to Camera A
We match in the disciplined order: tone, then color, then skin. We are not correcting B to some ideal — we are matching it to A.
Tone first. Raise B's lift slightly so its blacks drop to ~3% to match A, and pull B's gain down a touch so its whites sit where A's do (~100%, not hotter). Now both waveforms have the same floor and ceiling. A surprising amount of the "mismatch" was just brightness — with tone matched, B already looks much closer.
Then color. On the parade, B reads warm (red/green high, blue low) relative to A's neutral. Nudge temperature cooler and adjust tint until B's collar shows the same level R, G, B that A's does. The parades now agree.
Then skin. The final, sensitive check. On the vectorscope, overlay or compare: B's skin arm should land at the same point on the skin-tone line as A's — same angle, same distance out. Nudge until the two faces are siblings.
FIGURE CS31.9 — Camera B, AFTER: matched to the hero
WAVEFORM: B now sits on A PARADE: B's collar now matches A's
100%┤· · · ▓▓ · · · · · 100│▓▓ ▓▓ ▓▓ ← level, like A
80%┤ ▓▓▓▓▓▓ │▓▓ ▓▓ ▓▓
60%┤ ▓▓▓▓▓▓▓▓▓ 50│▓▓ ▓▓ ▓▓
40%┤ ▓▓▓▓▓▓▓▓▓▓ │██ ██ ██
20%┤▓▓▓▓▓▓▓ 0│██ ██ ██
3%┤▓▓ ← blacks match A at ~3% ‾‾ ‾‾ ‾‾
0%┤· · · · · · · · · ·
B's floor, ceiling, and channel balance now match A's. VECTORSCOPE: both skin
arms land on the SAME point of the skin-tone line. Cut A→B→A and her face and
the room hold perfectly steady. The two cameras have become one.
We do the final test: play the assembled take at full speed and watch the cuts between A and B. Her skin holds steady. The wall behind her holds steady. The brightness holds steady. The two cameras have disappeared into one seamless conversation. That is a matched scene, and it is the difference between "two clips" and "a film."
⚠️ Common Mistake: matching color before matching brightness. The instinct is to attack the warm cast on B first, because it's the obvious difference. But B was also brighter than A, and brightness differences masquerade as color differences — a brighter shot reads as a different tone. If you chase the color first, you'll dial in a balance that stops being right the moment you fix the brightness. Match the waveform (black and white points) first, and the true color difference on the parade becomes clear and stable. Tone before color, always.
The edit pass
With both cameras corrected and matched, the cut itself is easy — which is the point. Correction done, the editing is about performance and rhythm, not fighting the color.
FIGURE CS31.10 — The testimonial cut: A and B intercut on one performance
V1 (Cam A, wide) [ intro: "I started this place..." #### ][ ...tag: "...every morning." ## ]
V1 (Cam B, tight) [ the heart: "...my grandmother's recipe." ##### ]
A1 (her lav audio) [ one continuous synced performance ############################### ]
A2 (room tone/bed) [ soft bakery ambience ................................................ ]
^open wide ^cut to tight on the emotional line ^back to wide to close
Because A and B are matched, we cut on her PERFORMANCE (to the tight shot for the
heartfelt line, back to wide to close) with zero color penalty. The correction is
what buys us the freedom to cut wherever the story wants.
We open on the wide (Camera A) to establish her and the shop, cut to the tight (Camera B) on the emotional line about her grandmother's recipe, and return to the wide to close. Because the two angles match, we cut purely on performance and rhythm — exactly where the story wants the cut — and pay no color price for it. This is the reward of correction: the freedom to edit for story, because the technical layer is solved and invisible.
Note what we have not done: we haven't graded. The piece is neutral and true. The warm, cozy bakery look — the amber that makes it feel like the smell of fresh bread — is the next chapter's job, and it will drop onto this matched baseline and land identically on both cameras precisely because we matched them first.
✂️ In the Edit: the grey card paid for itself. Look back at how fast Step 2 went: one click on the grey card and Camera A's cast was gone. That three-second habit on set turned a fiddly manual balance into a single click and gave us a rock-solid neutral reference to match Camera B against. The shoot did the correction a favor. This is you shoot for the edit made literal — and the lesson for your next shoot is exact: lock both cameras to the same manual white balance, match their picture profiles, and shoot a grey card. You will thank yourself at the color page every single time.
Discussion questions
- Both cameras saw the same window key and bounce, yet they still didn't match. Why is that normal, and why does it mean you can't rely on matched lighting alone to get matched footage?
- We matched brightness (waveform) before color (parade). Explain, with the specific example of Camera B being both brighter and warmer than A, why that order matters.
- We corrected Camera A to neutral, not to the warm bakery look, before matching B to it. Why match to a neutral hero rather than to a graded one?
- The grey card turned a manual balance into a one-click fix. What are two other things you could capture or lock on set that would make the correction pass even faster?
- Suppose you had shot Camera A on a Log profile and Camera B on a standard profile. Why would that make matching dramatically harder, and what's the on-set rule that prevents it?
- We cut on performance because the color was solved. Describe a piece you've made where color problems forced an editing decision (you avoided a cut because the shots didn't match). How would matching have freed that edit?
Your turn: shoot the mismatch, then fix it
Reproduce this case study with whatever two cameras you have — two phones, or one camera you reposition for a second synced-by-clap take.
The brief: shoot a 30-second, two-angle piece to camera of a person (a friend, yourself, a local business owner) under one window key. Then correct one angle to neutral and match the other, using the scopes, before/after.
Constraints and guidance:
- Lock both cameras to the same manual white balance and the same picture profile. Do not use auto. Shoot a grey card at the head of each.
- Roll both at once so it's one performance you can intercut.
- In post: correct the hero first (tone on the waveform, cast on the parade with the grey card, skin on the vectorscope), then match the second angle to it (tone → color → skin).
- Save before/after stills of your scopes for each camera, like the figures above. Seeing your own mismatch on the parade and then watching it snap into alignment is the moment this skill becomes yours.
- Test by cutting between the angles at speed. If her face or the wall jumps on any cut, you're not done matching. Chase it until the cut is invisible.
- Stop at neutral. No look yet — that's Chapter 32. The whole exercise is proving that correction and matching, alone, turn two mismatched cameras into one seamless piece.
Show it to someone and ask: "How many cameras was this shot on?" If they can't tell, you've matched it. That is the professional's invisible skill, and you now have it.
Key takeaways
- Two cameras never match out of the box — even under identical light, even the same model. The mismatch is normal and unavoidable; matching in correction is how you reconcile it.
- Lock manual white balance to the same number on both cameras, match picture profiles, and shoot a grey card. These three on-set habits do 90% of the matching before you open the editor and make the last 10% a one-click start.
- Correct the hero to neutral first, then match everything to it. A real reference shot beats an abstract ideal; the scene's internal consistency is what the viewer perceives.
- Match tone (waveform) before color (parade) before skin (vectorscope). Brightness differences masquerade as color differences; fixing tone first reveals the true cast and keeps skin honest.
- Matched footage buys editing freedom. When color is solved and invisible, you cut on performance and story, not around mismatched shots.
- Stop at neutral. Correction makes the piece true and consistent; the grade (next chapter) makes it feel like something — and it lands identically on both cameras because you matched them first.