The one-page reference to screenshot before an interview shoot. Definitions, the setup checklist, the question-design guide, the two-mic-safety diagram, and the "when to use what" decisions — dense and skimmable.
The five terms, defined
Term
Definition
The interview setup
The repeatable physical arrangement of subject, camera, interviewer, light, and mic — framing, eyeline, and interviewer position — built before the conversation starts.
Off-axis eyeline
The subject looks slightly off the lens axis, at an interviewer beside the camera, so the viewer becomes an observer of a real conversation (vs. direct-to-lens address).
Question design
Writing and sequencing questions so they produce usable, story-carrying answers — open rather than closed, specific rather than vague, phrased to invite a full sentence.
The non-answer
A reply that can't stand alone in the edit — a yes/no, a fragment, an echo of the question, an "it depends," or an evasion. The raw material of mush.
Active listening
Genuinely attending to the current answer (not the next question) so you can follow the thread, ask the right follow-up, and use silence to reach the real answer.
The interview-setup checklist (build every time)
✓
Element
Default
Fix if wrong
☐
Frame
Medium / medium-close-up, subject on a third
Too wide → face too small; too tight → lose shoulders/hands
☐
Eye height
Lens at subject's eye level
Too low → up-the-nose; too high → looking down on them
☐
Eyeline
Off-axis, small angle
Move in beside the lens if they look lost/profile
☐
Interviewer
Seated as close to the lens as possible
Too far → wide eyeline reads as evasive
☐
Looking room
More space on the side they face
Centered = mugshot; edge-facing = cramped
☐
Background
Deep, soft, darker than subject; nothing growing from head
Wall at the head = flat "hostage video"
☐
Key light
Soft, on the looking side, just above eye level
Window/lamp behind them = silhouette
☐
Audio
Two mics, two channels, headphones
One mic = single point of failure
☐
Room tone
30 s recorded at wrap
No tone = no way to patch edits
☐
Release
Signed before subject leaves
No release = can't use the footage (Ch.38)
Lighting the interview (apply Part III)
Choice
Setting
Tone it gives
Key side
The side the subject faces (looking side)
Motivated, natural modeling
Key quality
Soft, just above eye level, feathered down
Flattering, dimensional
Key : fill ratio
2:1–3:1
Warm documentary
~1:1
Bright, clean, corporate/testimonial
4:1+
Weighty, uneasy — only if earned
Separation
Rim/hair light + darker, deeper background
Three-dimensional, not pasted-on
No-budget path
Window on the looking side + white bounce fill
Free, soft, professional
Two-mic safety (the one non-negotiable)
(( LAV )) close under collar ───────────────► CH 1 ─┐
├──► RECORDER ──► two independent
(( BOOM )) just above frame, at mouth ────────► CH 2 ─┘ takes of the words
|
monitor BOTH on 🎧 headphones
peaks ~-12 dBFS (never touch 0 dBFS = clipping = unfixable)
+30 s ROOM TONE at wrap
Lav fails on a line (rustle/dropout)? → editor uses the boom for that line.
Boom drifts/reflects? → editor uses the lav.
Both fail on the same word, same way? → almost never. That's the "safety."
Why a lav + boom is a safe pair, not a redundant-but-useless one: they fail in different, uncorrelated ways, so both rarely fail on the same sentence.
Question design: the open/specific/quiet guide
Principle
Do
Instead of
Open
"Walk me through what happened."
"Did that happen?" (yes/no)
Specific
"Tell me about the day you..."
"What does X mean to you?" (abstract)
Quiet (short)
One question, then stop
Three stacked questions in a breath
Quiet (clean)
Nod silently; kill the "mm-hmm"s
Verbal encouragements over their audio
Neutral
"How did you feel?"
"You were devastated, right?" (leading)
Rescue
"Can you give me an example?"
Accepting a platitude and moving on
Sequence
Easy → hard (middle) → soft close
Hard emotional question first, cold
The magic close
"Anything I didn't ask that I should have?"
Ending on your last listed question
Getting the answer in a full sentence
The problem
The move
The question is cut in the edit, so fragments answer nothing
Brief up front: "answer in full sentences that fold in my question"
A great thought comes out tangled
Re-ask: "That was perfect — say it once more as a single sentence."
The prepared answer is the shallow one
Hold 3 seconds of silence — they fill it with the truer line
You need clean cuts
Leave air after answers; don't talk over the tail
The threshold idea: in the edit, the question disappears — so the answer must stand alone. "Twenty years" is useless; "I've been doing this for twenty years" is a soundbite.
One camera vs. two cameras
Single-camera
Two-camera
Hide internal cuts with
B-roll / 4K punch-in / reframe past 30° between Qs
Cut A↔B (invisible, no B-roll)
Requires
Enough B-roll (Ch.20); 30° rule when reframing
Same side of line, ≥30° apart + a size step; matched WB/exposure/height; sync
Best for
Solo, run-and-gun, phone-first
Paid work, long interviews, minimal B-roll
The trade
Spend effort later (in B-roll)
Spend effort now (second camera)
Top common mistakes → fixes
Mistake
Fix
Stage
Interviewer parked far from the lens
Sit right beside the lens; small off-axis angle
Setup
Window/lamp behind the subject
Brightest source in front, on looking side
Pre / setup
One mic, unmonitored
Two mics, two channels, headphones on
Setup
Clipped audio (peaks at 0 dBFS)
Leave headroom; peaks ~-12 dBFS
Setup
Closed / leading questions
Open, neutral, specific phrasing
Pre
Talking over the answer's tail
Nod silently; leave clean air
On set
Filling the silence
Hold 3 seconds; let them fill it
On set
Fragment answers
Brief the full-sentence rule; re-ask
On set
Single-camera with no B-roll plan
Punch-in / reframe / commit to B-roll (Ch.20)
Pre / set
No room tone / no release
Record 30 s tone; sign release before wrap
Wrap
The craft principles at work
Story is the boss. The eyeline (off-axis vs. into-lens) and the lighting ratio are chosen for what you want the viewer to feel, not by rule.
Sound is half the picture. Two-mic safety exists because a great-looking interview with failed audio is worthless; the audio is the half most likely to sink you.
You shoot for the edit. Full-sentence answers, clean air, and a hideable-cut strategy are all you doing the editor's job on set (usually the editor is you).
Fix it in pre, not in post. Where the subject sits — for light and for sound — solves more than any post tool. Solve the room before the camera.
Motivate every choice. Key from the looking side because it reads as coming from the conversation or the window; break the off-axis default only on purpose (see Case Study 1).
Ethics & accessibility (non-optional)
Informed consent + release. The subject should know what the piece is and that they can pause or stop; get the model release signed before they leave (Ch.38).
Handle sensitive subjects with care. A person reliving something painful is doing you a favor; their wellbeing outranks the shot. If they ask to stop, you stop.
Caption every interview (Ch.36) — clean two-mic audio is what makes accurate captions possible. Choose an accessible location; offer questions ahead if it helps a nervous subject; never treat someone's voice or accent as something to "fix."
Honest editing. Reaction shots (noddies) may bridge a cut, never fabricate a false reaction.
Project increment (Production Checkpoint)
Project 2 — shoot your documentary's interview. Clear four bars: (1) framing on a third with an off-axis eyeline, eye height; (2) motivated soft light on the looking side with a controlled ratio and background separation; (3) two-mic-safe audio (lav + boom, two channels, headphones, ~-12 dBFS) plus 30 s room tone; (4) full-sentence answers from a designed, open/specific/quiet question list run with active listening and the silence technique. Get the release signed; note single- vs. two-camera so you know how much B-roll you owe (Ch.20). This sit-down is the spine your whole documentary hangs on (assembled in Ch.30).