An interview setup is four separate decisions, not one look. Eyeline, background, sound, and interviewer position, and which of them you can never fix later.

Category
Filmmaking
Author

Justkay
Documentary Filmmaker & Founder at Storyflow
Topics
2026-07-29
•
13 min read
•
FilmmakingTable of Contents
To set up an interview shoot, make four separate decisions before anything is unpacked. First, where the subject looks: into the lens if they are addressing the viewer directly, or slightly off-axis to the interviewer if the viewer should feel like they are overhearing a conversation. Second, what is behind them, which should carry depth and say something about who they are. Third, where the microphone goes, which is the decision that determines whether the material is usable at all. Fourth, where the interviewer sits, which fixes the eyeline and, if you are shooting more than one interview, decides whether the two can ever be cut together. Most guides on this subject are gear lists. Gear matters far less than the four decisions above, and gear is also the part that is easiest to fix, borrow, or work around. The decisions are not. Put simply: every other decision on an interview can be fixed later. Where they look cannot. I have shot documentary interviews for years, and the errors that actually cost films are never the ones people worry about. Nobody has lost a film to a slightly underexposed background. Plenty have discovered in the edit that their two key contributors are both looking camera-left, so the film can never cut between them as a conversation.
Full disclosure: Storyflow is our own product, so weigh its placement here with the skepticism you would apply to any tool a company recommends on its own blog. We rank it last, and honestly it is the least relevant tool on this page. An interview is decided by a microphone, a room, and where two people sit, and Storyflow touches none of that: it records nothing, transcribes nothing, and has essentially no role on the day. Its narrow contribution is before the shoot, making visible which parts of the story still have no material.
The real tools for an interview are a close microphone and a treated room. These cover the planning and processing around them, and no software substitutes for either.
| Tool | Best For | AI Features | Price |
|---|---|---|---|
A document or notes app | The question list | Varies | Free |
Transcription service | Searchable interviews | Core feature | Varies by volume |
StudioBinder or a template | Releases and paperwork | Limited | ~$29 mo |
Storyflow | Questions beside the research | Reads the whole board | Free / $7.99 mo |
Keep the research, the themes the film has to cover, and the question list on one board, so you walk into the room knowing which gaps this interview has to fill.

Into the lens. The subject addresses the viewer directly. It is confrontational in the useful sense, it creates intimacy and authority, and it works when the person is speaking to the audience. Achieving it comfortably usually needs an off-axis prompter device so the subject sees the interviewer's face while looking down the lens, because most people cannot talk naturally to a piece of glass.
Slightly off-axis, to the interviewer beside the camera. The viewer becomes an observer of a conversation. It is softer, it suits reflection and storytelling, and it is the documentary default. The angle should be small, roughly ten to twenty degrees off the lens axis. Wider than that and the subject reads as looking at something offscreen rather than at a person.
Why this is the decision that cannot be undone: it defines the viewer's relationship to the subject and it cannot be changed in post. Everything else about the frame is a matter of degree; this is a binary that reaches the audience directly.
Depth beats decoration. Distance between the subject and the background is what makes an image read as three-dimensional. A subject two metres in front of a bookshelf looks better than a subject pressed against a nicer bookshelf.
The background should say something. A workshop, a kitchen, a room they actually inhabit. Background is free characterisation, and a plain wall throws it away.
Avoid competing detail. Anything bright, high-contrast, or apparently growing out of their head. Check the frame edges, which is where the distracting object always is.
Consistency matters if you will intercut. Two interviews with the same person in different rooms cannot easily be cut as one continuous conversation.
This is the most recoverable decision. A weak background is a shame. Bad audio is a lost interview. Spend accordingly, which is the opposite of how most people allocate their attention on the day.
Get the microphone close. Almost every audio problem is a distance problem. A modest microphone close to the subject beats an expensive one far away, every time.
Use two sources if you can. A lavalier on the subject and a shotgun overhead, recorded separately. One is your backup for a rustle, a cough, or a battery failure, and you will use it.
Listen on headphones, continuously. Not at the start. Throughout. Refrigerator compressors start, air conditioning cycles, aircraft pass, and a phone in a pocket buzzes against a lavalier cable.
Record room tone. Thirty seconds of the empty room, everyone silent, at every location. It is what the editor uses to patch gaps invisibly, and it costs nothing except remembering.
Treat the room before you blame the microphone. Soft furnishings, curtains, rugs, and bodies absorb reflections. Empty hard-surfaced rooms sound like bathrooms regardless of what you record with.
This is the decision that determines whether you have an interview at all. Audiences tolerate a soft image. They will not sit through speech they have to work to hear.
As close to the lens as possible for a natural off-axis eyeline, without appearing in frame.
At the subject's eye height. An interviewer below eye level makes the subject look up and read as superior; above makes them look down and read as diminished. Both are unintended editorial statements.
Alternate sides between contributors, deliberately. This is the setup decision people miss, and it is the one that constrains the whole edit. If contributor A looks camera-left and contributor B also looks camera-left, the two can never be intercut as a conversation, because both appear to be addressing the same offscreen point. Sitting on opposite sides for alternating contributors makes them appear to face each other and lets the film construct a dialogue that never happened.
Keep the same position for the whole interview. Moving mid-session changes the eyeline and breaks continuity within the interview.
Lighting matters, and it matters less than the four decisions.
One good source plus fill will carry almost any interview. A large soft source slightly off to one side, and something bouncing light back into the shadow side.
Bigger and closer is softer. A large source close to the subject wraps around the face. The same source further away hardens.
Window light is free and excellent. A subject beside a large window with a bounce on the opposite side is a professional-looking interview at no cost. The catch is that it changes over the session as the sun moves, which is a continuity problem on a long interview.
Separate the subject from the background. A little light on the background, or a subtle rim on the subject, prevents them merging into a flat plane.
Do not light for beauty at the cost of comfort. A subject squinting into a hard key gives you a worse interview, and performance beats lighting quality every time.
Do a pre-interview, on the phone or in person, days earlier. Find out what their story actually is and which moments matter. This is where an interview is won. Arriving without one means using the session to discover what you should have known in advance, and the good material arrives after the camera is packed.
Do not over-rehearse. Knowing what they will say is the goal. Getting them to say it fully in the pre-interview is not, because people rarely deliver the same thing twice with the same freshness.
Send practical details in advance. Where, when, how long, what to wear (mid-tones with texture, no fine repeating patterns, no logos), and whether they need to bring anything.
Handle consent properly, before you roll. Explain what the project is, where it will be shown, what will happen to the material, and what happens if they later change their mind. Get the release signed. A contributor who did not understand what they agreed to is a problem for them and a risk to the film, and doing this on the day rather than in the edit is the only honest sequence.
Arrive early enough to set up without an audience. Being watched while you fumble with a light stand does not help the person about to be vulnerable on camera.
The part that decides whether the setup was worth it.
Ask them to answer in complete sentences, incorporating the question. "I started the business in 1998" rather than "1998." Without this, every answer needs the question audible, which constrains the edit permanently.
Ask one thing at a time. Double questions get one answer, usually to the second half.
Then stop talking. The most useful technique available. After they finish, wait. The pause is uncomfortable and it is where the real answer arrives, because people fill silence with the thing they were not sure they should say.
Do not verbally agree. No "mmm" or "right" over their answers. It is instinctive, it makes the audio unusable, and a nod does the same job silently.
Ask for the story, not the summary. "What happened that morning" produces usable material. "How did you feel about that period" produces abstraction.
Return to the important question later, differently phrased. The second answer is often better, and you can use either.
Ask the obvious things you already know. The interview must stand alone for a viewer who was not in the room and has not read the research.
End by asking what you did not ask. It regularly produces the best moment in the session.
Worth isolating, because it is the one setup error that cannot be repaired and it only becomes visible weeks later.
A film with several contributors usually wants, at some point, to cut between two of them as if they were in conversation: she makes a claim, he responds to it, back to her. That construction only reads if the two appear to be facing each other.
| Contributor | Interviewer sits | Subject appears to look | Can they be intercut? |
|---|---|---|---|
A | Camera left | Camera left | Reference |
B | Camera left | Camera left | No. Both address the same offscreen point |
C | Camera right | Camera right | Yes, reads as facing A |
D | To-camera | Down the lens | No. Addresses the viewer, not another contributor |
The rule is simple: alternate the interviewer's side across contributors who might ever be cut together. On a film with six contributors, sit camera-left for the odd-numbered ones and camera-right for the even-numbered ones, unless you have a specific reason to do otherwise.
Two refinements from practice. Keep the two most important contributors on opposing sides, because those are the ones most likely to need to appear in dialogue. And if a contributor is deliberately shot to-camera, accept that they now sit outside the conversational grammar of the film entirely: they address the audience, which is a strong effect and one that cannot be mixed with overhearing in the same sequence.
This is the same constraint described in the spatial half of continuity, arriving weeks earlier and costing far more, because an eyeline established at the interview cannot be changed at any later point.
Two cameras solve the jump-cut problem. A wide and a tight on the same subject let you cut within an answer invisibly, which is the single biggest practical advantage available. If the second camera is affordable, take it.
Offset the framings clearly. Two similar sizes cut together as a jump. Make one noticeably wider.
With one camera, plan for cutaways. You will remove pauses and digressions, and every removal needs covering, which means shooting inserts and detail afterwards is not optional.
Shoot the cutaways before they leave. Hands, the room, objects they mentioned, them listening. Five minutes at the end saves the sequence.
Do not reframe during an answer. Zooming or repositioning mid-answer removes your ability to use that section cleanly.
Every contributor looking the same direction. The film can never intercut them as a conversation.
Distant microphone. The origin of nearly all bad interview audio.
No room tone. The editor cannot patch anything invisibly.
Interviewer sitting too far from the lens. Produces the odd, disengaged wide eyeline.
Talking over answers. Renders otherwise perfect material unusable.
No pre-interview. The session becomes discovery, and the good material arrives after the camera is off.
Answers that do not include the question. Every clip becomes dependent on hearing the interviewer.
Consent handled casually. An ethical failure and a legal exposure.
The real tools here are a close microphone and a treated room. These support the planning and processing around them.
Best for the question list and structure: a plain document or notes app. The question list is a document, it needs to be readable on a phone or a printed page in the room, and nothing beats a simple list. Free.
Best for transcription afterwards: an automated transcription service. For any interview-led project this is the highest-value post-production purchase available, because it makes hours of speech searchable and turns the transcript into the working surface for structure. Pricing varies by service and volume (verify current pricing).
Best for production paperwork and releases: StudioBinder or a standard release template. Release forms, call sheets, and contributor details in one place. Free tier, paid from around $29/month (verify current pricing).
Best for holding the questions beside the story: Storyflow. The canvas where the research, the themes you need to cover, and the question list sit together, so you can see which parts of the story still have no material. Free plan is $0; Plus is $7.99/month billed annually. The honest limit: it records nothing, transcribes nothing, and has no role on the day beyond being a document you might read.
| Tool | Best for | Used on the day | Free tier | Starting price |
|---|---|---|---|---|
A document or notes app | The question list itself | Yes | Yes | Free |
Transcription service | Making the interview searchable | No | Varies | Varies by volume |
StudioBinder or a template | Releases and paperwork | Yes | Yes | ~$29/mo |
Storyflow | Questions beside the research | Rarely | Yes, unlimited boards | $7.99/mo annual |
Pricing checked July 2026. Competitor pricing changes often and varies by plan and volume, so verify on each vendor's page before quoting it.
An interview setup is not a look you copy. It is four decisions, and they are not equally important or equally recoverable.
Sound decides whether you have an interview. Eyeline decides the viewer's relationship to the person speaking, and it decides whether two contributors can ever be cut as a conversation. Background is characterisation and depth, and it is the one everybody spends their attention on despite being the easiest to live with when it is imperfect.
Get a microphone close, sit next to the lens, alternate your side between contributors, record room tone, and then stop talking after each answer. Every other decision on an interview can be fixed later. Where they look cannot.
Make four decisions before unpacking: where the subject looks (into the lens to address the viewer, or slightly off-axis to the interviewer so the viewer overhears), what is behind them and how far away it is, where the microphone sits, and where the interviewer sits. Then light with one soft source and a fill, record room tone, and confirm consent before rolling.
It depends on the relationship you want. Looking into the lens addresses the viewer directly and creates intimacy and authority, and usually requires a prompter device to be comfortable. Looking slightly off-axis to the interviewer makes the viewer an observer of a conversation, which is the documentary default. The off-axis angle should stay within roughly ten to twenty degrees of the lens.
Get the microphone close, because almost every audio problem is a distance problem. Use two sources where possible, typically a lavalier plus an overhead shotgun recorded separately, monitor on headphones throughout rather than only at the start, treat the room with soft furnishings, and record thirty seconds of room tone before you leave.
Room tone is the sound of the empty location with nobody speaking, recorded for about thirty seconds. Editors use it to fill gaps, smooth joins, and patch removed sounds so the silence between words matches the rest of the interview. Without it, every edit in the audio is audible as a change in the background.
As close to the lens as possible without being in frame, at the subject's eye height. For multi-contributor projects, alternate which side of the camera you sit on between contributors, so their eyelines oppose and the film can intercut them as a conversation later.
One large soft source slightly off-axis, with fill bouncing light into the shadow side. Bigger and closer is softer. A window with a bounce opposite is excellent and free, though the light will change over a long session. Add a little separation between subject and background so they do not merge, and never light so brightly that the subject is squinting.
Do not talk over answers, even with sounds of agreement, because it makes the audio unusable. Do not ask two questions at once. Do not fill the silence after an answer, because that pause is where the best material arrives. Do not reframe or zoom during an answer. And do not skip the consent conversation.
Because unless the answer contains the question, every clip depends on the interviewer's voice being audible, which permanently constrains the edit. "I started the business in 1998" can stand alone; "1998" cannot. Ask for this explicitly at the start, and remind them gently once or twice if it slips.
A conversation days before the shoot, often by phone, to find out what the person's story actually is and which moments matter. It is where the interview is won, because it lets you arrive with the right questions rather than using the session to discover them. Avoid over-rehearsing, since people rarely deliver the same thing twice as freshly.
Not necessarily, but two solve the biggest practical problem: cutting within an answer without a visible jump. If you use two, make the framings clearly different in size. With one camera, plan cutaway material and shoot it before the subject leaves, because every pause you remove will need covering.
Only in preparation. Keeping the research, the themes the film needs to cover, and the question list on one canvas makes it visible which parts of the story still have no material. Free to start.
Three places, and they are most of the task. It records nothing, so camera and audio are entirely elsewhere. It does not transcribe, which is the highest-value processing step for interview material. And it has essentially no role on the day, when what you need is a printed question list, headphones, and a close microphone.
Every Storyflow board starts from real structure and an AI that reads the whole canvas. Open one of these templates and make it yours.
A visual AI workspace where every feature lives inside one canvas. No tab-switching, no context lost.
Build your entire board from a single message
Type what you need in the AI chat at the bottom of your canvas. The AI adds cards, headings, and structure directly onto your board.
Use expert frameworks as AI context
Type @ in the AI chat and choose any Tactic. The AI tailors every response to that framework instead of giving generic advice.
Turn your board into a mind map in seconds
Ask the AI to restructure your canvas as a mindmap. It connects your ideas into a visual hierarchy so you can see how everything relates.
Storyflow actually began as a personal tool while working on creative and research projects.
We kept running into the same problem: ideas were scattered everywhere: notes, documents, and whiteboards.
Nothing helped us see how everything connected.
So we started building a workspace designed around how ideas actually grow.
→ Read how Storyflow was created
Justkay
Documentary Filmmaker & Founder at Storyflow
Published: 2026-07-29
Transform your creative workflow with AI-powered tools. Generate ideas, create content, and boost your productivity in minutes instead of hours.
Ask Storyflow to