Case Study 2: Cutting "The Beekeeper" — an Interview Edit From Transcript to Lock
Case Study 1 took apart a famous film. This one builds a small one, from scratch, at the timeline — so you can watch the whole Chapter 30 method run once, end to end, on ordinary footage a phone could have shot. We are going to edit a three-minute documentary profile of a beekeeper (a talking-head subject, one of the book's four recurring setups) from a raw interview and a card of B-roll, following the pipeline exactly: transcript → radio edit → selects → string-out → spine → B-roll layer → pace → fine cut → lock. Every stage gets a timeline diagram, because interview editing only makes sense when you can see what each move does to the tracks.
There are no named characters here and no magic footage — just the disciplined method. Follow along with your own Project 2 footage open in another window; by the end you will have done the Production Checkpoint.
The brief and the constraints
The piece: a 3-minute documentary short profiling a small-scale beekeeper — a person who keeps a dozen hives on the edge of town, harvests and sells honey, and has a quiet, deeply-felt reason for doing it. The goal: a viewer who has never thought twice about bees should, by the end, understand this person and feel something about the vanishing, patient craft they practice.
What we are starting from (shot across Chapters 19–20, ingested and synced in Chapter 27):
- The A-roll: one single-camera interview, about fourteen minutes, shot as a talking head by the hives — subject on a third, off-axis eyeline, soft window-and-sky key, two-mic-safe audio (lav + shotgun), full-sentence answers because the subject was briefed (Chapter 19). Because it is single-camera, every internal trim will jump — we will lean hard on B-roll.
- The B-roll: about forty minutes, shot against a coverage plan (Chapter 20): sequences of the subject opening a hive, pulling a frame, smoking the bees, spinning the extractor, jarring the honey; inserts of hands, bees on comb, honey threading; cutaways of the apiary, the sign, the truck, the hills; plus wild sound on everything and sixty seconds of room tone.
- The constraint: solo edit, modest laptop, free editor. The 4K footage stutters, so we generated proxies (Chapter 27) and are cutting on them; we will relink to originals at export (Chapter 36).
⚙️ Settings Box: the interview-edit project (a starting point).
Setting Choice Why Timeline tracks V1 interview, V2 B-roll (above), A1 interview audio, A2 wild sound, A3 (later) music B-roll covers V1 from above; the voice on A1 stays on top of the mix Proxies On, cutting on half-res proxies 4K stutters on the laptop; relink to originals at export (Ch.27, Ch.36) Transcript Auto-transcribed, timecoded The raw material of the radio edit — and the caption file later (Ch.36) Selects bin One bin, labeled by content Turns footage into a deck of soundbite cards (§30.2) Interview audio level Clear and on top; wild sound and music dip under it The spine is the voice — nothing may fight it Loudness targets in LUFS and the real mix are audio post (Chapter 33). Here we cut for story, not for a finished sound.
Phase 1 — The transcript and the radio edit
We do not open the interview and start scrubbing. We open the transcript and start reading, highlighter in hand. Fourteen minutes of talking is a few pages of text, and reading is roughly ten times faster than watching. We mark every self-contained keeper and cross out every non-answer, tangent, and fragment. Here is a slice of the marked transcript:
FIGURE CS2.1 — The marked transcript (read, highlight keepers, cut the mush)
TC TRANSCRIPT (the QUARRY) KEEP?
------ ---------------------------------------------------------------- -----
00:20 "So, um, how did I get into it... my grandfather kept bees..." ✓ (origin — trim the "um")
01:05 "It's hard to explain to people who don't..." (vague) ✗
02:40 "You can't rush bees. They do everything on their own time, ✓★ (the THEME — strong)
and you learn to slow down to theirs."
03:55 [long tangent about equipment brands] ✗
05:30 "Every third bite of food you eat, roughly, needs a pollinator. ✓ (the stakes / facts)
People don't know that."
07:10 "I lost four hives over one winter. I sat in the truck and ✓★ (the low point — emotional)
I cried, honestly."
09:45 "It's not really about the honey. The honey's a bonus." ✓★ (the turn — reframes it all)
11:20 "Would I do it again? Every single year, I do it again." ✓ (the button — closes it)
12:30 "Does that make sense? Sorry, I'm rambling." ✗ (self-talk, never usable)
Ratio on this page: ~5 keepers to ~4 cuts — normal. Most of any interview is throat-clearing.
The three ★ lines are the emotional spine; the film will be built to deliver them.
Now the radio edit itself: we write each keeper on a card and deal them into story order — not the order the subject spoke. The subject opened with "my grandfather kept bees," but that is not our hook. Watch what we choose instead:
FIGURE CS2.2 — Dealing the cards: chronological order vs. STORY order
AS SPOKEN (chronological) → STORY ORDER (the radio edit's spine)
-------------------------------- --------------------------------------------------
1. grandfather kept bees (00:20) 1. HOOK: "You can't rush bees..." (02:40) ← said in the middle,
2. can't rush bees (02:40) but it's the thesis — it goes FIRST
3. every third bite (05:30) 2. ORIGIN: "my grandfather kept bees" (00:20)
4. lost four hives / cried (07:10) 3. STAKES: "every third bite of food..." (05:30)
5. not about the honey (09:45) 4. LOW: "I lost four hives... I cried" (07:10)
6. do it again (11:20) 5. TURN: "It's not really about the honey" (09:45)
6. BUTTON: "Every single year, I do it again" (11:20)
The subject never told it this way. The HOOK ("you can't rush bees") was buried at 02:40;
dragging it to the front gives the film a thesis in the first six seconds. The LOW POINT
(the cry) sets up the TURN ("not about the honey"), which sets up the BUTTON. That shape —
thesis → who → stakes → low → turn → resolution — is Chapter 17's arc, built from real sentences.
That reorder is the film. We built a three-act shape — a person, a threat, a low point, a reframe, a resolution — out of sentences the subject scattered across fourteen minutes. And we did it on paper, before touching a single frame of picture, exactly as §30.1 demands. If this order didn't work, fixing it here costs thirty seconds; discovering it was wrong after we'd laid in B-roll would cost an afternoon.
What we adjusted: our first pass put "every third bite of food" (the pollinator fact) as the hook, because it is a striking statistic. Reading the spine aloud, it felt like a public-service announcement, not a person. We demoted it to "stakes" in the middle and promoted "you can't rush bees" to the hook — because a film about a person should open on their voice and their philosophy, not a fact. The fact serves the person; the person is the boss.
Phase 2 — Selects and the soundbite log
Now we cross back to the timeline and pull the six keeper soundbites as selects — each one trimmed on the breath, in and out clean, labeled by content:
FIGURE CS2.3 — The soundbite log (selects, labeled by what they SAY)
LABEL | SOURCE TC | LEN | ★ | JOB
-------------------------+-------------+-----+-----+------------------------------
SB_cant-rush-bees | 02:40–02:49 | 9s | ★★★ | HOOK — the thesis
SB_grandfather | 00:20–00:31 | 11s | ★★ | origin / who they are
SB_third-bite | 05:30–05:39 | 9s | ★★ | stakes / the facts
SB_lost-four-hives | 07:10–07:22 | 12s | ★★★ | the low point (emotional)
SB_not-about-honey | 09:45–09:52 | 7s | ★★★ | the turn — reframes it all
SB_do-it-again | 11:20–11:26 | 6s | ★★ | the button — closes the film
-------------------------+-------------+-----+-----+------------------------------
Total spoken runtime of selects: ~54s. That's our audio spine — the film hangs on these six.
Six soundbites, about fifty-four seconds of talking. That is the whole spine of a three-minute film — the rest of the runtime will be B-roll and breathing room over these voices. Notice we did not keep the pollinator tangent, the equipment brands, or the lovely-but-vague "hard to explain" line. Ruthlessness at the select stage (§30.2) is what keeps the film from becoming a twelve-minute lecture.
Phase 3 — The radio string-out
We lay the six selects end to end, in spine order, audio only — the string-out — and we run the test the whole method is named for: we press play, close our eyes, and listen.
FIGURE CS2.4 — The radio string-out (audio only — CLOSE YOUR EYES and listen)
V1 [head:rush][head:grandfa][head:third-bite][head:lost-hives][head:not-honey][head:again]
A1 ["can't rush"]["grandfather"]["third bite"]["lost four hives...cried"]["not the honey"]["again"]
└──────────────── ~54 seconds, in story order, no B-roll, no music ────────────────┘
EYES-CLOSED VERDICT: Does it hold as radio? → Mostly. The arc is there and the low-point-to-turn
lands. BUT: the jump from "grandfather" straight to "every third bite" feels abrupt — we go from a
warm personal origin to a cold statistic with no bridge. FIX (in audio): swap their order? No —
instead we'll let a beat of wild sound and B-roll breathe between them in Phase 5, and trim the
head of "third-bite" so it lands softer. The spine is sound; note the rough edge and move on.
The eyes-closed pass did its job: it caught a rough seam (origin → statistic) that we would not have felt while distracted by picture. We note the fix and — crucially — we do not start laying B-roll to paper over a story problem. If the spine had a real hole (a missing beat, a confusing order), we would fix it here, in the audio, before spending an hour on picture. It holds well enough. We proceed.
Phase 4 — Building the spine
We commit the order, cut the questions out entirely (the viewer never hears the interviewer), and tighten within the soundbites — honestly. Two examples of the tightening, and one line we refused to cross:
- We trimmed the "um" off the front of "grandfather" and snipped a redundant clause out of the middle of "lost four hives." Each snip creates a jump on the picture — noted, to be covered.
- We did NOT frankenbite. The subject said "I lost four hives over one winter" and, separately, "I sat in the truck and I cried." We joined them — but they were about the same event, spoken seconds apart, so the join is honest: it says exactly what the subject meant. We were tempted to cut "the honey's a bonus" onto the end of the low-point line to speed things up, but that would have implied the cry was about money, which it wasn't. We left it where it belonged. (§30.3's test: would the subject agree that's what they meant? On the honest join, yes; on the tempting one, no.)
FIGURE CS2.5 — The spine, committed — done as audio, deliberately BROKEN as picture
V1 [rush]|[grandfa]|[third]|[lost-hives]|[not-honey]|[again]
A1 ["can't rush bees"]|["grandfather kept bees"]|["third bite"]|["lost 4 hives // I cried"]|["not the honey"]|["again"]
^ ^ ^ ^ (internal snip) ^ ^
every | is a splice between two different moments — the PICTURE JUMPS at all of them,
PLUS the internal "//" snip inside the low-point bite jumps too.
AUDIO: a finished 54-second story. PICTURE: a talking head that jumps ~7 times. That is CORRECT.
The spine is supposed to be broken as picture. Phase 5 covers every seam.
This is the moment interview editing feels wrong to a beginner and right to a professional. The audio is a finished, moving little story; the picture is a mess of jump cuts. We did not fail — we built exactly what §30.3 describes: a spine that is done as audio and broken as picture, with the seams left deliberately naked because we are about to cover every one of them.
Phase 5 — Layering the B-roll
Now the film comes alive. We lay the Chapter 20 B-roll onto V2, above the interview, matched to each beat, so the interview audio (A1) runs unbroken while the picture roams and every seam disappears. This is the B-roll-driven narrative of §30.4, built for real:
FIGURE CS2.6 — The covered spine: B-roll over the interview (voice leads, picture proves)
V2 ....[ B-roll: hands lifting a frame, bees ][ B-roll: apiary + hills ][seq: extractor spinning]....[ jar ]
V1 [FACE]|__(picture cut away — seams hidden)__|[FACE]|________________|[FACE]|__(covered)__|[ FACE ]
A1 [ SPINE, unbroken: "can't rush bees // grandfather // third bite // lost hives, cried // not the honey // again" ]
A2 [ wild sound: bees humming, the smoker puffing, the extractor whir, wind, birds ...................... ]
A3 [ ............ (temp music bed, low, enters on the turn) ........................................... ]
^open on ^prove "can't rush" ^stakes: ^RETURN to face ^prove "not ^button
the face (slow B-roll of bees apiary/hills for the CRY — the honey" — on the
(hook) on comb — patience) (scale) must be SEEN the jars, the FACE
said gift
Every seam from FIG CS2.5 now lives UNDER B-roll — all seven jumps hidden. The audio never stops.
The bees-on-comb over "can't rush" both HIDES a cut AND proves the line. One shot, two jobs.
Study the matching. Over "you can't rush bees," we lay slow, patient B-roll of bees moving on the comb — it hides the opening seam and proves the philosophy (show-don't-tell). Over "every third bite of food," we lay the wide of the apiary and the hills, giving the statistic scale and, incidentally, the breathing beat the radio test told us this bridge needed. Over "not really about the honey," we lay the jars and a shot of the subject handing honey to a neighbor — the picture reframes the line from product to gift. Every piece of B-roll is doing two jobs: hiding a cut and proving a point. That is why the matched coverage plan from Chapter 20 was worth so much more than pretty footage.
What we adjusted: our first B-roll pass covered everything, including the low-point line "I sat in the truck and I cried." Watching it back, the cry had no weight — we were looking at bees while the most emotional line of the film played. We pulled the B-roll off that line and let the face carry it. Which is Phase 6.
Phase 6 — Pacing: reveal and return
The covered spine hides every seam, but a first B-roll pass almost always over-covers, burying the person. We now do the reveal-and-return pass (§30.5): open on the face, roam the B-roll, and return to the face for the lines whose power is in the expression.
FIGURE CS2.7 — Pacing: which lines stay on the FACE (reveal → roam → return)
BEAT ON FACE or B-ROLL? WHY
----------------------- --------------------- --------------------------------------------
"you can't rush bees" FACE (open), then Meet the person first; hand to B-roll on the tail
B-roll on the tail
"grandfather kept bees" B-ROLL (old-hands seq) Warm illustration; the line is informational
"every third bite" B-ROLL (apiary/hills) A fact — give it scale, let the person rest
"I lost four hives... FACE — HELD The CRY. Its power is the expression. Never cover
I cried" through the pause this. Hold through the 2-second pause after "cried."
"not about the honey" B-ROLL (jars → gift) The turn — the picture reframes it (product→gift)
"I do it again" FACE (the button) End on the human, not on a cutaway
Two moments live on the FACE: the CRY (the emotional low) and the BUTTON (the resolution).
Everything else roams. We open on the face, return for the cry, breathe, and close on the face.
THE EXHALE: after "not about the honey," we hold 2s of quiet B-roll (a bee landing) before the button.
The single most important edit in the whole piece is the one we undid: we took the B-roll off "I cried" and let the subject's face — holding the two-second pause the interviewer earned by not filling the silence (Chapter 19) — carry the emotional low of the film. If we had left the bees over it, the film would have been slick and hollow, a beekeeping infomercial. By returning to the face for that one line, it becomes a portrait of a person. That is the difference between wall-to-wall B-roll and a paced film, and it is a choice, not an accident of coverage.
✂️ In the Edit. Notice that we could only make this choice because the interview was shot to allow it: the subject was framed and lit well enough (Chapter 19) that holding on the face for eight seconds works — the light models the face, the eyeline is right, the emotion reads. If the interview had been a flat, badly-lit hostage frame, holding on it would have been unbearable and we'd have been forced to bury it under B-roll, losing the moment. The pacing freedom we have in the edit was bought on the shoot. Shoot the interview so the face is worth holding, and the edit can return to it when it matters.
Phase 7 — Fine cut, the two passes, and lock
We tighten the whole thing — words and pictures together (§30.6). We enter every B-roll shot late and leave early; we trim a soundbite that repeated the one before it; we pull two cutaways that overstayed. The piece drops from a loose 3:25 to a tight 3:00, and every trimmed frame makes it stronger — except the cry and the button and the one exhale, which we protect at full length. Then we run the two passes:
FIGURE CS2.8 — The two quality passes on the locked cut
PASS 1 — EYES CLOSED (the radio test): listen only, whole film.
VERDICT: story holds — thesis, person, stakes, low, turn, button all land as audio. PASS.
(Caught one thing: a music swell was stepping on the word "cried." Dipped the bed under it.)
PASS 2 — SOUND OFF (the mute test for picture): watch only, muted, whole film.
VERDICT: the picture track tells a real story on its own — bees, hives, the person, the jars,
the gift. A sound-off / caption viewer sees a film, not a static head. PASS.
(Caught one thing: a 4-second stretch of only the talking head with no B-roll — added a cutaway.)
BOTH PASSES CLEAN → the piece serves the ear alone AND the eye alone. → READY TO LOCK.
PICTURE LOCK: freeze the cut. Hand off to → color (Ch.31–32), audio post + music license
(Ch.33, Ch.38), lower-third for the subject's name (Ch.34), captions from the transcript (Ch.36).
Both passes clean — the story holds with the eyes closed and shows with the sound off — so we lock the picture. We export a reference, note the two hand-offs the finish will need (a name lower-third; the transcript trimmed to a caption file), and we stop. The storytelling is done. Project 2 — "The Beekeeper," three minutes, built from fifty-four seconds of scattered soundbites and a card of matched B-roll — is a film.
Here is the finished shot the whole edit was built to deliver, rendered as a Described Shot so you can see how the shoot and the cut met on the one moment that matters most:
FIGURE CS2.9 — "The line we did not cover" [constructed teaching example]
THE FRAME Medium close-up, the beekeeper on the right third, veil pushed back off the face, the hives
and the hills falling soft behind at f/2.8. We are close enough to read the eyes.
THE MOVE Locked off. After a film of roaming B-roll, the stillness of the held face is a change of
gear the viewer feels — everything stops.
THE LIGHT Soft sky-and-window key from camera-left (the looking side, Chapter 19); the far cheek a stop
down; the late-afternoon warmth on the skin.
THE SOUND The spine, on their own voice: "...I lost four hives over one winter." A two-second pause —
only wild sound, a single bee, the wind. Then, quietly: "I sat in the truck and I cried."
The music bed sits low and does not swell over the word.
THE CUT We came to the face off a wide of the empty hives (which proved the loss) and we HOLD —
through the line, through the pause, through the admission — and only cut away, to a quiet
bee landing on the comb, once the sentence is fully out. The exhale.
THE EFFECT Because we held the face and refused to cover it, the viewer sees a real person admit a real
grief, and the held pause makes them lean in. The one moment we did NOT hide is the one they
remember.
THE LESSON B-roll hides the cuts you want hidden; it must never hide the face when the face is the
story. The most powerful edit in an interview piece is often the cutaway you choose NOT to make.
Discussion questions
- The radio edit moved "you can't rush bees" from the middle of the interview to the very front. What did that reorder buy the film, and why is "the subject's chronology is irrelevant" a liberating idea rather than a dishonest one?
- In Phase 3, the eyes-closed pass caught a rough seam (warm origin → cold statistic). The editor chose not to fix it by reordering, but by letting B-roll and wild sound breathe between the lines in Phase 5. When is a story problem really an audio problem (fix the spine) versus a pacing problem (fix the cover)?
- The editor was tempted to join "I cried" onto "the honey's a bonus" to speed the film up, and refused. Walk through why that specific join would have been a frankenbite, using §30.3's test.
- The single most important decision in the whole edit was removing B-roll from the "I cried" line. Explain, using Murch's Rule of Six (Chapter 29), why holding the face beat covering it — and which priority made the call.
- Both the eyes-closed and sound-off passes had to be clean before lock. Describe a real interview piece you've seen that would fail one but not the other, and what that failure would tell its editor to fix.
- This was a single-camera interview, so every internal trim jumped. How would the edit have differed if it had been shot two-camera (Chapter 19, §19.6)? What would you have gained, and what B-roll pressure would have eased?
Your turn: cut your own three-minute profile
Run this exact pipeline on your Project 2 footage — this is the Production Checkpoint, walked one more time:
- Transcript + radio edit. Auto-transcribe your interview. Read it, highlight keepers, cross out mush, and deal the keepers into story order on paper (FIGURE CS2.2). Find your hook — it may be buried in the middle.
- Selects + soundbite log. Pull each keeper as a select, trimmed on the breath, labeled by content (FIGURE CS2.3).
- Radio string-out. Lay them end to end, close your eyes, and confirm the story holds as audio. Fix the spine now if it doesn't.
- Spine. Commit the order, cut your questions out, tighten honestly. Accept the jumping picture.
- B-roll layer. Cover every seam with matched B-roll on the track above, wild sound underneath. Voice leads; picture proves.
- Pace. Reveal, roam, and return to the face for your two most important lines. Give it one exhale.
- Fine cut + two passes + lock. Tighten words and pictures; run eyes-closed and sound-off; then freeze the picture.
Show the locked cut to one person who has not seen the footage. Ask only two things: "Did you follow the story?" and "Did you feel anything for the person?" If yes to both, you have done what this case study did — turned a real person's scattered words into a film — and Project 2 is locked.
Key takeaways
- The whole method runs on ordinary footage. A single-camera interview and a card of matched B-roll — phone-shootable — became a three-minute film by following the pipeline, not by having special gear.
- The story was built on paper, in the words, before any picture. The radio edit reordered the subject's scattered sentences into an arc (hook → origin → stakes → low → turn → button) they never spoke.
- The spine is done as audio and broken as picture — on purpose. Fifty-four seconds of soundbites, jumping at every seam, then covered completely by B-roll on the track above.
- Matched B-roll does two jobs at once: the bees-on-comb over "you can't rush bees" hid a cut and proved the line. That is why Chapter 20's coverage plan was worth more than pretty footage.
- The most important edit was a cutaway not made: removing B-roll from "I cried" so the face carried the film's emotional low. B-roll must never hide the face when the face is the story.
- A finished piece passes both passes: it holds with the eyes closed (the spine) and shows with the sound off (the matched pictures). Only then do you lock.
- Lock is the finish line the whole book pointed at. The storytelling ends; color, sound, titles, and captions (Part VII, Chapter 36) make the finished film shine.