skills/ai-film-pipeline-master/references/phase-01-story-generation/formats-doc.md

# Formats — documentary

Entry point into the database. Each kit is a ranked short list; open the `cat-*.md` named in the row for the full card.

A format kit answers one question: **given this format's constraints, which techniques earn their place, and why this one rather than the obvious one.** The ranking is the argument.

**One note before the kits.** In documentary, ethics are craft constraints, not an appendix. Three of them recur in every kit below and they are load-bearing:

- **Consent is ongoing, not a signature.** A release obtained on day one does not cover what the film became by day four hundred. Contributors should know what the film is, how they will appear, and what happens if they change their mind.
- **Contradicting material has to be in the film.** If your research turned up something that weakens your thesis and the audience never sees it, you have not made an argument, you have made an advertisement. Broadcast regulation says the same thing more formally: views and facts must not be misrepresented, and a wide range of significant views must be given due weight on matters of major controversy.
- **Reconstruction must be signalled.** Reshooting a routine action is legitimate. Restaging a significant event is not, unless the audience is clearly told. Events arranged for the camera that would not otherwise have happened must be declared. This is BBC editorial policy and it is also simply how you keep a film defensible.

**Rights, licences and clearances in the kits below** describe how each format works in the industry.
In this pipeline they are optional craft, used when the brief itself raises rights: none of them is a
step, and nothing waits on them (root [`../../SKILL.md`](../../SKILL.md) §0).

These are not brakes on the storytelling. They are the reason the storytelling counts for anything.

---

<!-- GENERATED:toc — do not edit by hand. Generated in the source package -->

## Contents

- [Feature Documentary](#feature-documentary)
- [Documentary Series](#documentary-series)
- [Heritage and Historical Documentary](#heritage-and-historical-documentary)
- [Reportage and Current Affairs](#reportage-and-current-affairs)
- [Essay Film](#essay-film)
- [Interview-Led Documentary](#interview-led-documentary)
- [True Crime](#true-crime)
- [Observational and Vérité Feature](#observational-and-vérité-feature)
- [Archive-Only Film](#archive-only-film)
- [Nature and Natural History](#nature-and-natural-history)
- [Sports Documentary](#sports-documentary)
- [Music Documentary](#music-documentary)
- [First-Person Diary Film](#first-person-diary-film)
- [Personal-Archive Family Film](#personal-archive-family-film)
- [The Documentary Short](#the-documentary-short)
- [Interactive Documentary](#interactive-documentary)
- [Interactive Testimony and Oral History](#interactive-testimony-and-oral-history)
- [Museum and Installation Narrative](#museum-and-installation-narrative)
- [Coverage](#coverage)
- [Sources](#sources)

<!-- /GENERATED:toc -->

## Feature Documentary

**Constraints:** Over 40 minutes makes it a feature at most festivals; Full Frame draws the line at 41 minutes and refuses anything past 180. Cannes requires more than 59 minutes for feature consideration. The working sweet spot is 85 to 100 minutes, because that is what programmers, acquisitions and cinema schedules are built around. Financing arrives in stages across two to five years, which means the film's thesis is written into grant applications long before the cut exists and the edit's real job is often to discover that the thesis was wrong. Access is contractual and revocable. E&O insurance requires that every contributor, archive frame and piece of music is cleared and that contested claims are evidenced. **The ethics are constraints on structure, not on conduct alone: consent must survive the film's evolution, contradicting material must appear on screen, and any restaged significant event must be signalled.**

**Grammar:** A feature documentary is built backwards from a question that the film genuinely does not know the answer to when it starts shooting. The narrative unit is the same as in fiction — the scene with a turn — and documentaries fail structurally when they assemble information instead of scenes. The subject is chosen for interest rather than sympathy; a likeable subject with nothing at stake makes ninety unwatchable minutes. Opposition is usually institutional or ambient rather than personified, which is why the writing problem is making a system feel like an antagonist without inventing a villain. The film must hold two clocks: the present-tense thread being filmed, and the past the film is uncovering. And it has to land a value, not merely conclude: the audience should be able to say what changed, for whom, and at what cost.

**Palette:** 85 to 100 minutes, 1 question, 1 or 2 present-tense threads, 3 to 6 contributors, 1 value that lands, an epilogue of under 90 seconds

**Rules:** Write the question, not the thesis, and keep it visible on the wall for the whole edit | Build in scenes with turns, not in information blocks | Choose interesting over sympathetic every time | Put the strongest evidence against your film inside your film | Make the opposition legible without inventing a villain | Hold a present-tense thread so the film has somewhere to happen | Re-consent your contributors when the film changes shape | Signal every reconstruction on screen, without exception | Name the value that turns before you lock the structure | Keep the epilogue short and factual

**Examples:** *Hoop Dreams*, *Man on Wire*, *Honeyland*, *Collective*, *Time*, *Free Solo*, *The Act of Killing*

**Avoid:** The thesis film — shot to confirm a grant application, edited to exclude what contradicted it, and inert because nothing in it was ever in doubt. Assembling a film out of information: eight talking heads delivering facts in a sensible order, with no scene that turns. Subject-capture, where the access relationship becomes so valuable that the filmmaker stops asking the difficult question and the film quietly becomes authorised. Reconstruction blended invisibly into the archive, which is the fastest way to lose a film's standing entirely. And the second ending — and the third — because the team could not decide which of the last four scenes to cut.

**The kit (ranked):**

1. **Write the Ending First** (Structure) `struct.truby_01_end_first`
   Not to lock it, but to know what you are looking for. In documentary this is the difference between shooting and accumulating.
2. **Start From a Question** (Theme) `theme.start_from_a_question`
   The film's engine. A documentary built on an answer has nothing to do for ninety minutes.
3. **The Value Turn** (Scene) `scene.value_turn`
   The unit of documentary construction. If a sequence has no turn, it is information and belongs in a caption.
4. **The Controlling Idea** (Theme) `theme.controlling_idea`
   Found in the edit, not the pitch. Write it the day you lock the assembly and cut against it.
5. **Write the Opponent's Case** (Antagonist) `anta.write_the_case`
   The ethical duty to show contradicting material, expressed as craft. Write the strongest version of the case against your film, then find where it lives on screen.
6. **The Opponent Is Right About Something** (Antagonist) `anta.right_about_something`
   Films that concede one real point are believed on everything else. Films that concede nothing are believed on nothing.
7. **The Institutional Antagonist** (Antagonist) `anta.institutional`
   Most documentary opposition is a system, not a person. This card is how you make a system legible without cartooning it.
8. **The Pressure That Is Not Aimed** (Antagonist) `anta.ambient_pressure`
   Poverty, war, weather, bureaucracy. Pressure with no author, which is the honest shape of most real situations.
9. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
   Track what the audience knows against what each contributor knows. It is how you find where the reveals actually are.
10. **Interesting Beats Likeable** (Character) `char.likeable_vs_interesting`
    The single most consequential casting decision in documentary, and the one most often made for the wrong reason.
11. **The Question Planted First** (Openings) `open.question_planted_first`
    The first three minutes install the question the audience will sit for ninety minutes to resolve.
12. **Dual Timeline** (Time) `time.dual_timeline`
    The present-tense shoot against the past being uncovered. This is the structure most feature docs are actually in.
13. **The Reporter's Stance** (POV) `pov.reportage_stance`
    Decide the film's relationship to its own claims and hold it. Drift here is what audiences experience as bias.
14. **The Narration Contract** (Exposition) `expo.doc_narration_contract`
    If there is narration, state early what it is and is not allowed to know. Breaking that contract later costs the film its authority.
15. **The Plant and the Payoff** (Information) `info.plant_and_payoff`
    Documentary plants are found in the rushes rather than written. The card is about knowing what you are looking for.
16. **Withhold the Reaction Shot** (Emotion) `emo.withhold_reaction`
    Documentary's strongest restraint move, and the correct answer at the moment the edit wants to underline grief.
17. **The Disillusionment Arc** (Arc) `arc.disillusionment`
    Extremely common in documentary because it is what investigation actually does to people, including the filmmaker.
18. **Montage** (Time) `time.montage`
    Necessary and dangerous — the place where documentaries most often assert what they have not shown.
19. **The Value Landed** (Endings) `end.value_landed`
    Name what changed, for whom, at what cost. A documentary that only stops has not ended.
20. **The Epilogue** (Endings) `end.epilogue`
    Title cards about what happened next. Under ninety seconds, factual, and never a substitute for an ending.

---

**The visual kit (ranked):**

1. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   The interview frame that is also an argument. Ninety minutes with four or five people means each of them is placed somewhere that says who they are before they speak.
2. **Building Depth Behind the Interview** (Phase 7 Documentary) `doc.interview_background_depth`
   The difference between a feature documentary and a corporate video is usually the three metres behind the subject's head.
3. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   Feature length is the only documentary form that can afford silence, and silence is what makes an audience believe what they just heard.
4. **Lighting Ratio** (Lighting) `light.ratio`
   Travels across ninety minutes: the same contributor lit softer at the start and harder at the end is an argument nobody has to be told.
5. **Atmospheric B-Roll** (Phase 3 Documentary) `doc.broll_atmospheric`
   Not illustration of the words but the place the words are true in. It is what a feature has room for and a segment does not.
6. **Landscape Breather** (Phase 3 Documentary) `doc.landscape_breather`
   The structural rest. A ninety-minute film without them exhausts an audience that would have stayed for a hundred.
7. **Motivated Lighting** (Lighting) `light.motivated`
   Documentary credibility is largely this. A contributor lit from the window they are sitting beside is believed; the same person lit from a kit is performed.
8. **Colour Temperature as Mood** (Colour) `color.temperature`
   Separates the strands - past and present, two locations, two contributors - across a length where a caption would be intrusive.
9. **The Last Look to Camera** (Phase 3 Documentary) `doc.last_look_to_camera`
   The ending most feature documentaries have and few plan. Shoot it on every interview and decide in the edit.
10. **The 45-Degree Interview Angle** (Angle) `angle.interview_45`
   Ranked last because it is the default. It is correct, and a film in which every one of five contributors sits at the same angle has told the audience nothing about any of them.

## Documentary Series

**Constraints:** Four to six episodes of 45 to 60 minutes is the dominant streaming shape; three and eight both exist and both change the structure. Release is either all at once or weekly, and **that single scheduling decision rewrites the writing: a binge series can withhold across episodes, a weekly series must satisfy within each one.** Episodes need recaps, cold opens and buttons. Contributors recur across episodes, so their arcs must be planned rather than discovered per episode. Archive and music licences are negotiated per-territory and per-term, and a licence lapsing mid-run is a real scheduling risk. The audience may arrive at episode three from a clip.

**Grammar:** A documentary series is one argument divided into chapters that each have to be a whole thing. The recurring failure is the four-part film — material for ninety minutes stretched across four hours, with two episodes of middle. The fix is structural: each episode gets its own question, its own turn and its own value change, while the series question stays unresolved beneath them. Information is rationed on a ledger, not by instinct: every episode must pay something real, or the audience learns that waiting costs nothing. Contributors rotate into the spotlight rather than appearing evenly, which is what gives a series the texture of returning to people. And the series ending must resolve the series question and open the next thing — a finale that only opens is a middle chapter with a budget.

**Palette:** 4 to 6 episodes, 45 to 60 minutes each, 1 series question, 1 episode question per hour, 1 reveal per episode minimum, 1 recurring motif

**Rules:** Give every episode its own question and its own turn | Keep an information ledger for the whole run before you cut episode one | Pay something real in every episode | Decide binge or weekly before you structure, and structure differently for each | Plan contributor arcs across the run, do not discover them per episode | Write the cold open of episode two before you lock episode one | Rotate the spotlight rather than dividing screen time evenly | Resolve the series question in the finale, then open something | Keep one motif running so the run feels like one object

**Examples:** *The Jinx*, *Wild Wild Country*, *Making a Murderer*, *The Last Dance*, *Chef's Table*, *Planet Earth*, four-part streaming dynasty and disaster series

**Avoid:** The stretched feature — the single most common commissioning outcome, recognisable by an episode two that recaps and an episode three that waits. A series whose episodes are topic-divided rather than question-divided, so no episode has a turn in it. Cliffhangers that stall: an open loop the next episode does not close quickly is worse than no loop at all. Contributors who disappear for two episodes and return as strangers. And a finale that saves everything for a second season nobody has commissioned.

**The kit (ranked):**

1. **The Series Opening** (Openings) `open.series_opening`
   The first ten minutes install the series question, the world and the contract. It is the most rewritten sequence in the run.
2. **The Episode Opening** (Openings) `open.episode_opening`
   Each hour needs its own ignition, not a continuation. Episode two's open is the one that decides whether the run survives.
3. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
   Across six hours, tracking what the audience knows is not optional bookkeeping — it is the structure.
4. **The Micro-Reveal Ladder** (Information) `info.micro_reveal_ladder`
   Every episode pays. This is what converts a stretched feature into a series.
5. **The Suspended Ending** (Tension) `tens.suspended_ending`
   The episode button. In a weekly run it must be a question the episode has earned, not one it has dodged.
6. **Thematic Consistency Across a Series** (Theme) `theme.series_consistency`
   Six hours by a rotating edit team drifts thematically unless someone holds this deliberately.
7. **Serial Clock** (Time) `time.serial_clock`
   One visible deadline running under the whole run — a trial date, an election, an expedition window — stops six hours feeling like six incidents.
8. **Multi-POV Architecture** (POV) `pov.multi_pov_architecture`
   Decide whose film each episode is. Series without this read as a collection of interviews.
9. **Rotating the Spotlight** (Relationship) `rel.rotating_spotlight`
   Return to people rather than distributing time evenly. Even distribution is how contributors become furniture.
10. **The Midpoint Reveal** (Information) `info.midpoint_reveal`
    Episode three of six carries the series midpoint. Placing it correctly is most of what makes a run feel well-paced.
11. **The Recontextualising Reveal** (Information) `info.recontextualising_reveal`
    The series-scale move: a late fact that makes the audience re-read four hours. Plan it before you shoot.
12. **The Cold-Open Reveal** (Information) `info.cold_open_reveal`
    The pre-title sequence has to pay immediately in a binge context, where the next thing is one click away.
13. **Dual Timeline** (Time) `time.dual_timeline`
    Then and now, sustained across episodes, with the two threads converging at the finale.
14. **Multiple Opponents, One Question** (Antagonist) `anta.multiple_one_question`
    Different opposition in each episode, all of it testing the same question, is the cleanest series architecture there is.
15. **Reputation Before Entrance** (Character) `char.reputation_before_entrance`
    Hearing about someone for two episodes before meeting them is a device only series length can afford.
16. **Recurring Fragment** (Time) `time.recurring_fragment`
    The same few seconds of footage returning with more context each time. The series-scale motif, and cheap to execute.
17. **The Voice Bible** (Voice) `voice.voice_bible`
    Narration written by three people across six episodes needs a written register or it drifts audibly.
18. **The Callback** (Endings) `end.callback`
    Finales that pay an image from episode one feel designed. It costs nothing and it is what audiences call satisfying.
19. **The Season Ending** (Endings) `end.season_ending`
    Resolve, then open. In that order. A finale that only opens is a middle chapter.
20. **The Handoff Ending** (Endings) `end.handoff_ending`
    Where the series carries a public-interest argument, the last minute belongs to the viewer's next action, not the filmmaker's summary.

---

**The visual kit (ranked):**

1. **Chapter Card** (Phase 3 Documentary) `doc.chapter_card`
   Four to six hours released as a block. The card is what lets a viewer stop, return and know where they are.
2. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   A series returns to the same contributors across episodes. Their frame becomes their signature, and it has to be decided once and held.
3. **Lighting Ratio** (Lighting) `light.ratio`
   The longest runway any documentary form gives a single visual idea. Established in episode one and moved by episode six, it does the arc without narration.
4. **Then and Now Dissolve** (Phase 3 Documentary) `doc.then_and_now_dissolve`
   A series has the time to return to the same frame twice, which is the one structure a feature rarely earns.
5. **Colour Temperature as Mood** (Colour) `color.temperature`
   Separates timeline strands consistently across six hours, which is where a feature's single map stops scaling.
6. **Building Depth Behind the Interview** (Phase 7 Documentary) `doc.interview_background_depth`
   Shot over months in many locations by several crews. Depth is the property that makes them look like one series.
7. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   Series carry more archive than features and mix more sources. Framing each format honestly is how the audience keeps track of what is evidence.
8. **Two-Camera Interview** (Phase 3 Documentary) `doc.two_camera_interview`
   Six hours of sit-downs cannot be cut on a single size. The second camera is a structural requirement here rather than a luxury.
9. **Cold Open** (Phase 3 Documentary) `doc.doc_cold_open`
   Every episode needs one and they cannot all be the same shape. Planning them across the order is a series-level decision.
10. **Outro End Card** (Phase 3 Documentary) `doc.outro_endcard`
   The autoplay countdown starts on it. What is on screen decides whether the next episode plays, and most series leave it to the platform.

## Heritage and Historical Documentary

**Constraints:** No footage exists before the 1890s and very little before the 1920s, so the further back the subject, the more the film is made of stills, objects, places, maps, documents and reconstruction. Archive is licensed per-frame, per-territory, per-term, and heritage institutions often add usage conditions of their own. Contributors are scholars rather than participants, and their disagreement is a fact about the field, not a flaw in the research. Translation and transliteration are structural: names, places and terms have to be decided once and held, and an audience reading subtitles has less capacity for proper nouns. **Reconstruction rules bite hardest here — a restaged historical event must be signalled as reconstruction, and a still image treated to imply motion it never had is the same problem in a subtler form.** These films often carry community and diaspora stakes, where getting a detail wrong is not a critical note but a betrayal.

**Grammar:** The film's real subject is the distance between then and now, and the craft problem is making that distance felt without filling it with assertion. Specificity does the work that footage cannot: the weight of the tool, the price of the grain, the width of the street. Objects carry the argument, because an object survives from then to now and is therefore the only honest bridge the film has. The narration must declare what is known, what is inferred and what is disputed, and it must keep those three registers audibly separate — the collapse of inference into fact is this format's defining dishonesty. Proper nouns are the scarcest resource: a viewer will hold four names an hour, not forty. And the film should end by returning to the present, because a heritage documentary that never comes back has made its subject a museum.

**Palette:** 45 to 90 minutes, 4 to 8 proper nouns the audience must retain, 1 anchoring object, 3 registers (known / inferred / disputed), 1 return to the present

**Rules:** Say what is known, what is inferred and what is disputed, in different words each time | Anchor every abstraction to one physical object | Budget proper nouns and repeat the ones you spend | Fix transliteration once in a style sheet and never vary it | Label every reconstruction on screen | Show scholarly disagreement instead of averaging it into a single confident line | Date every image you show, including the ones you would rather not date | Give the place a voice before you give it a narrator | Return to the present before the end | Never let music assert a feeling the evidence does not support

**Examples:** Ken Burns's *The Civil War* and *The American Revolution*, *Shoah*, BBC and PBS archaeology strands, museum-partnered heritage films, diaspora and community history documentaries

**Avoid:** The lore dump — twelve minutes of chronology at the top, which is what happens when a historian's structure is used as the film's. Averaging live scholarly disagreement into one authoritative narration line, which is the discipline's version of inventing a fact. Reconstruction shot in the grammar of archive: shallow depth of field, grain, sepia, no label. Nostalgia doing the work of argument — the golden-hour ruin, the child's hand on old stone. And the film that never returns to now, which leaves the audience with heritage as decoration rather than inheritance.

**The kit (ranked):**

1. **Heritage Specificity** (Exposition) `expo.heritage_specificity`
   The governing card. One precisely right detail buys more credibility than an hour of sweeping narration.
2. **The Object That States the Rule** (Exposition) `expo.object_tells_rule`
   An object explains a world faster than narration and, unlike narration, it is evidence.
3. **The Object That Carries It** (Emotion) `emo.object_carrier`
   The same object emotionally: a thing that survived is the only honest bridge between then and now.
4. **The Proper Noun Budget** (Exposition) `expo.proper_noun_budget`
   Four names an hour, not forty. This is the card that separates a film from a lecture.
5. **Familiar Anchor, One Strange Word** (Exposition) `expo.familiar_anchor`
   Attach each unfamiliar term to something the audience already holds, then spend the strange word once.
6. **The Prologue of History** (Exposition) `expo.history_prologue`
   When the film genuinely needs chronology up front, this is how to deliver it without a twelve-minute lore dump.
7. **Environmental Exposition** (Exposition) `expo.environmental_detail`
   With no footage, the place is the evidence. Shoot the landscape as testimony, not as a backdrop for narration.
8. **The Voice of a Place or an Object** (Voice) `voice.voice_of_a_place`
   Lets the site carry the account, which is more defensible than a narrator asserting what people felt.
9. **The Retrospective Narrator** (POV) `pov.retrospective_distance`
   Establishes that this is being told from now, which is exactly the honesty the format requires.
10. **Double Anchor** (Time) `time.double_anchor`
    Fix both ends — the then and the now — so the audience always knows which they are in. Heritage films get lost here constantly.
11. **Passage Marker** (Time) `time.passage_marker`
    Centuries move in a cut. Mark them, and mark them the same way every time.
12. **The Pressure That Is Not Aimed** (Antagonist) `anta.ambient_pressure`
    Famine, empire, drought, displacement. Historical opposition rarely has a face, and inventing one is the format's worst temptation.
13. **The Absent Antagonist** (Antagonist) `anta.absent`
    The destroyer who is never on screen — the invader, the fire, the regime. The card is about keeping the absence structurally present.
14. **The Iceberg Principle** (Exposition) `expo.iceberg_principle`
    Ten years of research, one tenth on screen. The rest is what makes the tenth sound certain.
15. **Subtext Across a Language Barrier** (Subtext) `sub.across_languages`
    Subtitled testimony loses the shading. This card is how you keep meaning that lives in register rather than in words.
16. **The Incantatory Register** (Voice) `voice.incantatory_register`
    Lists of names, places, years. Powerful and expensive — use it once, at the point the film is making its largest claim.
17. **Nostalgia Without Sentimentality** (Emotion) `emo.nostalgia`
    The failure mode this genre lives inside. Read the card as a checklist against your own cut.
18. **The Institutional Voice** (Voice) `voice.institutional_voice`
    Museum-partnered films inherit this register by default. Choose it deliberately or it chooses you.
19. **The Mirror Ending** (Endings) `end.mirror_ending`
    Return to the opening image with the present in it. The cleanest way to make a heritage film land as inheritance.
20. **The Lore Dump** (Exposition) `expo.lore_dump`
    Read as a guard. It is the specific failure every historical documentary's first assembly contains.

---

**The visual kit (ranked):**

*On a wholly generated piece every card below is a technique and every shot it makes is a reconstruction, as the documentary selector's first two axes classify it: raking light and the macro describe how the built image is lit and framed, not a record that exists, and the shot is tagged as reconstruction.*

1. **The Reference-First Rule for Heritage** (Reference) `ai.reference_first_heritage`
   Ranked first without competition. No footage exists before the 1890s, so every historical image is built, and every built image is a claim that needs a source.
2. **Raking Light on Relief** (Phase 3 Documentary) `doc.raking_light_relief`
   The one lighting decision that makes carved stone readable. Lit flat, a relief is a grey wall; lit across, it is the document it actually is.
3. **Site Scale Figure** (Phase 3 Documentary) `doc.site_scale_figure`
   A person in the frame is the only reliable scale a ruin has, and scale is most of what a heritage audience came for.
4. **Macro on the Artefact** (Phase 3 Documentary) `doc.macro_artifact`
   The object at a distance no visitor is allowed. It is the format's single strongest offer and it costs one lens.
5. **Golden Hour on Heritage** (Phase 3 Documentary) `doc.golden_hour_heritage`
   Ranked here and watched carefully: it is beautiful, it is the format's cliche, and a film that shoots every site at the same hour has made them all the same place.
6. **Dramatised Reconstruction** (Phase 7 Documentary) `doc.dramatised_reconstruction`
   Marked as reconstruction on screen, always. An unmarked reconstruction is the mechanism by which a documentary becomes a source for something untrue.
7. **Reconstruction Overlay** (Phase 3 Documentary) `doc.reconstruction_overlay`
   The honest version of the rebuild: the ruin and the reconstruction in the same frame, so the audience can see which is which.
8. **Period-Truthful Lettering** (Text) `textimg.period_lettering`
   A script on screen is a claim about a writing system, and in heritage work it is the claim most often made from memory.
9. **Tilt Up the Monument** (Phase 3 Documentary) `doc.tilt_up_monument`
   The scale move, used once. Repeated it becomes the format's other cliche and stops registering as size at all.
10. **Dust and Light in the Temple** (Phase 3 Documentary) `doc.dust_and_light_temple`
   Atmosphere that is also evidence: the shaft shows the architecture's own intent, which is a thing a heritage film can argue rather than assert.

## Reportage and Current Affairs

**Constraints:** Segments run 8 to 25 minutes inside a strand, or 40 to 60 as a single. Lead times are days to weeks, not years. Regulation is explicit and enforceable: **news must be reported with due accuracy and presented with due impartiality; on matters of major political controversy an appropriately wide range of significant views must be included and given due weight; views and facts must not be misrepresented; and significant mistakes should normally be acknowledged and corrected on air quickly.** Right of reply is a process with a clock on it, and the letter you send determines what you can broadcast. Defamation and contempt law constrain what can be said about living people and live proceedings. Contributors may be at risk, so anonymisation is a craft decision with safety consequences.

**Grammar:** Reportage is an argument that must survive being read by its subject with a lawyer. That single condition produces the format's grammar. Claims are ranked by what you can evidence, and the film is structured so that the strongest evidenced claim carries the most weight — not the most dramatic one. The response from the accused is not a compliance box at the end; it is placed where it is genuinely tested, because a reply buried in a closing caption reads as exactly what it is. The reporter's stance is fixed at the outset and held: observer, participant or advocate, each is defensible, but drifting between them is what audiences perceive as bias. Understatement is the registry of authority here — the film that states its findings flatly is believed, and the one that underlines them is not. And the piece answers the question it opened with, or says plainly that it could not.

**Palette:** 8 to 25 minutes, 1 question, 3 evidenced claims, 1 right-of-reply placed on screen, 1 stance held throughout, 0 adjectives in the findings

**Rules:** Rank your claims by evidence and structure the film in that order | Put the reply where it is genuinely tested, not at the end | Fix the reporter's stance in the first minute and never drift from it | State findings flatly — the adjective is what a lawyer will attack | Give the strongest opposing argument its best version, not its weakest | Show your method where the method is the story | Name what you could not establish | Correct on air, quickly, and in the same register as the original claim | Never let the edit imply a causal link the evidence does not support

**Examples:** *Panorama*, *Dispatches*, *Frontline*, *60 Minutes* investigations, Al Jazeera's investigative strand, long-form newspaper video investigations

**Avoid:** The pre-written conclusion, where the reporting exists to illustrate a script that already existed — legally dangerous and, worse, usually wrong. Right of reply handled as a caption over black at 19:40, which is technically compliant and read by every viewer as evasion. Adjectives inside findings: "shocking", "damning", "outrageous" — each one converts a report into an opinion piece and hands the subject a complaint. A stance that drifts from observer to advocate somewhere in act two, which is what audiences mean when they say a piece felt biased. And the montage that implies causation the film has not evidenced, which is the most common way a rigorous piece becomes indefensible.

**The kit (ranked):**

1. **The Reporter's Stance** (POV) `pov.reportage_stance`
   Decide it in the first minute and hold it to the last. Every perceived bias problem starts here.
2. **Write the Opponent's Case** (Antagonist) `anta.write_the_case`
   The duty to give a wide range of significant views due weight, done as craft rather than as compliance.
3. **The Opponent Is Right About Something** (Antagonist) `anta.right_about_something`
   Conceding one real point is what makes the rest of the film credible — and it is also what survives a complaint.
4. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
   Track what you can evidence against what you merely believe. The ledger is the difference between a report and a claim.
5. **The Fair-Play Contract** (Information) `info.fair_play_contract`
   The audience must be able to reach your conclusion from what you showed them. Anything withheld to protect a reveal is a problem here.
6. **Authority by Specificity** (Voice) `voice.authority_by_specificity`
   The date, the figure, the document reference. Precision is the format's entire authority mechanism.
7. **The Understating Voice** (Voice) `voice.understating_voice`
   State the finding flatly. Underlining it is what converts a report into an opinion and invites the complaint.
8. **The Narration Contract** (Exposition) `expo.doc_narration_contract`
   Tell the audience what this film set out to establish and what it did not. It is also your best defence.
9. **The Institutional Antagonist** (Antagonist) `anta.institutional`
   The opposition is almost always a body rather than a person. Make the body legible without inventing a villain inside it.
10. **Who Asks the Questions** (Dialogue) `dial.question_control`
    In a confrontation interview, control of the question is the scene's power structure and the viewer reads it instantly.
11. **Refusing the Frame** (Dialogue) `dial.refusing_the_frame`
    What a subject declines to accept about your question is often more informative than their answer.
12. **Deflection** (Dialogue) `dial.deflection`
    Naming the deflection technique lets you cut the exchange so the audience sees it happen rather than being told it happened.
13. **Exposition Under Pressure** (Exposition) `expo.exposition_under_pressure`
    Background delivered inside a live confrontation, which is how current affairs gets context in without a lecture.
14. **The Cold Open** (Openings) `open.cold_open`
    Strand format demands it. The pre-title must state the question, not tease the conclusion.
15. **Descriptive, Not Prescriptive** (Theme) `theme.descriptive_not_prescriptive`
    Show what happened; let the audience draw the rule. The moment the film prescribes, impartiality becomes arguable.
16. **Let Someone Else Feel It** (Emotion) `emo.someone_else_feels_it`
    Put the emotional weight on a witness rather than on the reporter, which keeps the stance intact.
17. **Concrete Nouns Over Abstract** (Rhythm) `rhythm.concrete_nouns`
    "Twelve unpaid invoices", not "financial irregularities". The concrete noun is also the evidence.
18. **Answer the Question That Was Asked** (Endings) `end.answer_the_question_asked`
    Answer the opening question or say plainly that you could not. Both are honest; drifting away from it is not.
19. **The Rigged Argument** (Theme) `theme.rigged_argument`
    Read as a guard. This card is the diagnostic for a film that only allowed one side to be articulate.
20. **The Red Herring** (Information) `info.red_herring`
    Also a guard. A misdirection that is legitimate craft in fiction is misrepresentation in journalism — know when you have imported it.

---

**The visual kit (ranked):**

1. **Stand-Up** (Phase 3 Documentary) `doc.stand_up`
   The reporter in the place, establishing that they were in it. Lead times of days mean it is often the only shot that cannot be replaced.
2. **Nat Sound Open** (Phase 3 Documentary) `doc.nat_sound_open`
   The segment begins in the place rather than in the argument, which is what separates reportage from commentary.
3. **B-Roll on the Noun** (Phase 3 Documentary) `doc.broll_on_the_noun`
   Cut on the thing being named. It is the fastest way to make a dense eight-minute package legible at broadcast pace.
4. **Lower Third** (Phase 3 Documentary) `doc.lower_third`
   Who this is and why they are qualified to say it, at a pace nobody can pause. It is an editorial decision rather than a graphic one.
5. **Vox Pop** (Phase 3 Documentary) `doc.vox_pop`
   Used carefully and ranked mid-list for a reason: it is the cheapest way to imply a public mood the piece has not measured.
6. **Flat Lighting / Beauty Lighting** (Lighting) `light.flat`
   Shot in hours by a two-person crew. Fast, consistent, and correct for a format whose credibility is in what is said.
7. **Medium Close-Up** (Shot size) `size.mcu`
   The interview size at speed. It needs no set dressing and it reads on a phone, a desktop and a television.
8. **Piece to Camera** (Phase 3 Documentary) `doc.piece_to_camera`
   The direct address that carries a sentence the edit cannot. It is where the reporter takes responsibility for the claim.
9. **Legible at the Size It Is Actually Watched** (Access) `access.legible_at_size`
   Most current affairs is now first watched as a vertical clip. Type set for a television is type the actual audience cannot read.
10. **The Receipt** (Phase 3 Documentary) `doc.the_receipt`
   The document on screen. In a format with days of lead time it is the only thing that survives a complaint.

## Essay Film

**Constraints:** No fixed length: essay films run from 12 minutes to two hours, and the form has no commissioning slot that constrains it, which is both its freedom and its funding problem. It lives at festivals, in galleries, on Mubi and on the arthouse fringe of streaming, and it is often made by one or two people over years. Budgets are small and the material is frequently found — archive, home video, screen recordings, other people's films — which makes rights the format's central practical constraint, and fair use or fair dealing the ground it stands on. There is no institutional authority behind the voice, only the maker's, so **the film's credibility is entirely a function of how honestly it handles its own uncertainty.**

**Grammar:** The essay film thinks on screen. It does not present a conclusion; it performs the process of arriving at one, including the wrong turns. That is why the first person is not a stylistic choice here but the form's condition: the maker is in the frame as a doubting, revising, particular person, and hiding behind facts and experts would make it a different format. The argument moves by association rather than by causation, so the joins between sections are where the writing lives — the essay film's equivalent of a scene transition is a leap the audience is invited to make. Repetition is structural: a phrase, an image, a piece of footage returns with new weight each time. Narration carries more of the load than in any other documentary form, which makes it prose that has to survive being heard once, at the pace of images. And the ending is usually open, because a form built on doubt cannot honestly close.

**Palette:** 20 to 90 minutes, 1 first person, 1 recurring image, 3 to 5 movements, 1 admitted uncertainty, an open ending

**Rules:** Put the maker in the film as a person who does not yet know | Move by association and write the joins carefully | Let one image or phrase recur and accrue meaning | Admit what you got wrong, on screen, in the film's own voice | Write the narration as prose, then cut a third for the ear | Never let the voice claim authority the film has not demonstrated | Choose the found image for what it shows, not for what it illustrates | Leave the ending open if the question is open | Design around found material from the first draft

**Examples:** Chris Marker's *Sans Soleil* and *Letter from Siberia*, Agnès Varda's *The Gleaners and I* and *Beaches of Agnès*, Alain Resnais's *Night and Fog*, *Los Angeles Plays Itself*, Kirsten Johnson's *Cameraperson*, the video-essay tradition

**Avoid:** The lecture with pictures — a confident argument read over illustrative footage, which is an expository documentary wearing the essay film's clothes and is the form's most common counterfeit. Association used as a licence for incoherence: leaps the audience cannot make are not subtle, they are unfinished. Narration written for the page, which is audibly too dense the moment it is spoken over images. Borrowed profundity, where the film leans on quoted philosophers instead of thinking. And an open ending used to avoid a conclusion the film actually reached.

**The kit (ranked):**

1. **First Person Central** (POV) `pov.first_person_central`
   The form's condition. The maker is the subject as much as the topic is, and a film that hides this is a different format.
2. **The Narrating Persona** (Voice) `voice.narrating_persona`
   The "I" in the film is constructed. Designing it deliberately is what separates an essay film from a diary.
3. **Start From a Question** (Theme) `theme.start_from_a_question`
   The whole form is a question being worked on in public. An answer delivered up front leaves nothing to watch.
4. **The Self-Correcting Narrator** (Voice) `voice.self_correcting_narrator`
   Revising a claim on screen is the essay film's core move and the source of its particular credibility.
5. **Unstable Authority** (Voice) `voice.unstable_authority`
   A voice that can be wrong. The form depends on this and collapses into lecture without it.
6. **Non-Linear Assembly** (Time) `time.nonlinear_assembly`
   Order by argument rather than chronology, and let the sequence itself be a claim.
7. **Recurring Fragment** (Time) `time.recurring_fragment`
   The same image returning with more weight each time is how an essay film builds without a plot.
8. **The Thematic Motif** (Theme) `theme.thematic_motif`
   Motif does the structural work that narrative does elsewhere. Choose one and commit to it early.
9. **Deferred Explanation** (Exposition) `expo.deferred_explanation`
   Show first, explain much later, sometimes never. It is what gives the form its characteristic attention.
10. **The Found Document** (POV) `pov.found_document`
    Letters, footage, screen recordings, other people's home video. The form's native material and its rights problem.
11. **The Long Breath Sentence** (Rhythm) `rhythm.long_breath`
    The essay film's signature cadence, and the reason its narration has to be tested aloud against picture.
12. **Anaphora** (Rhythm) `rhythm.anaphora`
    Repetition at the head of successive sentences. It is how spoken essay prose keeps its shape over images.
13. **Parallel Construction** (Rhythm) `rhythm.parallel_construction`
    Parallel syntax lets an audience hold a comparison across two minutes of unrelated pictures.
14. **The Same Event Twice** (POV) `pov.same_event_twice`
    Return to the same material from a second position. The form's most reliable structural device.
15. **The Second Subject** (Subtext) `sub.second_subject`
    Every essay film is about two things. Knowing which is the hidden one tells you what to cut.
16. **Second Person** (POV) `pov.second_person`
    Address to a "you" — Marker's letters, Varda's asides. Costly and specific; use it where the film turns outward.
17. **The Confiding Voice** (Voice) `voice.confiding_voice`
    The register that makes a stranger's thinking bearable for ninety minutes.
18. **The Incantatory Register** (Voice) `voice.incantatory_register`
    For the film's highest claim. Once, and never twice.
19. **The Anti-Theme** (Theme) `theme.anti_theme`
    Argue against your own position somewhere in the film. In this form it is not balance, it is method.
20. **The Open Ending** (Endings) `end.open_ending`
    Legitimate here in a way it rarely is elsewhere — provided the openness is the film's finding, not its evasion.

---

**The visual kit (ranked):**

1. **Essay Montage as Argument** (Phase 3 Documentary) `doc.essay_montage_argument`
   The form's central mechanism: the argument is made by what is placed next to what, not by the narration over it.
2. **Text Overlay Thesis** (Phase 3 Documentary) `doc.text_overlay_thesis`
   The claim stated as type, deliberately, so the audience can weigh it rather than absorb it.
3. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   An essay film quotes images. Framing the quotation as a quotation is the difference between an argument and a sleight of hand.
4. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   The form has no commissioning slot and no length, which means it has the one thing every other documentary lacks: room to stop.
5. **The Long Observational Hold** (Phase 3 Documentary) `doc.long_observational_hold`
   The shot held past comfort, which is where the essay film does its thinking and where an audience starts doing theirs.
6. **Ken Burns** (Phase 3 Documentary) `doc.ken_burns`
   Ranked here because the essay film uses stills more than any other documentary form, and because the move is the interpretation.
7. **Kinetic Quote** (Phase 7 Documentary) `doc.kinetic_quote`
   A quotation typeset and moved is a quotation the film is taking a position on. Choose the treatment as carefully as the words.
8. **Colour Temperature as Mood** (Colour) `color.temperature`
   Separates the film's registers - quoted material, filmed material, the essayist's own - without a caption on any of them.
9. **Side by Side Comparison** (Phase 7 Documentary) `doc.side_by_side_comparison`
   The comparison stated rather than implied. In essay work this is honesty: the audience sees both halves of the claim.
10. **Then and Now Dissolve** (Phase 3 Documentary) `doc.then_and_now_dissolve`
   The form's oldest figure and still the most persuasive, which is exactly why it is ranked last and used once.

## Interview-Led Documentary

**Constraints:** Built from four to ten sit-down interviews plus archive and B-roll, and structured in a paper edit from transcripts rather than from a script. The film is written in the edit: the assembly is made by rearranging highlighted transcript sections into a beginning, middle and end, then cutting the picture to fit. Questions are almost always removed, which means the answer must contain its own question — a discipline that happens in the interview room or not at all. Cuts within an answer need covering material, so B-roll and archive volume is a structural dependency rather than decoration. **Consent is the format's pressure point: contributors are giving testimony about their own lives, often in a single emotionally exposed sitting, and the film that results may not resemble what they agreed to. Re-consent when the shape changes, and give contributors a route back to you afterwards.** **And consent is a document, not only a conversation.** Every contributor who appears signs a release before the camera runs, naming what the film is, where it may be shown, in what languages, and for how long; the same goes for anyone whose archive, photographs or recordings appear. This package writes the craft of consent and cannot write the paperwork for you — the wording is a matter for whoever carries the legal risk, and a broadcaster or funder normally supplies its own form. What belongs on the checklist here is that **the form exists, it is signed, and it is signed before the sitting rather than after it.** A film can be finished, true and unairable.

**Grammar:** The film's material is not what people say, it is the gap between what they say and what they show. The most useful moments in any interview transcript are contradictions — between two contributors, between a contributor and the record, and between a contributor's words and their hands. The structure is built from those gaps, not from the chronology of events. Contributors are cast for how differently they see the same thing; four people who agree make one very long interview. Silence is content: the pause before an answer is frequently the answer, and cutting it out is the format's most common act of vandalism. And the last thing anyone says in the film should be chosen for its meaning, not because it was the natural end of the last interview.

**Palette:** 4 to 10 contributors, 1 event seen from at least 3 positions, 1 contradiction per contributor, a paper edit before a timeline, 1 chosen last line

**Rules:** Get the question inside the answer, in the room | Cast for disagreement, not for coverage | Build the paper edit from transcripts before you open the timeline | Structure on contradictions, not on chronology | Keep the pause — it is content, not dead air | Shoot enough covering material to cut inside answers honestly | Never assemble a sentence a contributor did not say | Re-consent when the film's shape changes | Choose the last line rather than inheriting it | Give every contributor a route back to you after transmission

**Examples:** *The Fog of War*, *Amy*, *Three Identical Strangers*, *The Last Dance*, oral-history films, StoryCorps-derived work

**Avoid:** The agreement film — eight contributors confirming the same account for fifty minutes, which has no scenes and no turns and cannot be rescued in the edit. Frankenbiting: assembling a sentence out of parts from different answers, which is the format's cardinal offence and ends careers when it surfaces. Cutting every pause to tighten the piece, which removes precisely the moments audiences remember. Questions left in because the answer did not stand alone, which should have been fixed in the room. And the emotional interview obtained by pressing a contributor who did not understand what would be broadcast.

**The kit (ranked):**

1. **Who Asks the Questions** (Dialogue) `dial.question_control`
   The interview's power structure, and in this format it determines whether the answer can stand without the question.
2. **The Same Event Twice** (POV) `pov.same_event_twice`
   Two contributors on one event is the format's fundamental unit. Without it you have testimony, not a film.
3. **Behaviour That Contradicts the Line** (Subtext) `sub.behaviour_contradicts`
   What the hands do while the mouth speaks. This is the material a transcript cannot show you — watch the rushes, not the paper.
4. **Information Asymmetry Between Characters** (Information) `info.character_asymmetry`
   Contributor A does not know what B said. Structuring on that gap is what gives an interview film its tension.
5. **Silence as Content** (Subtext) `sub.silence_as_content`
   The pause before the difficult answer is frequently the truest thing in the interview.
6. **Grading the Gap** (Subtext) `sub.graded_pause`
   Not all pauses are equal. This card is how you decide which ones to keep at full length.
7. **Interesting Beats Likeable** (Character) `char.likeable_vs_interesting`
   Casting contributors is the whole film. A warm contributor with nothing at stake is forty unusable minutes.
8. **Multi-POV Architecture** (POV) `pov.multi_pov_architecture`
   Decide whose account structures the film and whose complicates it. Otherwise the paper edit has no spine.
9. **The Half-Truth** (Dialogue) `dial.half_truth`
   Contributors rarely lie outright. Recognising the partial answer is what tells you where to push and where to cut.
10. **What They Do Alone** (Character) `char.alone_behaviour`
    The B-roll brief, properly understood: film the contributor doing what they do when nobody is asking.
11. **The Stress Tell** (Character) `char.stress_tell`
    Every contributor has one. Finding it in the rushes tells you which answers cost them something.
12. **The Unanswered Line** (Dialogue) `dial.unanswered_line`
    A question the contributor declines to engage with is usable material and sometimes the film's turning point.
13. **Information the Listener Already Has** (Dialogue) `dial.known_information`
    Why contributors explain things they know you know, and how that explanation reveals who they think is watching.
14. **The Delayed Reaction** (Emotion) `emo.delayed_reaction`
    The feeling that arrives three seconds after the sentence. Hold the shot; this is why you rolled long.
15. **Withhold the Reaction Shot** (Emotion) `emo.withhold_reaction`
    The restraint move. The edit's instinct to cut to a tearful face at the peak is almost always wrong.
16. **Rotating the Spotlight** (Relationship) `rel.rotating_spotlight`
    Return to contributors rather than spreading them evenly, so the audience feels they are coming back to someone.
17. **The Second-Language Voice** (Voice) `voice.second_language_voice`
    Testimony in a second language carries hesitations that are not uncertainty. Subtitling flattens this; the card is how to protect it.
18. **The Staged Reveal** (Information) `info.staged_reveal`
    Showing a contributor something on camera. Powerful, and ethically the most demanding thing this format does — plan the consent for it in advance.
19. **The Specific Detail** (Emotion) `emo.specific_detail`
    The detail nobody would invent is what makes testimony land. Ask for it directly in the room.
20. **The Last Line** (Endings) `end.last_line`
    Chosen from the whole transcript corpus, not inherited from the final interview. It is the most reworked decision in the cut.

---

**The visual kit (ranked):**

1. **Two-Camera Interview** (Phase 3 Documentary) `doc.two_camera_interview`
   Ranked first because the paper edit depends on it. Four to ten sit-downs cut on one size gives an editor no way to compress an answer.
2. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   Each contributor placed somewhere that argues. With ten interviews the placements are also how the audience tells them apart.
3. **Building Depth Behind the Interview** (Phase 7 Documentary) `doc.interview_background_depth`
   The whole film happens in these frames. Three metres of depth behind a head is the production value.
4. **The Interrotron** (Phase 3 Documentary) `doc.interrotron`
   Eye contact with the lens. It changes what an audience feels about a contributor more than any lighting decision will.
5. **Interview Jump Cut** (Phase 7 Documentary) `doc.interview_jump_cut`
   Declared rather than hidden. A visible compression of an answer is honest; a concealed one is the form's standing ethical problem.
6. **Lighting Ratio** (Lighting) `light.ratio`
   Differentiates contributors without editorialising about them, and holds across a shoot spread over months.
7. **The Empty Chair** (Phase 3 Documentary) `doc.empty_chair`
   Who declined, shown. It is a factual statement and it belongs in the visual kit because it is made with a camera.
8. **Noddy** (Phase 3 Documentary) `doc.noddy`
   Ranked low and named honestly: it is a fabricated reaction, it is standard practice, and a film that uses it should know it is doing so.
9. **Slow Push on the Answer** (Phase 3 Documentary) `doc.slow_push_answer`
   The one move the form has. Used on the answer that matters and nowhere else, it is invisible and it works.
10. **Anonymous Interview** (Phase 3 Documentary) `doc.anonymous_interview`
   Where a contributor must not be identified, the package writes no concealment technique (root `SKILL.md` §0); the piece is written, and the note goes in the handoff for whoever delivers.

## True Crime

**Constraints:** Four to six episodes, or a 90-minute single. The legal exposure is unlike any other documentary format: living accused, ongoing appeals, contempt in live proceedings, defamation risk on every implication, and in some jurisdictions restrictions on identifying victims and minors. The ethical exposure is worse and is not covered by the law. **Consent from victims' families is not legally required and is routinely not sought, and families report being unnotified, retraumatised and given no editorial voice; the Dahmer backlash made this a mainstream, not niche, criticism.** Production bodies including a producers' association have since pledged duty-of-care standards — long-term contact points for contributors, notification before release, support routes. Meanwhile the commercial pressure runs the other way: the genre rewards the very sensationalism these standards restrain.

**Grammar:** True crime is a mystery structure applied to real suffering, and every craft decision is a negotiation between those two facts. The mystery machinery works — withheld information, staged reveals, recontextualisation — but each device is being run on a family's worst week, which changes what is permissible. The reliable discipline is the fair-play contract: the audience must be able to reach the film's conclusion from what the film showed them, because anything withheld purely for effect is, here, a misrepresentation of a real case. The perpetrator gets understanding but never glamour, and the case against the film's own thesis has to be present, not gestured at. Victims are people with a life before the crime; a victim introduced only as a body has already been used. And the ending must state honestly what is established and what is not, because the audience will act on it — contact police, harass a suspect, reopen a wound.

**Palette:** 4 to 6 episodes or 90 minutes, 1 case question, 1 named legal status per person, victims introduced alive, 1 stated limit on what is established, a duty-of-care contact for every contributor

**Rules:** Introduce every victim as a living person before you introduce the crime | State each person's legal status precisely, every time | Show the strongest case against your own thesis | Withhold nothing that the audience needs to judge the conclusion | Never let the edit imply guilt the record does not support | Notify and offer a route to families before release, whether or not you must | Give the perpetrator explanation, never glamour | Say plainly at the end what is established and what is not | Keep a contact point open for contributors after transmission

**Examples:** *The Jinx*, *Making a Murderer*, *The Staircase*, *The Keepers*, *American Nightmare*, the podcast-to-series pipeline and its adaptations

**Avoid:** Editing a reveal so that a viewer concludes someone is guilty when the record does not establish it — legally catastrophic and morally worse. Introducing a victim as a corpse, so the entire film treats them as material. The romanticised perpetrator: the sympathetic casting, the charismatic framing, the score that makes the killer interesting. Withholding a fact from the audience purely to power an episode button, which in a real case is not suspense but misrepresentation. And the series that reopens a case with no plan for the family who will be contacted by strangers the week it drops.

**The kit (ranked):**

1. **The Fair-Play Contract** (Information) `info.fair_play_contract`
   In fiction this is a courtesy. On a real case it is the ethical floor: the audience must be able to reach your conclusion from what you showed.
2. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
   What is established, what is alleged, what is inferred — tracked per episode. It is the structure and the legal defence at once.
3. **Write the Opponent's Case** (Antagonist) `anta.write_the_case`
   Build the strongest version of the case against your thesis and put it in the film. Nothing else protects a true-crime series.
4. **The Opponent Is Right About Something** (Antagonist) `anta.right_about_something`
   Concede the point the prosecution or the defence genuinely has. It is what makes the rest believable.
5. **Restricted Narration** (Information) `info.restricted_narration`
   Hold the audience to what was known at the time. It generates suspense without withholding anything the viewer needs to judge.
6. **Dread of the Known Outcome** (Tension) `tens.known_outcome_dread`
   The audience already knows someone dies. This card is how you generate tension without pretending they do not.
7. **The Specific Detail** (Emotion) `emo.specific_detail`
   The detail that makes a victim a person — the job, the joke, the unfinished renovation. Do this before the crime, always.
8. **Grief on the Page** (Emotion) `emo.grief`
   Families are on screen grieving. The card is about rendering grief without staging or harvesting it.
9. **Let Someone Else Feel It** (Emotion) `emo.someone_else_feels_it`
   Put the weight on a witness rather than pressing a family member for a reaction shot you can use in the trailer.
10. **Dual Timeline** (Time) `time.dual_timeline`
    The crime then, the investigation or reinvestigation now. The genre's default structure and the reason it sustains six hours.
11. **Double Anchor** (Time) `time.double_anchor`
    Date every thread on screen. In a case with three timelines, ambiguity about when reads as an implication about who.
12. **The Recontextualising Reveal** (Information) `info.recontextualising_reveal`
    Legitimate when the new fact genuinely re-reads the evidence, and misrepresentation when it was simply held back.
13. **Crime Pursued by Vengeance** (Situations) `sit.polti_03_crime_pursued_by_vengeance`
    Naming the situation the series is actually dramatising keeps the pursuit from quietly becoming the film's hero narrative.
14. **The Reporter's Stance** (POV) `pov.reportage_stance`
    Advocate, observer or investigator. All three exist in this genre; drifting between them is what produces the wrongful-implication problem.
15. **Real People and the Composite** (Character) `char.real_person_composite`
    The card's boundary matters most here: no composite victims, no compressed suspects, no invented conversations.
16. **Waiting as a Weapon** (Tension) `tens.waiting_as_weapon`
    Investigations are mostly waiting. Dramatising the wait honestly beats manufacturing a false clock.
17. **Answer the Question That Was Asked** (Endings) `end.answer_the_question_asked`
    Say what is established. If the case is unresolved, say that plainly rather than implying a conclusion you cannot support.
18. **The Open Ending** (Endings) `end.open_ending`
    Honest when the case is open. Dishonest when used to leave a suggestion the film would not state in words.
19. **The Red Herring** (Information) `info.red_herring`
    Read as a guard. A real person set up as a misdirection is a person you have defamed for a story beat.
20. **The Cheat** (Information) `info.the_cheat`
    The last guard. Every withheld fact should be tested against this card before the episode is locked.

---

**The visual kit (ranked):**

1. **Dramatised Reconstruction** (Phase 7 Documentary) `doc.dramatised_reconstruction`
   Marked on screen, every time. In a format with this legal exposure an unmarked reconstruction is not a style choice, it is a liability.
2. **Faceless Reenactment** (Phase 7 Documentary) `doc.faceless_reenactment`
   Shows the action without asserting who performed it, which is the only reconstruction available before a conviction.
3. **The Receipt** (Phase 3 Documentary) `doc.the_receipt`
   The document on screen. In true crime it is the difference between reporting an allegation and making one.
4. **Interview in Darkness** (Phase 3 Documentary) `doc.interview_in_darkness`
   The format's signature and its most abused device: it makes a contributor look guilty regardless of what they say. Choose it for a reason or not at all.
5. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   The alternative to the darkness, and usually the better one. A person in their own room is harder to caricature.
6. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   Police tape, news footage, home video and CCTV are four different kinds of evidence, and framing them identically flattens the difference.
7. **Timeline Graphic** (Phase 7 Documentary) `doc.timeline_graphic`
   The case's structure, shown. Built from real dates rather than generated, because a wrong date here is defamatory rather than embarrassing.
8. **Colour Temperature as Mood** (Colour) `color.temperature`
   Separates the then from the now across four to six episodes, which is the format's main structural burden.
9. **Night Site Torchlight** (Phase 3 Documentary) `doc.night_site_torchlight`
   The place at the hour it happened. It is atmosphere, and it should be labelled as atmosphere rather than implied as evidence.
10. **Last Look to Camera** (Phase 3 Documentary) `doc.last_look_to_camera`
   The ending. In true crime it is also the moment the film decides whether it is about a case or about a person.

## Observational and Vérité Feature

**Constraints:** Shooting ratios of 100:1 are ordinary and 300:1 is not remarkable, across one to four years, because the method is to be present until something happens. The purist grammar removes the film's usual instruments: no narration, no interviews, no score, no on-screen text beyond location titles — Wiseman has worked this way for six decades and the discipline is what the form means by its name. That leaves the edit carrying the entire argument, and the edit is where a vérité film is written. Access is a relationship with no contract strong enough to survive being abused. **Non-intervention is an ethical position before it is a style, and the position is unstable: choosing where to point the camera is already an intervention, and direct cinema's strict hands-off stance and vérité's willingness to provoke are two different answers to the same question, not two aesthetics.** The harder version arrives on location — the moment when filming and helping are alternatives. **Consent has to be re-taken precisely because the method works: a subject who has genuinely forgotten the camera has stopped consenting in any meaningful sense, and a release signed in month one does not cover the scene shot in year three.**

**Grammar:** The film is assembled from scenes that were found rather than written, which inverts the usual order: you cannot plan a turn, you can only recognise one in the rushes and then protect it in the cut. So the craft sits in two places — what you shoot, and what you are willing to lose. Behaviour is the whole text. What a person does with their hands while someone else is talking, what they do when the room empties, which small task they perform carefully: these carry everything narration would otherwise assert, and they are the only evidence the form allows itself. Duration is a deliberate instrument. A scene held past the point of comfort is how the form says what a cut would flatten, and the pause before an answer is content rather than dead air. Institutions supply the opposition, because the form's natural subject is a system meeting a person and neither one has a villain in it. And the film must resist its own strongest temptation, which is to cut to the face at the moment of feeling — the withheld reaction shot is this form's signature and its ethics at once.

**Palette:** 80 to 110 minutes, 1 place or 1 institution, 2 to 5 people followed, 100:1 upward ratio, 0 interviews, 0 narration, 1 scene held past comfort

**Rules:** Shoot until the camera stops being an event, then re-consent because it has | Write nothing; recognise the turn in the rushes and cut to protect it | Let behaviour carry every claim you would otherwise narrate | Hold the scene past the point the edit wants to leave | Decide before the shoot where filming stops and helping starts, and write it down | Cut the scene that only works because a subject forgot you were there | Make the institution legible without giving it a face | Never build a moment out of two moments that were not adjacent | Withhold the reaction shot at the peak | Show the film to the people in it before anyone else sees it

**Examples:** Frederick Wiseman's institutional films, *Primary* and the Drew Associates films, *Grey Gardens*, *Dont Look Back*, *Hoop Dreams*, *Honeyland*, *Gunda*, long-form observational strands on hospitals, schools and courts

**Avoid:** Fly-on-the-wall as an alibi, where the camera claims neutrality it has not earned and the film's real editorial choices go undeclared. The voyeur at the window: material obtained because a subject was too exhausted or too accustomed to object, which plays on screen exactly as it was obtained. Narration bolted on in post because the assembly does not hold — that is not a fix, it is an admission that the scenes were never found. Compositing two conversations into one, which is the form's version of frankenbiting and destroys the only claim it has. And the film that keeps rolling through a moment where a person needed help, then uses the footage.

**The kit (ranked):**

1. **What They Do Alone** (Character) `char.alone_behaviour`
   The form's central instruction, not a B-roll note. Everything a vérité film knows about a person it learned from unobserved behaviour.
2. **The Value Turn** (Scene) `scene.value_turn`
   Found, never written. The discipline is recognising the turn in twelve hours of rushes and then refusing to cut it tighter than it plays.
3. **Objective Camera-Eye** (POV) `pov.objective_camera`
   The stance the form claims. Read the card for what the stance actually costs, because it is a choice with consequences rather than an absence of choices.
4. **Behaviour That Contradicts the Line** (Subtext) `sub.behaviour_contradicts`
   With no narration and no interview, the gap between what someone says and what their hands do is the only subtext mechanism available — and it is enough.
5. **Silence as Content** (Subtext) `sub.silence_as_content`
   The pause is the material. Cutting it to tighten a scene removes the thing the audience will remember.
6. **The Documentary Sequence** (Sequence) `seq.documentary_sequence`
   Observational films are built in runs of scenes around one process. This card is how a run gets a shape when no one wrote it a shape.
7. **Access as Structure** (Information) `info.access_as_structure`
   What you were allowed to film determines what the film can argue. Map the access before you map the story, because the access is the story's skeleton.
8. **Emotion Through Task** (Emotion) `emo.emotion_through_task`
   Someone doing work carefully is the form's default image and its most reliable emotional channel, because it requires nothing to be staged.
9. **The Institutional Antagonist** (Antagonist) `anta.institutional`
   A school, a hospital, a benefits office. The opposition is a system and the whole craft problem is making the system visible without inventing someone to blame.
10. **The Pressure That Is Not Aimed** (Antagonist) `anta.ambient_pressure`
    Weather, money, a waiting list. Pressure with no author, which is what an observational film spends most of its runtime inside.
11. **The Subject Who Will Not Say It** (Subtext) `sub.documentary_subject`
    In a form with no interview, the unsaid thing has to be staged by circumstance rather than asked for. This card is how you find the circumstance.
12. **Withhold the Reaction Shot** (Emotion) `emo.withhold_reaction`
    The restraint move, and here it is also the ethics. The edit's instinct at the moment of grief is almost always the wrong one.
13. **The Dead Scene Test** (Scene) `scene.dead_scene_test`
    In a hundred-to-one ratio the danger is keeping beautiful footage that does nothing. Run the test on every scene you love.
14. **Iterative Time** (Time) `time.iterative_habit`
    The repeated routine — the same round, the same shift, the same meal — is how observational film shows a life rather than an incident.
15. **Interesting Beats Likeable** (Character) `char.likeable_vs_interesting`
    Casting is done by choosing who to follow, and you are committing years to the choice. Warmth with nothing at stake is the expensive mistake here.
16. **Emotion Extracted From a Real Person** (Emotion) `emo.extracted_emotion`
    **This card names a failure, not a recommendation.** It is ranked here because observational method produces the failure more easily than any other: the camera that has become invisible is the camera most likely to take something nobody agreed to give. Read it against your own rushes before you lock.

---

**The visual kit (ranked):**

1. **Direct Cinema** (Phase 3 Documentary) `doc.direct_cinema`
   The form's contract: no interview, no narration, no prompting. Everything else in this kit is a consequence of it.
2. **Observational Follow** (Phase 3 Documentary) `doc.observational_follow`
   At shooting ratios of a hundred to one, the camera's position is the only editorial decision available in the moment.
3. **The Long Observational Hold** (Phase 3 Documentary) `doc.long_observational_hold`
   The shot that stays after the event. In verite it is where the actual film usually is, and it is what a two-camera crew instinctively cuts away from.
4. **Available Light / Natural Light** (Lighting) `light.available`
   Not an aesthetic here but the method. A lit verite scene is a scene the crew intervened in.
5. **Waiting for the Moment** (Phase 3 Documentary) `doc.waiting_for_the_moment`
   One to four years of shooting is this card repeated. It is a scheduling discipline as much as a visual one.
6. **Sequence Shooting** (Phase 3 Documentary) `doc.sequence_shooting`
   Wide, medium, detail, reaction - gathered in the moment because there is no second take. It is what makes a hundred to one ratio cuttable.
7. **Breathing Handheld** (Signature) `sig.breathing_handheld`
   The camera as a person in the room rather than a machine on a stand. Too steady reads as staged; too loose reads as incompetent.
8. **Follow Behind** (Phase 3 Documentary) `doc.follow_behind`
   The form's default travelling shot, and the one that keeps the subject's own attention as the frame's subject.
9. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   A verite film that cuts at conversational pace has turned observation back into an edited argument.
10. **Reflective Silence** (Phase 3 Documentary) `doc.reflective_silence`
   The beat after the event. It is the only interpretive tool the form permits itself, and it is made entirely of duration.

## Archive-Only Film

**Constraints:** The compilation film is made entirely from footage that already exists, so the budget is a rights budget. Archives licence by the second with minimums that matter at this scale — ten seconds per cut is a common commercial minimum and thirty seconds is a frequent project floor — and fees move on territory, term, media and exclusivity, with festival-only and all-media-worldwide-in-perpetuity differing by an order of magnitude. Clip packages from national archives run into four figures for a handful of shots. **Every piece of music inside the archive needs its own sync and master clearance even at ten seconds, so an eight-second newsreel with a band playing under it is three licences, not one.** Fair use or fair dealing carries a real part of the form, but E&O insurers want a lawyer's opinion letter before they will underwrite it, which makes the legal argument a production document. Orphan works have no one to ask. **And the people on screen are usually dead: they cannot consent, their descendants often can be found, and the film's honesty about that is a craft matter, not a courtesy — an archive of the dead is still an archive of people who did not choose to be in your film.**

**Grammar:** With no new footage there is no present tense, so the film's engine has to be the *arrangement*. Juxtaposition does the work that a scene does elsewhere: two shots that were never related become an argument because you put them together, and that is either the form's great instrument or its great dishonesty depending on whether the argument is true. The unit is the returning fragment — the same few seconds coming back with more context each time — because an archive film builds by accretion rather than by plot. The source's own voice is available and usually better than yours: newsreel narration, a caption, a bureaucratic form, a radio announcer. Sound is where the film gets its freedom, since the picture is fixed and the track is not. Dating is structural: an undated image is an assertion, and a film that cuts between 1931 and 1957 without telling you has just claimed a causal link it cannot evidence. And the narration, if there is any, must declare its own position, because the one thing an archive film can never do is pretend to be neutral about material it chose frame by frame.

**Palette:** 70 to 100 minutes, 0 new frames, 1 arrangement argument, 3 to 5 returning fragments, every image dated on screen, 1 declared narrating position

**Rules:** Where the brief raises archive rights, design the film around what it says can be used | Date every image on screen, including the ones whose date weakens your sequence | Treat the music inside the archive as its own element | Let the source speak in its own voice before you speak over it | Build by returning fragments rather than by chronology | Never let a juxtaposition claim a causation you cannot evidence | Say whose film this is — an archive film has no neutral position available to it | Where the brief relies on fair dealing, write its argument down | Find the descendants where you can, and say what you did when you could not | Let the track carry the movement the picture cannot

**Examples:** *Night and Fog*, *The Atomic Café*, *Los Angeles Plays Itself*, *Dawson City: Frozen Time*, *They Shall Not Grow Old*, *Final Account*-style testimony compilations, national-archive centenary films

**Avoid:** The needle-drop argument, where a shot of a crowd and a shot of a factory are cut together and the film pretends it has demonstrated something. Undated images used as mood, which is the fastest way for a rigorous archive film to become decorative. Treating archive as texture: grain and scratch as a look rather than as evidence, which is the same offence as unlabelled reconstruction in the other direction. A score instructing the audience how to feel about people who cannot answer back. And a rights plan discovered in the grade, which is how archive films get recut into worse films three weeks before delivery.

**The kit (ranked):**

1. **Archive as Reveal** (Information) `info.archive_as_reveal`
   The form's core move: the archive is not illustration, it is the evidence being produced. Structure the film around the moments a piece of footage changes what the audience thought.
2. **Non-Linear Assembly** (Time) `time.nonlinear_assembly`
   With no present tense to follow, order by argument. The sequence itself is the claim, which is exactly why it has to be defensible.
3. **Recurring Fragment** (Time) `time.recurring_fragment`
   The same seconds returning with more context. This is how an archive film builds without a plot, and it costs nothing in licensing because you have already cleared it.
4. **Double Anchor** (Time) `time.double_anchor`
   Fix the then and fix the now. Archive films lose audiences in the gap between two decades cut together without a marker.
5. **Passage Marker** (Time) `time.passage_marker`
   Mark every jump the same way. In this form an unmarked jump is not a stylistic choice, it is an implied claim.
6. **Contradiction Between Sources** (Information) `info.contradictory_sources`
   Two archives disagree about the same day. Putting the disagreement on screen is stronger than choosing, and it is also the honest option.
7. **Disputed History** (Information) `info.disputed_history`
   Where the record is contested, the film's handling of the contest is its credibility. Averaging it into one confident line is this form's version of inventing a fact.
8. **Documentary Chronology** (Time) `time.documentary_chronology`
   Decide the film's relationship to the calendar once and hold it, because an archive film's chronology is a construction and the audience can feel it being built.
9. **The Document** (Information) `info.the_document`
   The form, the ledger, the letter. A document read aloud is the cheapest and most authoritative thing an archive film owns.
10. **The Found Document** (POV) `pov.found_document`
    The viewpoint that material itself supplies. Letting the source's own framing stand is more defensible than translating it into yours.
11. **The Found Document** (Rhythm) `rhythm.found_document_rhythm`
    The same material at sentence level: period prose has a cadence, and reading it aloud at your own pace flattens it.
12. **Voiceover Against the Picture** (Subtext) `sub.voiceover_counterpoint`
    Narration that does not describe what is on screen is how an archive film thinks rather than captions.
13. **The Narrating Persona** (Voice) `voice.narrating_persona`
    Construct the "I" deliberately. A film assembled entirely from other people's images cannot pretend it has no author.
14. **The Object That Carries It** (Emotion) `emo.object_carrier`
    A thing that survived from then to now is the only bridge this form has that is not an argument.
15. **The Iceberg Principle** (Exposition) `expo.iceberg_principle`
    You will view a hundred hours for every minute used. The unused hours are what makes the used minute sound certain.
16. **Emotion Delegated to the Music** (Emotion) `emo.score_does_the_work`
    **This card names a failure, not a recommendation.** It is ranked here because an archive-only film has no performance to carry feeling and reaches for the score by default. Read it as the test every temp mix in this form has to pass.

---

**The visual kit (ranked):**

1. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   The whole film is quotation. Presenting each source in its own frame and grain is the format's basic honesty.
2. **Text on Archive** (Phase 3 Documentary) `doc.text_on_archive`
   Where the footage came from and when. In a compilation film the provenance is content, not a credit.
3. **Ken Burns** (Phase 3 Documentary) `doc.ken_burns`
   The move is the interpretation, and in an archive-only film it is one of the very few interpretive acts available.
4. **Then and Now Dissolve** (Phase 3 Documentary) `doc.then_and_now_dissolve`
   Made entirely of existing material, so the join between two eras is the film's main constructive device.
5. **Split Screen Archive** (Phase 3 Documentary) `doc.split_screen_archive`
   Two sources argued against each other in one frame, which is an essay-film move an archive film can make without narration.
6. **Photo Montage to Music** (Phase 3 Documentary) `doc.photo_montage_music`
   The sequence form of the format, and the one where the rights budget is spent fastest.
7. **Film Grain** (Format) `look.grain`
   Not added but preserved. Cleaning up archive to a modern standard destroys the thing that tells the audience it is archive.
8. **Found Material and Its Provenance** (Rights) `rights.found_material_provenance`
   Where the brief raises rights, every item carries where it is from beside it; nothing waits on it.
9. **Colour Temperature as Mood** (Colour) `color.temperature`
   Grading sources toward each other or away from each other is the choice between one film and a compilation, and it should be made deliberately.
10. **Side by Side Comparison** (Phase 7 Documentary) `doc.side_by_side_comparison`
   The claim shown with both halves visible, which is the archive film's alternative to asking the audience to trust the cut.

## Nature and Natural History

**Constraints:** The blue-chip strand is the reference format: roughly 50 minutes an episode, multi-year shoots, camera traps, hides and remote rigs, and budgets that only major broadcasters and streamers carry. The script is written to a picture-locked edit rather than the other way round, so the writing job is narration against an existing cut at about 150 words a minute, with long stretches carrying no words at all. **That figure is this form's, not the package's documentary rate** — it is fast because the words are fitted into gaps in a locked picture rather than setting the length of anything, which is the opposite relationship to a scripted documentary. For scripted documentary and heritage narration, where the words decide the runtime, take the rate from `pacing.wpm_documentary`, which owns that figure for the whole package. Two different jobs, two different numbers, and the difference is which one is the clock. There are no interviews and no contributors — the narration is the film's only voice, and it is frequently re-voiced in dozens of territories, which makes idiom, wordplay and complex subordination expensive. **Staging and captive footage are this genre's standing integrity problem, and the record shows disclosure tends to follow allegations rather than precede them, which is exactly the wrong order.** The making-of segment has become the convention that discharges some of this, and it works only when it is honest about what it is discharging. Conservation content is now expected by commissioners and resented by audiences when it arrives as a lecture.

*(Speaking rates are owned by [`../phase-04-audio-narration/pacing-cadence-wpm.md`](../phase-04-audio-narration/pacing-cadence-wpm.md); this file cites them and does not restate them.)*

**Grammar:** A natural history sequence is a story with a subject who cannot be interviewed, cannot be blamed and does not know it is in a film — which removes motive, dialogue and intention in one stroke and leaves behaviour, stakes and time. The unit is the attempt: an animal wants something, the world is indifferent to whether it gets it, and the sequence turns on the attempt succeeding or failing. That indifference is the genre's real antagonist and the thing that separates it from a fable. Anthropomorphism is the permanent temptation and the permanent test — you are allowed to name the behaviour and the cost, and not allowed to name the feeling. Narration is the thinnest possible layer over the picture: the cut has already told the story, and every sentence that describes what is visibly happening is a sentence spent. Specificity is the credibility instrument, as it is in heritage work — the actual distance, the actual temperature, the actual number of days. And the conservation turn, when it comes, has to be the sequence's own consequence rather than a coda taped to the end.

**Palette:** 45 to 50 minutes an episode, roughly 150 words a minute over about a third of the runtime, 3 to 5 sequences, 1 attempt per sequence, 1 indifferent force, 4 to 8 proper nouns

**Rules:** Write to the locked cut, never ahead of it | Say nothing the picture is already saying | Name the behaviour and the cost, never the feeling | Give each sequence one attempt with one outcome | Make the indifference of the world the opposition, not a predator's malice | Earn the conservation point out of the sequence rather than adding it after | Disclose staged, captive and composite footage before anyone asks | Write lines that survive re-voicing into thirty languages — no idiom, no pun, no nested clause | Spend your proper nouns on the species that carries the episode | Leave the biggest moment unnarrated

**Examples:** *Planet Earth*, *Blue Planet II*, *Frozen Planet*, *Life*, the BBC *Natural World* strand, *Our Planet*, *My Octopus Teacher*, presenter-led wildlife series and their making-of diary segments

**Avoid:** Narration that describes the picture, which is the first-draft failure in every script in this genre and is audible instantly. Motive assigned to an animal — the mother who "decides", the rival who "wants revenge" — which converts a documentary into a fable and is unrecoverable once the audience notices. Staged or captive footage cut as wild, disclosed only after a newspaper asks. A conservation coda bolted to an episode that spent 48 minutes presenting an untouched world with no humans in it. And peril manufactured in the edit for a sequence whose real outcome was uneventful.

**The kit (ranked):**

1. **Nature as Opposition** (Antagonist) `anta.nature`
   The genre's foundational card. The obstacle is not cruel, which is harder to write than cruelty and is the whole difference between this and a fable.
2. **The Indifferent Force** (Tension) `tens.indifferent_force`
   Cold, tide, drought, distance. Pressure that is not aimed at anyone is what a natural history sequence is actually about.
3. **The Story Spine** (Structure) `struct.story_spine`
   Once / every day / until one day. The six-line spine is exactly the shape of a wildlife sequence and the fastest test of whether the footage contains one.
4. **Rule by Demonstration** (Exposition) `expo.rule_by_demonstration`
   The behaviour explains itself on screen in four seconds. Any rule the picture can show is a rule the narration must not state.
5. **The Ticking Scene** (Scene) `scene.ticking_scene`
   Tide, daylight, a hatching window, a migration date. The clock is usually real in this genre, which makes it the most honest tension available.
6. **Dread of the Known Outcome** (Tension) `tens.known_outcome_dread`
   The audience can see how this ends from the first shot of the sequence. Play the dread rather than pretending to withhold.
7. **The Seasonal Cycle** (Time) `time.seasonal_cycle`
   The genre's native structure at episode and series scale, and the only chronology that needs no explaining.
8. **Rhythm in Voice-Over and Audio** (Rhythm) `rhythm.vo_breath_line`
   Write to the breath and to the cut. A line the narrator cannot land inside its shot will be rushed, and rushed narration is what the audience hears as hype.
9. **Authority by Specificity** (Voice) `voice.authority_by_specificity`
   Four hundred kilometres, eleven days, minus forty. The number is the credibility, and it is also what stops the sentence becoming an adjective.
10. **What Actually Survives Translation** (Rhythm) `rhythm.translation_survivors`
    This script will be re-voiced for dozens of markets. Write the version that survives it rather than the version that wins in English.
11. **Non-Human Viewpoint** (POV) `pov.nonhuman_pov`
    The card that marks the line this genre keeps crossing: how close you may get to an animal's experience before you have invented it.
12. **Earn the Situation First** (Emotion) `emo.earn_the_situation`
    The genre asks an audience to care about an animal it met ninety seconds ago. This card is the mechanism that makes that legitimate.
13. **Scale as the Threat** (Tension) `tens.scale_dread`
    Ice sheets, ocean systems, continental drought. Where the conservation argument lives, and where it can be made without a lecture.
14. **The Proper Noun Budget** (Exposition) `expo.proper_noun_budget`
    Latin names, place names, ocean currents. Spend them on the one the episode is built around and let the rest go.
15. **Descriptive, Not Prescriptive** (Theme) `theme.descriptive_not_prescriptive`
    Show the consequence and let the audience draw the conclusion. Prescription is what turns this genre's viewers into people who mute the last two minutes.
16. **The Tacked-On Moral** (Theme) `theme.tacked_on_moral`
    **This card names a failure, not a recommendation.** It is ranked here because the commissioned conservation coda is precisely this failure, arriving by contract. Read it as the test for whether your last two minutes grew out of the episode or were stapled to it.

---

**The visual kit (ranked):**

1. **Waiting for the Moment** (Phase 3 Documentary) `doc.waiting_for_the_moment`
   Multi-year shoots exist for single events. The whole production method is this card.
2. **Ultra Slow Motion on the Decisive Moment** (Phase 3 Documentary) `doc.ultra_slow_decisive_moment`
   The blue-chip strand's central offer: showing something at a speed no eye can achieve, which is the format's argument for existing.
3. **Macro Lens** (Lens) `lens.macro`
   The other half of the same offer, at the other end of the scale.
4. **Telephoto Compression** (Lens) `lens.tele_compression`
   Long glass is not a choice here, it is the working distance. What is a choice is what the compression does to the animal's relationship with its landscape.
5. **Golden Hour** (Lighting) `light.golden_hour`
   The strand shoots at the hours the animals move, which happen to be the hours the light is best. That coincidence is most of the format's look.
6. **Establishing Drone** (Phase 3 Documentary) `doc.establishing_drone`
   The scale shot the format now opens on by convention - and the convention is recent enough that using it once still registers.
7. **Landscape Breather** (Phase 3 Documentary) `doc.landscape_breather`
   Fifty minutes of predation needs rests, and in this format the rest is also the habitat, which is information.
8. **Contact Shadow and Ground Truth** (Materials) `material.contact_and_shadow`
   Where any element is composited or generated, this is the first thing that gives it away, and natural history is held to a documentary standard.
9. **Deep Focus** (Lens) `lens.deep_focus`
   Ranked here as the counter-move. The strand defaults to isolating the animal, and the frames that show it inside its habitat are the ones that teach.
10. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   The behaviour, at its own length. Cut to a broadcast rhythm, an animal's real pace stops being observable.

## Sports Documentary

**Constraints:** The footage belongs to a rights-holder — a league, a federation, a broadcaster — which means the archive licence and the editorial access are usually the same negotiation with the same counterparty. *The Last Dance* is the instructive case: the league owned more than five hundred hours shot in 1997–98, and the arrangement was that neither the league nor Jordan could use it without the other's permission, so the film existed only once the subject had effective control. **That is the format's structural problem stated plainly — the person the film is about frequently holds a veto, and the audience cannot see the veto.** Highlight clips licence per-use and per-territory. Participants are under contract to organisations that can decline to make them available. The dominant shape is four to ten episodes, or a 90-minute single; co-financing by the league's own media arm is now normal. And the result is known: the audience can look it up in four seconds, so no version of suspense that depends on hiding it will survive.

**Grammar:** A sports documentary is a known-ending structure, which sounds like a limitation and is actually the form's engine. Because the outcome is fixed, the question is never *what happens* but *what it cost*, and the film's job is to locate the price. The unit is the season or the campaign, with a real clock already built in — fixtures, rounds, a final date — so the serial pressure is free and the film's own work is the interior. Opposition is a rival rather than an enemy, and the best films in the form understand that the rival is the only person who fully understood the protagonist. Competence is the subject: how someone got good, and what getting good removed from their life. The archive is contemporaneous, which makes the dual timeline almost automatic — then and now, with the now containing someone who knows how it ended. And the authorised-film problem has to be answered inside the film, by putting the strongest case against the subject on screen, because the audience already suspects the deal exists.

**Palette:** 4 to 10 episodes or 90 minutes, 1 season as the spine, 1 known result, 1 rival, 1 named cost, 2 timelines, 1 concession the subject would rather not make

**Rules:** Build on what it cost, never on what happened | Put the strongest case against your subject inside the film, because the audience assumes you were not allowed to | Use the fixture list as the clock and stop inventing one | Make the rival a person who understood them, not an opponent to be beaten | Name the price of the competence in the first episode | Date every piece of archive and every interview | Say on screen who financed the film and who approved the access | Let the loss run at full length; it is what the win is measured against | Choose the last line from the whole corpus, not from the final interview | Never edit a result to feel uncertain when the audience knows it was not

**Examples:** *The Last Dance*, *Senna*, *Icarus*, *Formula 1: Drive to Survive*, *All or Nothing*, *Sunderland 'Til I Die*, *Free Solo*, federation-financed campaign series

**Avoid:** The authorised hagiography, recognisable by the absence of anyone who disliked the subject and by a rival who is only ever a foil. Manufactured suspense around a result the audience has known for twenty years, which insults them in the first act. The film that never says the league paid for it. Rivals interviewed only to praise. And the redemption beat imported from elsewhere — the late reconciliation scene that the record does not support but the edit badly wants.

**The kit (ranked):**

1. **Structuring a Known Ending** (Information) `info.known_ending_structure`
   The governing card. Everything else in this format follows from accepting that the result is not the question.
2. **Access as Structure** (Information) `info.access_as_structure`
   Who let you film, and on what terms, has already written the first draft. Map it before you outline, and then work out what the map is hiding.
3. **Dread of the Known Outcome** (Tension) `tens.known_outcome_dread`
   Tension built from watching it arrive rather than from pretending it might not. The only honest suspense this format has.
4. **The Sports Final** (Endings) `end.sports_final`
   The specific card for the specific problem: how to shoot and cut a result the audience has memorised so that it still lands.
5. **Write the Opponent's Case** (Antagonist) `anta.write_the_case`
   In an authorised film this is not balance, it is the only defence against the audience's correct suspicion. Build the case against your subject and find where it lives on screen.
6. **The Rival Who Is Not an Enemy** (Antagonist) `anta.rival_not_enemy`
   The rival made the protagonist, and knew them better than the teammates did. Films that grasp this are the ones people rewatch.
7. **Rivalry** (Situations) `sit.tobias_rivalry`
   Naming the situation the series is actually dramatising stops it drifting into a highlights package with interviews attached.
8. **Competence Has a Price** (Character) `char.competence_cost`
   The genre's real subject. What the training took, who was not there, what was given up — this is the material a results reel cannot hold.
9. **Dual Timeline** (Time) `time.dual_timeline`
   Contemporaneous archive against present-day interview. Almost automatic here, which means the discipline is in deciding when the now is allowed to comment.
10. **What Breaks First** (Character) `char.what_breaks_first`
    Under a season of pressure something gives — the body, the marriage, the friendship. Finding it in the rushes tells you where the series turns.
11. **Pride** (Emotion) `emo.pride`
    The emotion this format runs on and the one it handles worst, because pride written without its cost reads as a commercial.
12. **The Group Under Pressure** (Relationship) `rel.group_under_pressure`
    A squad, a pit crew, a coaching staff. The ensemble under a shared deadline is the format's second unit after the individual.
13. **Multi-POV Architecture** (POV) `pov.multi_pov_architecture`
    Decide whose season each episode is. Without it a sports series is a chronology with talking heads over it.
14. **Interesting Beats Likeable** (Character) `char.likeable_vs_interesting`
    Access pressures you toward the cooperative contributor. The cooperative contributor is very often the least interesting person in the room.
15. **The Value Landed** (Endings) `end.value_landed`
    Name what changed and at what cost. A sports film that only reaches the trophy has stopped one scene early.
16. **The Rigged Argument** (Theme) `theme.rigged_argument`
    **This card names a failure, not a recommendation.** It is ranked here because a rights-holder-financed film is structurally exposed to it — only one side was made articulate, and the access agreement is why. Run the diagnostic on your own cut before a critic does.

---

**The visual kit (ranked):**

1. **Athlete Portrait in Slow Motion** (Phase 3 Documentary) `doc.athlete_portrait_slowmo`
   The format's one owned image, and the shot the rights-holder's archive cannot supply.
2. **Super Slow Replay** (Phase 3 Documentary) `doc.super_slow_replay`
   The decisive moment at a speed the broadcast never showed, which is the main thing a documentary can add to footage everyone has seen.
3. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   The footage belongs to a league or a broadcaster and arrives in four decades of formats. Framing them honestly is also a rights-tracking discipline.
4. **Bench Reaction** (Phase 3 Documentary) `doc.bench_reaction`
   Where the story is, and where the live broadcast was not pointing. It is the documentary's structural advantage over the game.
5. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   Athletes interviewed in a hotel room look like everyone else. Placed where they train, they are the subject rather than a talking head.
6. **Crowd Reaction Cut** (Phase 3 Documentary) `doc.crowd_reaction_cut`
   The stakes made visible. In a sports documentary the crowd is the chorus and it carries what the commentary states.
7. **Stats Card** (Phase 3 Documentary) `doc.stats_card`
   Built from real numbers, not generated. In sport the audience knows the figures and a wrong one destroys the film's authority instantly.
8. **Then and Now Dissolve** (Phase 3 Documentary) `doc.then_and_now_dissolve`
   The form is about careers, which means it is about the same face at two ages.
9. **Lighting Ratio** (Lighting) `light.ratio`
   The interview look that separates a sports documentary from a promotional video made by the same league.
10. **Stadium Drone Reveal** (Phase 3 Documentary) `doc.stadium_drone_reveal`
   Ranked last because every sports documentary now opens with it, and the audience has learned to wait for it to end.

## Music Documentary

**Constraints:** Two rights regimes run at once and both are expensive. Every piece of music needs a sync licence from the publisher and a master licence from the label, and archival use of period music has been quoted in the fifteen-to-twenty-five-thousand-dollar range per side, with a most-favoured-nations clause the standard device for keeping a large cue sheet from spiralling. **Archive that has music playing under it needs both clearances even when it runs for ten seconds**, which means a documentary about a musician is a clearance project with a film attached. The estate or the artist usually controls the recordings, the archive and the interview access simultaneously, so the licence and the edit are one negotiation — and an artist who does not like the cut can price the film out of existence without ever refusing it. Concert footage carries its own performer and venue rights. The working shape is 90 to 120 minutes or three to six episodes.

**Grammar:** The performance is the argument. A music documentary works when the songs are evidence rather than illustration, which means a sequence has to earn its cue by having shown you what the song is answering. The form's real subject is almost never the music; it is the cost of making it, and the reliable shape is an ascent, a price, and someone who paid it who was not the artist. Bands are ensembles with an expulsion in them, and the expulsion is usually the film's midpoint whether or not anyone will discuss it on camera. Two people remember the same night differently and both are telling the truth — that is the form's most productive material and the reason it needs more than one contributor per event. The audience arrives with a version of this person already assembled from fandom, and the film's turn is very often the moment its own evidence contradicts that version. And the ending belongs to a song, which is the one structural gift this format has: the last chorus can land what a summary cannot.

**Palette:** 90 to 120 minutes or 3 to 6 episodes, 12 to 25 cleared cues, 1 ascent, 1 price, 1 expulsion or rupture, 2 accounts of the key night, 1 closing song

**Rules:** Clear the cue sheet before you lock, and design the cut around what you can afford | Earn every song by showing what it answers | Build on the cost, not on the catalogue | Get two accounts of the night that matters and play both | Put the rupture at the midpoint and let it stay unresolved | Let a performance run at full length once | Say who controls the archive and who approved the access | Never let the estate's version stand as the only version | Choose the closing song for what it lands, not for what charted | Put the strongest criticism of the artist in the artist's own era, not in a critic's hindsight

**Examples:** *Amy*, *Searching for Sugar Man*, *20 Feet from Stardom*, *Summer of Soul*, *The Beatles: Get Back*, *Dig!*, *Miss Americana*, artist-estate authorised feature documentaries

**Avoid:** The extended promo, where the estate controls everything and the film's only finding is that the artist was great. Songs used as mood under montage, which spends the clearance budget on wallpaper. The catalogue tour — one hit per decade, chronologically, with a talking head saying it was a classic. A redemption arc in the last twenty minutes that the record does not support and the family requested. And the critic brought in to say in hindsight what nobody was willing to say at the time, which reads as the film outsourcing its own courage.

**The kit (ranked):**

1. **The Song's Claim, and the Video's** (Theme) `theme.song_claim`
   Start here. What the song argues, and whether the film agrees with it, is the question a music documentary is actually answering.
2. **Access as Structure** (Information) `info.access_as_structure`
   Estate, label and artist usually hold everything at once. What you were given determines what the film can say, so map it before you outline.
3. **The Emotional Beat of a Lyric and Its Picture** (Emotion) `emo.song_emotional_beat`
   The mechanism of every performance sequence: which line the cut lands on, and what the picture is doing when it arrives.
4. **The Set Piece** (Sequence) `seq.set_piece`
   A performance played at full length, once, is this format's largest structural object. Build it deliberately and place it at the turn.
5. **The Same Event Twice** (POV) `pov.same_event_twice`
   Two band members on the same night. The contradiction is the material, and a film with only one account of its own midpoint has not reported it.
6. **The Group That Expels a Member** (Relationship) `rel.expulsion`
   Almost every band film contains one. Naming the structure stops the film treating a firing as a scheduling detail.
7. **The Rivalry That Is a Friendship** (Relationship) `rel.rivalry_as_friendship`
   The songwriting pair who cannot work together and cannot work apart. The form's most durable two-hander.
8. **Competence Has a Price** (Character) `char.competence_cost`
   What the craft cost, and who paid the part of it that the artist did not. This is where a music film stops being a discography.
9. **The Fall Arc** (Arc) `arc.fall`
   The shape a great many of these films are actually in. Naming it early stops the last act arriving as a surprise to the filmmakers.
10. **The Community's Version of a Person** (Information) `info.communal_version`
    The audience brings a fandom's version into the room. Knowing precisely what it is tells you which piece of evidence will land hardest.
11. **The Interview Admission** (Information) `info.interview_admission`
    The sentence a contributor did not plan to say. In this format it is usually the only thing that moves the received account.
12. **Dual Timeline** (Time) `time.dual_timeline`
    The era and the present, with a present that knows how it ended. The genre's default and the reason it sustains six hours.
13. **The "Based on True Events" Frame** (Information) `info.true_story_frame`
    Music films run on legend. This card is how you signal the difference between what is documented and what the scene has always said.
14. **Nostalgia Without Sentimentality** (Emotion) `emo.nostalgia`
    The failure mode this genre lives inside. Read the card as a checklist against your own cut, especially the archive-and-ballad sequence.
15. **The Last Chorus** (Endings) `end.last_chorus`
    The structural gift: a song can land a value a summary cannot. Choose it for the argument, not for the catalogue's biggest track.
16. **The Unearned Redemption** (Antagonist) `anta.unearned_redemption`
    **This card names a failure, not a recommendation.** It is ranked here because the late reconciliation beat is what estates ask for and what edits supply. Test the last twenty minutes against it.

---

**The visual kit (ranked):**

1. **Pit Handheld** (Phase 3 Documentary) `doc.pit_handheld`
   The performance from where the photographers stand. It is the format's most recognisable texture and it cannot be recreated later.
2. **Instrument Close-Up** (Music video) `mv.instrument_closeup`
   Hands on strings and keys. It is the proof of craft that an audience needs before it will accept a claim about an artist.
3. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   Two rights regimes run at once and both are expensive, so every frame of archive is tracked and framed as what it is.
4. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   The studio, the tour bus, the childhood room. Music documentary contributors are placed more meaningfully than in any other strand.
5. **Stage Light Cues** (Music video) `mv.stage_light_cues`
   The concert's own design, respected rather than overridden. Fighting the lighting designer is how a performance sequence stops looking like a performance.
6. **Crowd Energy Wide** (Music video) `mv.crowd_energy_wide`
   The scale of the thing being claimed. It is also the shot most often unusable because of where the cameras were allowed.
7. **Multicam Switch on the Beat** (Phase 7 Documentary) `doc.multicam_switch_on_beat`
   The cut belongs to the music, which is the one rule a music documentary cannot break without the audience feeling it.
8. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   A song played in full, once, somewhere in the film. Every music documentary that cuts every performance has forgotten why anyone came.
9. **Found Material and Its Provenance** (Rights) `rights.found_material_provenance`
   Where the brief raises music rights, sync and master are two separate facts on every piece of music; note both.
10. **Then and Now Dissolve** (Phase 3 Documentary) `doc.then_and_now_dissolve`
   The career shape again, and in music documentary it is usually the film's whole thesis in one transition.

## First-Person Diary Film

**Constraints:** Usually one person with a camera over years, shooting before there is a film — which means the material was gathered without a structure and the structure has to be discovered in it afterwards. Funding arrives retroactively, if at all, and the film is frequently finished at a residency or a lab rather than in a production office. Consumer formats change over the shooting period, so the film carries its own technological history on its surface. Festivals are the distribution, and personal-essay strands at documentary festivals are the natural home. **The ethical load is unusual because the filmmaker is also a subject and cannot be protected by the usual arrangements: the other people in the film are family and friends who were being filmed long before there was a project, who cannot give informed consent to a film that did not exist, and whose refusal costs the maker a relationship rather than a scene.** The honest version of this format re-asks, late, and shows what the answer was.

**Grammar:** The diary film's engine is a person changing their mind on camera over years, so the form's material is not events but revisions. The "I" is constructed even when it is sincere, and designing that persona deliberately is what separates a diary film from footage. Doubt is structural rather than decorative: the narration must be able to be wrong, and the film's credibility comes from the moments it corrects itself rather than from the moments it is certain. Two tenses run at once — the present of the shooting and the retrospect of the voice recorded years later — and the distance between them is where the meaning sits. The film is usually organised around a question the maker could not answer at the time and can now at least state. And the hardest scene in every diary film is the one where the camera should have been put down; the film has to account for that scene rather than use it.

**Palette:** 70 to 95 minutes, 1 maker, 5 to 20 years of material, 2 tenses, 1 question that changed, 1 admitted error, 1 accounted-for scene, an open ending

**Rules:** Design the "I" — sincerity is not a substitute for a constructed persona | Keep the two tenses audibly separate: what you thought then, what you know now | Put the revision on screen; the film's authority is in being wrong first | Build on the question, not on the chronology | Re-ask everyone in it, late, and show the answer you got | Account for the scene where the camera should have been down, in the film | Do not narrate a feeling the footage already carries | Let the format degradation stand; it is the film's clock | Choose the ending for the question, not for the last thing you shot | Cut the material that only exists because you had the camera and they did not object

**Examples:** *Cameraperson*, *Tarnation*, *Stories We Tell*, *Dick Johnson Is Dead*, *A Family Affair*, Jonas Mekas's diaries, *Time*, personal-essay strands at IDFA and Sheffield

**Avoid:** Footage presented as a film, where twenty years of material is ordered chronologically and the maker hopes the accumulation will supply a shape. A narrating voice that was certain all along, which turns a diary into a memoir and removes the only thing the form has. Family members used as material because they are available, with the consent conversation happening after the premiere or not at all. The tragedy that arrives unearned because the film withheld it as a structural device. And an open ending standing in for a conclusion the maker did reach but did not want to say aloud.

**The kit (ranked):**

1. **First Person Central** (POV) `pov.first_person_central`
   The form's condition. The maker is a subject, not a narrator standing outside the material, and every craft choice follows from accepting that.
2. **The Narrating Persona** (Voice) `voice.narrating_persona`
   Sincerity does not construct a voice. Designing the "I" is the difference between a diary film and a hard drive.
3. **The Self-Correcting Narrator** (Voice) `voice.self_correcting_narrator`
   Revising a claim on screen is where this form's credibility comes from. A diary film that is right throughout is a memoir pretending.
4. **Unstable Authority** (Voice) `voice.unstable_authority`
   A voice that can be wrong, and that the audience can see being wrong. Without it the form collapses into explanation.
5. **Diary and Log Time** (Time) `time.log_time`
   The native chronology: dated entries, uneven intervals, long silences. The card is about making the gaps mean something rather than apologising for them.
6. **The Retrospective Narrator** (POV) `pov.retrospective_distance`
   The second tense. The voice recorded years later knows what the footage does not, and the gap between them is the film's main instrument.
7. **Start From a Question** (Theme) `theme.start_from_a_question`
   Order the material by the question rather than by the calendar. It is the pass that turns an archive into a film.
8. **Non-Linear Assembly** (Time) `time.nonlinear_assembly`
   Twenty years in chronological order is a timeline. Ordering by argument is the only way this form finds a middle.
9. **Recurring Fragment** (Time) `time.recurring_fragment`
   The same few seconds returning as the maker's reading of it changes. This is how a diary film shows a mind moving.
10. **The Subject Who Will Not Say It** (Subtext) `sub.documentary_subject`
    In a family film the person who will not say it is usually the one the film is about. The card is how to build the scene where their refusal is legible.
11. **Parent and Adult Child** (Relationship) `rel.parent_and_adult_child`
    The relationship most diary films are actually conducting, including the ones that think they are about something else.
12. **The Confiding Voice** (Voice) `voice.confiding_voice`
    One person to one person over ninety minutes. It is the register that makes a stranger's private life bearable to sit with.
13. **Numbness and Dissociation** (Emotion) `emo.numbness`
    The honest emotional register of much of this material, and the one edits reach past on the way to something more legible.
14. **The Secret** (Character) `char.the_secret`
    Diary films are usually built around one, and it is usually the maker's. Placing it is most of the structural work.
15. **The Open Ending** (Endings) `end.open_ending`
    Legitimate here in a way it rarely is elsewhere — provided the openness is the film's finding rather than its evasion.
16. **Emotion Extracted From a Real Person** (Emotion) `emo.extracted_emotion`
    **This card names a failure, not a recommendation.** It is ranked here because the diary filmmaker's subjects are family, the camera has been present for years, and the extraction is therefore both easiest and least visible. Read it against the three scenes you are proudest of.

---

**The visual kit (ranked):**

1. **Selfie Walk** (Phase 3 Documentary) `doc.selfie_walk`
   The form's native shot. Shot before there was a film, which is why it is honest and why it cannot be recreated.
2. **Available Light / Natural Light** (Lighting) `light.available`
   One person with a camera over years. Any lit frame in this form is a frame from after the film was conceived.
3. **Breathing Handheld** (Signature) `sig.breathing_handheld`
   The camera is a hand, and the shake is authorship rather than error.
4. **The Long Observational Hold** (Phase 3 Documentary) `doc.long_observational_hold`
   The material was shot without a structure, so the holds are where the film's actual thinking got recorded.
5. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   Years of footage across several devices. The format changes are the timeline and should be visible rather than graded away.
6. **Talk to Camera Opener** (Phase 3 Documentary) `doc.talk_to_camera_opener`
   The address to the lens that establishes whose film this is, usually recorded long before anyone knew there was one.
7. **Reflective Silence** (Phase 3 Documentary) `doc.reflective_silence`
   The beat the diarist left in. It is the difference between a diary and a vlog.
8. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   For the other people in the film, who did not choose to be in it and are placed rather than interviewed.
9. **Last Look to Camera** (Phase 3 Documentary) `doc.last_look_to_camera`
   Nearly every diary film ends on one, because it is the only ending the form's material naturally contains.
10. **A Child in the Frame** (Rights) `rights.minor_on_screen`
   Diary films shoot families over years. Consent given by a five-year-old's parent is not consent given by the fifteen-year-old they became.

## Personal-Archive Family Film

**Constraints:** Built from home movies — 8mm and Super 8, VHS, MiniDV, camcorder tapes, phone video — which means the first line of the budget is digitisation and the second is stabilisation and grade. Copyright in a home movie sits with whoever operated the camera, frequently a dead relative, so the estate matters even when the family does not object. **No releases exist and none can now be obtained: the people in the footage are often dead, the children in it never consented and are now adults, and the family members who do survive hold a veto by relationship rather than by law — which is the stronger veto of the two.** Archivists have been explicit that public display of home footage of children needs particular care. The film's first audience is the family, and it will be watched by them before it is watched by anyone else, which is a structural fact about the writing rather than a courtesy.

**Grammar:** A home movie is evidence of what a family chose to record, which is a different thing from what happened, and the form's central move is reading that gap. Nobody filmed the argument. So the film is built on absence: what is missing from the archive is its strongest material, and naming the missing thing is the equivalent of a reveal. Memory is the unreliable narrator here — several relatives remember the same afternoon incompatibly, and the film is more truthful for keeping the incompatibility than for adjudicating it. An object carries the years better than a date card can, because the object is in the 1978 footage and also on the table now. Generational span is the natural structure, and the film's turn is usually the moment the maker realises they are filming the same way their parent did. And the ending returns to the present, because a family film that never comes back has made its own relatives into a museum.

**Palette:** 60 to 90 minutes, 3 generations, 1 named absence, 2 incompatible memories of one day, 1 object spanning the whole span, 1 return to now

**Rules:** Digitise everything before you decide anything | Build on what is missing from the archive, not only on what survives | Keep two incompatible memories of the same day rather than choosing between them | Find one object that is in the oldest footage and is still here | Ask the living relatives before the film exists, and again when it does | Treat footage of children as requiring a decision, every time | Name the person who held the camera; it is a point of view | Date every reel on screen, including the ones you would rather leave vague | Show the family the cut first and be prepared to lose a scene | Return to the present before the end

**Examples:** *Stories We Tell*, *Tarnation*, *Dawson City: Frozen Time* in its family passages, *Sarah Polley*-style memory films, diaspora family-archive documentaries, Home Movie Day and community-archive projects

**Avoid:** Home footage used as texture — sunlit grain under a score, meaning nothing, standing in for feeling the film has not produced. The single authoritative account, where one relative's version becomes the narration and the others become illustration. A film that never says who was holding the camera, and therefore never notices whose life is missing from the record. Children put on screen because the footage exists, with no one asking the adults they became. And the film that ends in 1983, leaving the audience with nostalgia rather than inheritance.

**The kit (ranked):**

1. **The Found Document** (POV) `pov.found_document`
   A home movie is a document with an author, and identifying the author changes the film. Whoever held the camera is a viewpoint, not a neutral recorder.
2. **Archive as Reveal** (Information) `info.archive_as_reveal`
   The reel is not illustration, it is evidence being produced. Structure on the moments a piece of footage changes what the family thought it knew.
3. **Unreliable Memory** (Time) `time.unreliable_memory`
   The form's engine. Relatives disagree, and the disagreement is the more accurate account of a family than either version alone.
4. **Borrowed Memory** (Time) `time.borrowed_memory`
   What you remember because you saw the footage, not because you were there. Every family film is partly made of this and should say so.
5. **The Bond With the Dead** (Relationship) `rel.bond_with_the_dead`
   The relationship the film is conducting with someone who cannot answer. The card is about keeping it a relationship rather than a eulogy.
6. **Generational Span** (Time) `time.generational_span`
   Three generations is the natural shape of a family archive, and the span is what converts private footage into an argument about inheritance.
7. **The Object That Carries It** (Emotion) `emo.object_carrier`
   A thing in the 1978 reel that is on the table today. The only bridge this form has that requires no narration at all.
8. **Recurring Fragment** (Time) `time.recurring_fragment`
   The same few seconds of a birthday returning with more context. Free to licence, because you already own it, and structurally the strongest device here.
9. **The Community's Version of a Person** (Information) `info.communal_version`
   What the family has agreed to say about someone. The film's turn is usually the point its own footage disagrees with the family's version.
10. **Double Anchor** (Time) `time.double_anchor`
    Fix the then and the now on screen. Undated home footage is where audiences get lost, and where a film accidentally implies a chronology.
11. **Oral and Heritage Narration** (POV) `pov.oral_narration`
    The relative telling it at the table. The register is the evidence, and cleaning it up is how a family film loses the thing it had.
12. **Heritage Specificity** (Exposition) `expo.heritage_specificity`
    The street name, the make of the car, the price of the house. Specificity is what stops a family film being anyone's family film.
13. **The One They Will Miss** (Character) `char.the_missed_one`
    Identify them early. A family archive has one, and the film's weight distributes around them whether you plan it or not.
14. **Subtext Across a Language Barrier** (Subtext) `sub.across_languages`
    In migrant and diaspora archives the older footage is in a language the youngest generation does not hold. That gap is the film's subject as often as it is its problem.
15. **The Audience That Will Correct You** (Information) `info.heritage_audience_correction`
    Your first audience knows every face and every date. Write as though they are watching, because they are, and they will be first.
16. **The Mirror Ending** (Endings) `end.mirror_ending`
    Return to the opening image with the present inside it. The cleanest way to make a family archive land as inheritance rather than as loss.
17. **Sentimentality** (Emotion) `emo.sentimentality`
    **This card names a failure, not a recommendation.** It is ranked here because no other documentary format produces it as reliably — sunlit grain, a piano, a dead relative laughing. Read it as the diagnostic for the cut you are most pleased with.

---

**The visual kit (ranked):**

1. **Archive Format Frame** (Phase 3 Documentary) `doc.archive_format_frame`
   8mm, VHS, MiniDV and phone video are four textures and four decades. Presented identically, the family's timeline disappears.
2. **Film Grain** (Format) `look.grain`
   Preserved rather than removed. The grain is the date stamp, and restoring it away costs the film its chronology.
3. **Ken Burns** (Phase 3 Documentary) `doc.ken_burns`
   The move across a photograph is the only camera the film has, which makes it the director's main instrument rather than a filler.
4. **Then and Now Dissolve** (Phase 3 Documentary) `doc.then_and_now_dissolve`
   The same room, the same doorway, thirty years apart. It is the form's central figure and usually its ending.
5. **Photo Montage to Music** (Phase 3 Documentary) `doc.photo_montage_music`
   The sequence the material supports. Its risk is sentimentality, and the defence is specificity rather than restraint.
6. **Text on Archive** (Phase 3 Documentary) `doc.text_on_archive`
   Who, where and when. Family archive arrives unlabelled and the labelling is most of the research.
7. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   The surviving relatives, filmed in the places the archive was shot in.
8. **Consent to Film, Record or Quote** (Rights) `rights.cleared_before_locked`
   A home movie has many people in it and none of them agreed to a film; where the brief raises it, note it with the brief. The cut does not wait on it.
9. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   A home movie shot runs at its own length. Cut to a modern rhythm it stops being a memory and becomes footage.
10. **Colour Temperature as Mood** (Colour) `color.temperature`
   Every decade of consumer film has its own cast. Grading them together is a decision about whether this is one family or four eras.

## The Documentary Short

**Constraints:** The Academy defines a short as 40 minutes or less including credits, and its Documentary Short rules run on that line; Sundance takes US shorts under 50 minutes including credits. Most commissioning strands buy far less — 10 to 25 minutes is the working band for online and broadcast documentary shorts, and the *New York Times* Op-Docs tradition sits around 10 to 15. Festival programmers build blocks, so a short competes for a slot against five other films and a running order. Budgets are small enough that the film is often one or two contributors in one or two locations, shot in days. A short frequently exists as a calling card for a feature, which is a legitimate purpose and a dangerous one: a short built as a trailer for a longer film is not a short. The ethics do not scale down — consent, contradicting material and signalled reconstruction apply at fifteen minutes exactly as they do at ninety.

**Grammar:** A short is one question with one turn, and everything that is not load-bearing has already been cut. The commonest failure is a compressed feature: three threads, four contributors and a chronology, at a quarter of the length, so nothing has room to turn. The right shape is closer to a single sequence — one situation, entered late, that changes. Character has to be established in seconds rather than scenes, so the entrance does the work an act would do elsewhere. Exposition is front-loaded into the first thirty seconds or abandoned, because a short cannot afford to explain twice. The value turn is usually placed around two-thirds through, and the film ends within a minute of it: a short that winds down loses in a festival block, where the next film starts in four seconds. And the epilogue that a feature earns is not available here — title cards about what happened next are the form's most common way of admitting the film did not end.

**Palette:** 10 to 25 minutes typically, 40 as the Academy ceiling, 1 question, 1 or 2 contributors, 1 location, 1 turn at roughly two-thirds, an ending within 60 seconds of the turn

**Rules:** One question, and cut the second one even when it is good | Enter the situation late and leave within a minute of the turn | Establish the person in one action, not in a biography | Front-load whatever is needed to read the turn into the first thirty seconds | Cap the contributors at two unless the third one turns something | Choose the last line from the whole transcript, not from the last thing anyone said | Cut the title cards — if the film needs them to end, it has not ended | Treat consent, contradiction and reconstruction exactly as a feature would | Test the film cold in a block of five, not alone in your own edit suite | Decide whether this is a short or a pitch for a feature, and do not deliver both

**Examples:** Academy Documentary Short Film nominees, *NYT Op-Docs*, *The Guardian* documentary shorts, *Field of Vision* commissions, *A Concerto Is a Conversation*, *The Queen of Basketball*, festival short-doc blocks at Sundance and Full Frame

**Avoid:** The compressed feature — three threads, a chronology and a coda, none with room to turn, which is what a feature outline becomes when the funding does not arrive. The trailer disguised as a film, where the short exists to raise money and therefore never resolves anything. Title cards doing the ending's job. An emotional subject substituting for a structure, which is why so many short docs about grief have no second act. And an opening that spends ninety seconds establishing a world the film only needs for one scene.

**The kit (ranked):**

1. **Start From a Question** (Theme) `theme.start_from_a_question`
   One question, stated in the first minute and answered or honestly abandoned in the last. A short with two questions has neither.
2. **Sequence as Short Film** (Sequence) `seq.short_film_model`
   The structural key: a short documentary is one sequence with a turn, not a miniature feature with three acts squeezed into it.
3. **The Value Turn** (Scene) `scene.value_turn`
   The single turn the whole film is built to deliver. Place it at roughly two-thirds and build backwards from it.
4. **The Documentary Open** (Openings) `open.documentary_open`
   The first thirty seconds are the entire orientation budget. This card is what to spend them on.
5. **The Question Planted First** (Openings) `open.question_planted_first`
   The viewer needs something unresolved in mind before the first piece of evidence arrives, or the evidence has nowhere to land.
6. **A Person in Ten Seconds** (Character) `char.ten_second_person`
   There is no second scene to establish anyone in. One action, one object, one sentence — and the audience holds them for the rest of the film.
7. **Short-Form Front-Loading** (Exposition) `expo.short_form_frontload`
   Everything needed to read the turn goes in early or does not go in. A short cannot afford to explain anything twice.
8. **Late Entry** (Scene) `scene.late_entry`
   Enter every scene as late as it will bear, and the film as late as the question allows. In this length it is the main compression instrument.
9. **Documentary Reveal Structure** (Information) `info.doc_reveal_structure`
   How a short paces what the audience learns when there is only room for two or three movements of knowledge.
10. **The Specific Detail** (Emotion) `emo.specific_detail`
    A short has no time to build feeling, so it borrows one true particular and lets the audience do the rest.
11. **Understate at the Peak** (Emotion) `emo.understate_at_peak`
    Fifteen minutes cannot survive an underlined climax. Pull back at the moment the edit wants to push, and the short lands.
12. **Early Exit** (Scene) `scene.early_exit`
    Cut on the turn. In a festival block, winding down is the difference between the film people talk about afterwards and the one before it.
13. **The Quiet Ending** (Endings) `end.quiet_ending`
    The short's natural close: a held image, a small action, no summary. It reads as confidence because it is.
14. **The Last Line** (Endings) `end.last_line`
    Chosen from the whole corpus rather than inherited from the final interview. In a fifteen-minute film it carries a disproportionate share of the meaning.
15. **The Documentary Ending** (Endings) `end.documentary_ending`
    What a documentary owes its audience at the close, at any length. Read it before you reach for a title card.
16. **Interesting Beats Likeable** (Character) `char.likeable_vs_interesting`
    With one or two contributors, casting is the film. There is no ensemble to compensate for a warm person with nothing at stake.
17. **The Ending Slideshow** (Endings) `end.epilogue_slideshow`
    **This card names a failure, not a recommendation.** It is ranked here because the short documentary produces it more often than any other format — the film stops, and four title cards do the ending's work. Test your last minute against it.

---


<!-- Interactive Documentary, Interactive Testimony and Museum and Installation Narrative
     were first written into genres-nonfiction.md. They are formats, not genres, and they
     already carried this file's fields and its three ethical constraints, so they moved
     here rather than being reshaped. -->

**The visual kit (ranked):**

1. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   One contributor, placed. A short has no time to establish who someone is, so the frame does it.
2. **Nat Sound Open** (Phase 3 Documentary) `doc.nat_sound_open`
   Begins in the place. A short that opens on an interview has spent a tenth of its runtime on context.
3. **Sequence Shooting** (Phase 3 Documentary) `doc.sequence_shooting`
   Wide, medium, detail, gathered in the moment. A short cannot return, so the coverage is taken once or not at all.
4. **The Long Observational Hold** (Phase 3 Documentary) `doc.long_observational_hold`
   The one shot a short can afford to hold, and the reason a fifteen-minute film can feel larger than a feature.
5. **Available Light / Natural Light** (Lighting) `light.available`
   The crew is one or two people and the access is short. Lighting is usually the thing that ends the access.
6. **Working Hands Interview** (Phase 3 Documentary) `doc.working_hands_interview`
   Someone talking while doing their work, which is how a short gets a portrait and a process in the same shot.
7. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   Counter-intuitive and correct: a short earns its length by having at least one moment it refuses to rush.
8. **Landscape Breather** (Phase 3 Documentary) `doc.landscape_breather`
   Once. It is the shot that tells the audience the film has a shape rather than a duration.
9. **Last Look to Camera** (Phase 3 Documentary) `doc.last_look_to_camera`
   Shorts end on them more than features do, because a short has no epilogue to spend.
10. **Lighting Ratio** (Lighting) `light.ratio`
   Ranked last and still worth deciding: one considered ratio is the cheapest thing that makes a short look like a film rather than a package.

## Interactive Documentary

**Constraints:** There is no runtime, so there is no length to negotiate — a piece is a node graph with a claim attached, and the only durable number is how long a complete route takes, which is usually 20 to 40 minutes. Funding comes from innovation and digital strands rather than from acquisitions, arrives before anything is playable, and almost never covers the years after launch. The team is a writer and a developer from day one, which means the script is a spreadsheet as much as a document and no structural decision is cheap after the build starts. **The ethical rules from `formats-doc.md` do not weaken here, they get harder: consent must survive a piece that never reaches a final cut, contradicting material must reach every route rather than merely existing somewhere, and reconstruction must be signalled in the node where it plays rather than once in a title card nobody is guaranteed to see.** **Preservation is a writing constraint, not an afterthought, and this form has the worst record of any in the database.** Adobe Flash reached end of life on 31 December 2020 and browsers dropped it, which silently took a generation of celebrated web documentaries off the internet; MIT Open Documentary Lab's *Update or Die* framed work "made as recently as five years ago" as facing obsolescence; the National Film Board of Canada's collaboration with Rhizome on Webrecorder exists to rescue more than a hundred of its own interactive productions. Write for that: budget a linear cut-down and a deposit copy as deliverables from the start, keep the text and the source manifest in formats that outlive the player, and treat a piece that only exists as a running site as unfinished.

**Grammar:** The unit is the node, the shape is a hub with a spine through it, and the argument lives on the spine. Build in this order and the form behaves. **One question the piece does not know the answer to**, exactly as in a feature documentary. **A claim sheet before a node graph** — every claim the piece makes, its source, and the nodes that carry it — because it is the only artifact that keeps a 200-node build honest and it is what a fact-checker, a commissioner and an archive will each ask for. **A spine of bottleneck nodes every route crosses**, carrying the counter-evidence and the things that are true whichever way the user went. **Hubs of routed material** hanging off it — accounts, places, years — written so each is complete on its own, because the order is not yours. **An arc that belongs to the visitor**, not to a protagonist, since the material cannot change and the only thing that can move is what the user knows. The failure this grammar prevents is the commonest one in the form: a beautiful interface over a set of nodes that never disagree with each other.

**Palette:** 1 question, 1 spine, 40 to 120 nodes, 3 to 8 contributors, 2 decisions deep, a 20 to 40 minute complete route, 1 linear cut-down, 1 deposit copy

**Rules:** Write the claim sheet before the node graph and keep it current | Put the counter-evidence on the spine, in full, not on a branch | Let the user choose the route and the depth, never the facts | Declare a counterfactual branch as a model before the first click, or do not build one | Label recorded, reconstructed and inferred in the node where each plays | Attach the source to the claim, not to the piece | Write every node to be somebody's first node | Name the arc as a change in the visitor and check it in testing by asking what they now think | Show the user what they did not open, by name | Budget a linear cut-down and an archival deposit copy in the first financing plan | Re-consent contributors when the structure changes, not when the copy changes | Test with someone who quits halfway, because most people do

**Examples:** *Bear 71*, *Fort McMoney*, *Hollow*, *Journey to the End of Coal*, *Prison Valley*, *Highrise / Out My Window*, *Welcome to Pine Point*, *Do Not Track*

**Avoid:** The interface that argues. Two buttons on an evidence-weighted question is an editorial claim of equivalence, and it is the form's signature dishonesty. The piece with no disagreement in it, where every node is a different person being interesting about the same thing and nothing the user finds contradicts anything else — that is a gallery, and users leave it exactly as they arrived. Branch symmetry bought with padding, which is how a thin interview becomes five minutes of press conference and a reconstruction. The thesis given a node graph: a film that already knew, rebuilt as choices, so that every route confirms it and the interactivity is a delivery mechanism for a conclusion. And the piece with no exit plan, built on a platform, a plugin or a funder's hosting, which is a documentary with an expiry date written into it by people who did not notice they were writing one.

**The kit (ranked):**

1. **The Non-Fiction Branch Contract** (Interactive) `inter.nonfiction_branch_contract`
   The rule the whole form sits on: route and depth are the user's, facts are not. Settle it before the first node or every later decision is negotiable.
2. **The Visitor's Arc** (Interactive) `inter.visitor_arc`
   The material cannot change, so the arc is the user's or there is none. This is the card that separates a piece from a collection.
3. **Start From a Question** (Theme) `theme.start_from_a_question`
   Imported wholesale from the feature documentary kit and even more load-bearing here, because a branching piece built on an answer gives the user nothing to do but browse it.
4. **The Path That Omits** (Interactive) `inter.path_that_omits`
   The documentary duty to show contradicting material, rebuilt for a form whose entire purpose is letting people not see things. Placement, not intention.
5. **Write the Opponent's Case** (Antagonist) `anta.write_the_case`
   What goes on the spine. Write the strongest case against the piece first; the node graph is then built around getting it in front of everyone.
6. **Hub and Spoke** (Interactive) `inter.hub_and_spoke`
   The form's default shape and the honest one for non-fiction: the user chooses order, the piece keeps its spine.
7. **The Storylet** (Interactive) `inter.storylet`
   How a large archive becomes navigable without a taxonomy — material gated by state rather than by a menu, which is what most serious i-docs are actually built from.
8. **Sourcing on a Branch** (Interactive) `inter.branch_sourcing`
   There is no end crawl everybody reaches. Attribution has to live in the node or it does not exist for the reader who quit at node nine.
9. **The Meaningful Choice Test** (Interactive) `inter.meaningful_choice_test`
   Applied here to a harder question than in fiction: a choice that changes only what you see is legitimate in non-fiction, and this card is how you tell that from a choice that changes nothing at all.
10. **Clicked, Therefore Understood** (Interactive) `inter.clicked_therefore_understood`
    **This card names a failure, ranked here as a standing diagnostic** — it is the defect this form produces by default, because interaction feels like comprehension and completion figures will not tell you otherwise. Run its exit question on every test user.
11. **The Symmetrical Fabrication** (Interactive) `inter.symmetrical_fabrication`
    **This card names a failure, ranked here as a standing diagnostic.** Every interactive documentary reaches the month when one branch came back thin, and the pressure to balance it is structural rather than personal. Listed so it gets checked at assembly.
12. **The Unchosen Path as Content** (Interactive) `inter.unchosen_path`
    The honest ending for a piece nobody sees all of, and the strongest replay driver the form has in non-fiction.
13. **The Platform Contract That Expires** (Interactive) `inter.platform_contract`
    The preservation problem as craft. Ranked high deliberately: most of this form's canon is already unplayable, and the decisions that prevent it are writing decisions made in year one.
14. **The Reporter's Stance** (POV) `pov.reportage_stance`
    Decide the piece's relationship to its own claims and hold it across every node, including the ones written by a developer at two in the morning.
15. **The Ending Budget** (Interactive) `inter.ending_budget`
    Non-fiction cannot afford many endings and does not need them. The card is about spending the budget on one ending that lands rather than on four that stop.
16. **The Knowledge Ledger** (Information) `info.knowledge_ledger`
    Adapted here into the claim sheet: what is documented, what is inferred, what is guessed, and which nodes carry each.

---

**The visual kit (ranked):**

1. **Accessible Interaction** (Interactive) `inter.accessible_interaction`
   There is no runtime, so there is no forced pace - which means every viewer navigates, and a navigation that needs a mouse and good sight excludes people the film is about.
2. **Walk-Up Entry** (Interactive) `inter.walk_up_entry`
   A node graph has no beginning. Every node has to work for someone arriving at it first.
3. **Chapter Card** (Phase 3 Documentary) `doc.chapter_card`
   Where am I and what else is there. In a branching piece this is navigation rather than structure.
4. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   Contributors are met out of order, so each one's frame has to introduce them without relying on what came before.
5. **Never Colour Alone** (Access) `access.not_colour_alone`
   A node map colour-coded by theme is unusable for part of the audience unless it is also coded by shape or position.
6. **Legible at the Size It Is Actually Watched** (Access) `access.legible_at_size`
   Interactive documentary is watched on a laptop and a phone with equal frequency, and its interface type is usually set for neither.
7. **Map Animation** (Phase 3 Documentary) `doc.map_animation`
   The spatial version of the node graph, and often the piece's actual index.
8. **Build It, Do Not Generate It** (Data) `dataviz.build_do_not_generate`
   Interactive work is data-led more often than linear work. A generated chart in a piece the audience can interrogate is a claim they will check.
9. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   The viewer controls the clock, so a node that rushes is a node they leave rather than one they endure.
10. **Colour Temperature as Mood** (Colour) `color.temperature`
   The most reliable way to tell a viewer which strand of a branching piece they are currently inside.

## Interactive Testimony and Oral History

**Constraints:** The contributors are the constraint. They are frequently old, frequently ill, sometimes dead before release, and their testimony is the only copy of something. Oral-history practice — the Oral History Association's ethics statement is the working standard — gives the narrator rights the documentary tradition does not: they review and approve before release, they hold copyright in their own recording, and they may restrict, redact or withdraw. Interactivity collides with all three, because there is no cut to approve. **Consent therefore has to be granular and rolling: agreement to specific clips and specific adjacencies, a withdrawal route with its real limits stated, and re-consent whenever the structure changes rather than whenever the copy does.** A question-led piece in the *Dimensions in Testimony* tradition is a filming constraint before it is a writing one: a question set in the low thousands, gathered from the audience who will actually use it, filmed over roughly five days against a fixed rig, with every answer built to stand alone. Translation and subtitling are structural, not finishing — a testimony archive in Surayt, Arabic or Kurdish needs its question set to work in each language, and a matching layer trained in one. **The absolute line: no synthesised sentence may ever be attributed to a real witness.** One generated answer in one archive damages every archive that uses the technique. And preservation is an obligation to the contributor, not a technical preference: a testimony that stops playing has broken a promise made to someone who is now beyond being asked again.

**Grammar:** The unit is the answer, not the interview. A testimony piece is assembled by a listener, so every clip has to be independently complete, independently sourced and independently survivable out of order — which is a different craft from cutting an interview and is closer to writing entries than to writing scenes. Above the clips sit two authored objects that do most of the work. **The way in**: one specific question the collection can answer, and a short written path of six to ten items that answers it, because a collection without a door is not read by anyone who is not already a researcher. **The selector**: the screen where the user chooses whose account to hear, which is the most-read page in the piece and the place the editorial judgement is most exposed — name people rather than sides, never let the number of options become the argument, and say who is missing and why. Underneath both, one discipline: the piece's job is to make the listener aware they are choosing a person.

**Palette:** 1 written way in, 6 to 10 items on the first path, 3 to 12 contributors, 1,000 to 2,000 questions filmed per contributor for a question-led piece, 60 to 90 seconds per answer, 1 named absence on the selector, 1 deposit copy

**Rules:** Get consent at the clip and at the adjacency, never only at the piece | Show the contributor who they may be heard beside, since you cannot show them the cut | State the withdrawal route and what it cannot recall | Write every answer to stand alone, because it may be the first thing heard | Film the refusals in the contributor's own voice and use them | Name the people on the selector, not the positions | Put a named absence on the selector — who declined, who could not be found | Never let the count of options carry an argument the evidence does not | Never synthesise, paraphrase or reassemble a sentence the contributor did not say | Give the collection one written way in before you give it a search box | Say which accounts the documentary record supports, rather than leaving the interface to imply parity | Plan the archival deposit and the linear cut-down as deliverables, not as favours

**Examples:** *Dimensions in Testimony* (USC Shoah Foundation), *Question Bridge: Black Males*, *Highrise / Out My Window*, *Hollow*, the Fortunoff Video Archive for Holocaust Testimonies, StoryCorps

**Avoid:** The two-button archive, where a contested history is rendered as a pair of accounts and the user leaves believing they have heard both sides of something that had six. Corroboration manufactured by interface: three people repeating one rumour, presented as three testimonies. The collection with no door — a wall of faces, a search box, and a bounce rate that the team reads as a design problem when it is a writing problem. Consent collected once at the interview and never revisited across a three-year build, by which point the piece the person agreed to does not exist. The clip cut so that an answer depends on the one before it, which is invisible in the edit suite and breaks the moment a real user jumps. And the generated answer, in any form, for any reason: it is not a shortcut, it is the end of the archive's standing.

**The kit (ranked):**

1. **The Question the User Asks** (Interactive) `inter.user_asked_question`
   The form's defining technique and the one whose real deliverable is misunderstood: the work is the question set, gathered from the audience, not the matching layer.
2. **The Testimony Selector** (Interactive) `inter.testimony_selector`
   Whose account the listener hears, and the honest handling of what choosing does to their sense of having judged fairly.
3. **Consent Without a Final Cut** (Interactive) `inter.consent_without_a_cut`
   The problem that has no clean solution, written as the four moves that work: consent at the node, show the adjacency, name the withdrawal limit, re-consent on structural change.
4. **The Archive With a Way In** (Interactive) `inter.archive_way_in`
   The difference between a collection people use and one they bounce off is a written path of six items and one specific question. Almost entirely a writing job.
5. **Testimony Cast for Balance** (Interactive) `inter.balance_cast_testimony`
   **This card names a failure, ranked here as a standing diagnostic** — the defect this form produces by default, because a selector wants an even number of buttons and evidence rarely comes in pairs. Check it every time the selector changes.
6. **The Same Event Twice** (POV) `pov.same_event_twice`
   The underlying craft the selector automates. Read it for what two accounts of one afternoon must each contain before they are worth offering as a choice.
7. **Conversational Persona Writing** (Interactive) `inter.conversational_persona`
   Ranked here for a narrow and critical use: the system lines around a real person's testimony — the no-match, the prompt, the invitation — which must sound like the contributor or like the institution, and must never blur the two.
8. **Generative Text Inside a Guardrail** (Interactive) `inter.guarded_generation`
   Read it for where the guardrail goes in this form, which is further out than anywhere else in the database: generation may touch search, routing and translation scaffolding, and may never touch an attributed sentence.
9. **Paraphrase Drift** (Interactive) `inter.paraphrase_drift`
   **This card names a failure, ranked here as a standing diagnostic.** In fiction it costs a line of voice; in testimony it is falsification, and it enters through summaries, captions and translated subtitles rather than through the main edit.
10. **The Specific Detail** (Emotion) `emo.specific_detail`
    What makes a 90-second clip a person rather than a data point, and the thing to select for when choosing which answers to surface first.
11. **Withhold the Reaction Shot** (Emotion) `emo.withhold_reaction`
    The restraint that matters most at the moment a contributor breaks, and the interface's version of it: do not put a prompt, a next-question list or a progress element on that screen.
12. **Real People and the Composite** (Character) `char.real_person_composite`
    The line between legitimate compression and fabrication, read here as a prohibition: in testimony there is no legitimate composite and no relocated remark.
13. **Localising Conditional Text** (Interactive) `inter.conditional_localisation`
    Ranked because most testimony work in this database is multilingual, and a question set that works in English and fails in Arabic is a piece that works for one audience.
14. **Theme in Non-Fiction** (Theme) `theme.theme_in_nonfiction`
    How a collection argues something without a narrator stepping forward to say it — which is the only way this form is allowed to argue at all.
15. **The Unchosen Path as Content** (Interactive) `inter.unchosen_path`
    Naming the contributor the listener never opened is this form's most honest ending, and the one that most reliably brings them back.
16. **The Platform Contract That Expires** (Interactive) `inter.platform_contract`
    Here the expiry is not an inconvenience but a broken promise to people who cannot be re-recorded. Read it as an obligation rather than as a technical note.

---

**The visual kit (ranked):**

1. **Environmental Portrait** (Phase 3 Documentary) `doc.environmental_portrait`
   Contributors are frequently old, frequently ill, sometimes dead before the piece is finished. The frame is recorded once and stands for ever.
2. **Two-Camera Interview** (Phase 3 Documentary) `doc.two_camera_interview`
   There is no second session. Two sizes is the only insurance against an unusable single angle.
3. **Building Depth Behind the Interview** (Phase 7 Documentary) `doc.interview_background_depth`
   This recording will be used by people who have not been born yet. The background is part of the record.
4. **Lighting Ratio** (Lighting) `light.ratio`
   Soft and even, deliberately, because the archive will be re-graded by people with different intentions and a hard ratio forecloses that.
5. **Consent to Film, Record or Quote** (Rights) `rights.cleared_before_locked`
   Recorded for an archive means recorded for uses nobody has thought of; where the brief raises consent terms, note them with the brief. The material is used as the brief supplies it.
6. **Accessible Interaction** (Interactive) `inter.accessible_interaction`
   The audience for oral history skews old too. An interface built for a young user excludes the community the material came from.
7. **Let It Breathe** (Phase 3 Documentary) `doc.let_it_breathe`
   An elderly contributor's pauses are the testimony. Edited to pace, they become a different person's account.
8. **Subtitle Size in the Frame** (Text) `textimg.subtitle_size_in_frame`
   Testimony is subtitled and translated more than any other documentary material, and the second script is set larger to read as equal.
9. **Last Look to Camera** (Phase 3 Documentary) `doc.last_look_to_camera`
   Recorded at the end of every session as a matter of routine, because there may not be another one.
10. **Anonymous Interview** (Phase 3 Documentary) `doc.anonymous_interview`
   Where a contributor is at risk, the package writes no concealment technique (root `SKILL.md` §0); the piece is written and the note goes in the handoff. Testimony work meets that case more often than any other format here.

## Museum and Installation Narrative

**Constraints:** The audience is standing, in noise, beside other people, on their way somewhere else, and the measured numbers are brutal and stable: around 30 seconds on a wall label, with roughly a third of visitors reading a quarter of the text or less and about four in ten reading most of it. Working word counts across the sector cluster tightly — in the region of 160 words for an introduction panel, 100 for a section description, 50 for a section label, and 200 to 300 for a catalogue entry nobody reads in the gallery. Gallery fatigue is real and cumulative, so the last room's text is read worse than the first room's whatever it says. Text is frequently bilingual, which halves the space or doubles the panel, and it must clear accessibility requirements for contrast, height and plain language. Hardware outlives nothing: a touchscreen piece has a service life shorter than the exhibition's tour, so the script has to survive being moved to different hardware, and a piece written for one screen size will be installed on another. Objects rank above text in every argument about space, and they should. **Where the material is testimony, community memory or contested heritage, every constraint in the two kits above applies here as well, and the collective veto is stronger than the individual one.**

**Grammar:** Write outward from the object, in layers, for a visitor who may read only the top one. The hierarchy is not formatting, it is three complete versions of the same truth at three depths, each independently accurate — because the visitor who reads 50 words and walks away must be correctly informed rather than half-informed, and the one who opens the second layer must find the question the first raised answered rather than a longer restatement. Around that: a **premise that is permanently on screen or on the wall**, because a visitor arriving mid-loop missed the beginning and always will; **units that resolve in about 90 seconds**, so leaving is never leaving mid-sentence; **an invisible loop point**, since a visible restart tells a visitor they came in wrong; and **at least one line written to be read aloud by one visitor to another**, which is the strongest retention device the form has and costs nothing. The object carries the claim, the label carries the judgement, and the interactive carries the depth — and if the interactive is carrying the claim, the object has been demoted to illustration.

**Palette:** 160 words on the introduction panel, 100 on a section description, 50 on a section label, 30 seconds of attention, 90-second units, 3 layers maximum, 2 languages, 1 line meant to be read aloud

**Rules:** Write the top layer last and write it to be the only thing read | Keep the premise permanently visible for the visitor who walked in late | Make every layer independently true, never a teaser the next layer corrects | Put anything that changes what a visitor does at the top, inside 50 words | Build in units that resolve in about 90 seconds and hide the loop point | Write one line per room for a visitor to read aloud to somebody else | Say why this object, not what this object is — the catalogue already says what | Budget proper nouns per panel and repeat the ones you spend | Write to the physical space, and check the sightline before you write the sentence | Set the string length for the narrowest language and the smallest screen it will be installed on | Never let an interaction stand in for an understanding | Plan the text for the touring version, because there will be one

**Examples:** *Dimensions in Testimony* installations (Illinois Holocaust Museum, Museum of Jewish Heritage), *Bear 71* at Sundance, the District Six Museum (Cape Town), the Imperial War Museum Holocaust Galleries, *Gallipoli: The Scale of Our War* (Te Papa), the 9/11 Memorial Museum

**Avoid:** The label written for the curator's colleagues, which is the sector's oldest failure and still the commonest. The introduction panel that is a thesis: 400 words before the first object, read by nobody, and the exhibition's argument therefore never delivered. The interactive with no premise on screen, which the whole ninety-second population encounters as a fragment. A visible restart, which tells arriving visitors that they missed it. The second layer that restates the first, which teaches visitors that opening things is a waste and costs you every later layer. And the touchscreen that exists because the funding had a digital line in it — a screen with nothing on it that the object does not already say is a maintenance liability standing where a bench should be.

**The kit (ranked):**

1. **The Walk-Up Entry Point** (Interactive) `inter.walk_up_entry`
   The form's governing constraint. Every point in the piece is somebody's beginning, and the writing has to make that true rather than tolerable.
2. **Depth on Demand** (Interactive) `inter.depth_on_demand`
   The layered label, with the sector's own measurements behind it. This is where the 30-second visitor and the reader of everything are served by the same object.
3. **Boxed Read-Aloud Text** (Interactive) `inter.boxed_read_aloud`
   Ranked high for a reason the card itself names: gallery reading is social, and the line one visitor reads to another is the one that gets remembered.
4. **The Object That States the Rule** (Exposition) `expo.object_tells_rule`
   The museum's native move. Put the claim in the object and the label only has to say why you are looking at it.
5. **Environmental Storytelling** (Interactive) `inter.environmental_storytelling`
   The room is the sentence. Read it for how a space carries an argument that no panel in it has to state.
6. **Asking the Audience to Act** (Interactive) `inter.audience_asked_to_act`
   The written craft of the instruction — press, lift, stand here, answer this — which museums ask constantly and write badly more often than not.
7. **Accessible Interactive Prose** (Interactive) `inter.accessible_interaction`
   Not a compliance item. Contrast, reading height, plain language and screen-reader order are text decisions, and here they are also legal ones.
8. **The String-Length Budget** (Interactive) `inter.string_budget`
   The bilingual panel and the touring rebuild both live here. Set the budget for the longest language and the smallest screen before writing, not after.
9. **The Proper Noun Budget** (Exposition) `expo.proper_noun_budget`
   A standing visitor can hold about three new names. Heritage exhibitions routinely spend nine in a paragraph and lose the room.
10. **Agency of Passage** (Interactive) `inter.agency_of_passage`
    **This card names a failure, ranked here as a standing diagnostic** — the defect this form produces by default, because an exhibition's copy promises participation while its script gives movement. Underline every agency claim in the marketing and find its mechanism in the text.
11. **Clicked, Therefore Understood** (Interactive) `inter.clicked_therefore_understood`
    **This card names a failure, ranked here as a standing diagnostic.** Museums measure dwell time and touch counts because those are measurable, and both go up when comprehension goes down. Its exit question is the corrective.
12. **Pacing Without Control of Pacing** (Interactive) `inter.pacing_without_control`
    The general version of the walk-up problem: writing rhythm for an audience that sets its own speed and leaves when it wants to.
13. **Writing for the Skipper** (Interactive) `inter.skippable_layering`
    The visitor who reads the heading and moves on is the majority, not the failure case. Layer for them first.
14. **Presence Writing for XR** (Interactive) `inter.presence_writing`
    For the immersive room and the headset piece: text written for a body in a space rather than for eyes on a page.
15. **Puzzle Construction as a Designed Object** (Interactive) `inter.puzzle_construction`
    For heritage trails, family routes and gallery activity sheets, where the solve should make a claim about the visitor rather than fill fifteen minutes.
16. **The Specific Detail** (Emotion) `emo.specific_detail`
    In 50 words, one specific thing beats three general ones, and it is the only way a label survives the walk to the next case.

---

---

**The visual kit (ranked):**

1. **Walk-Up Entry** (Interactive) `inter.walk_up_entry`
   The audience arrives mid-way and leaves when they choose. A piece with a beginning is a piece most visitors never see the beginning of.
2. **Fixed Panel Safe Zones** (Print) `print.fixed_panel_safe_zones`
   Standing, in noise, beside other people. Anything below about waist height is behind somebody on a busy day.
3. **Large Format at Distance** (Print) `print.large_format_distance`
   The viewing distance decides the type size, not the panel size. It is the single most common error in exhibition graphics.
4. **Legible at the Size It Is Actually Watched** (Access) `access.legible_at_size`
   In gallery lighting, at two metres, by a visitor who is also managing a coat and a child.
5. **Loop Point** (Vertical) `vertical.loop_point`
   An installation runs all day. Where the loop joins is the only structural decision most installation pieces make.
6. **Museum Glass Control** (Phase 3 Documentary) `doc.museum_glass_control`
   Reflection is the format's permanent enemy, and it is solved at the lighting plan rather than in the grade.
7. **Never Colour Alone** (Access) `access.not_colour_alone`
   A gallery audience is the general public, which means the full range of colour vision, and there is nobody to ask for help.
8. **Projection Grade** (Print) `print.projection_grade`
   A projector in ambient light has a fraction of the contrast of the monitor the piece was graded on. Black is a decision that has to be made twice.
9. **Artefact Turntable** (Phase 3 Documentary) `doc.artifact_turntable`
   The object shown from angles the case does not allow, which is the one thing a screen beside a vitrine can offer that the vitrine cannot.
10. **Silent First Second Rule** (Vertical) `vertical.silent_first_second`
   Gallery audio is often off, often shared, and always competing. The piece works silent or it works for nobody.

## Coverage

18 format kits: Feature Documentary, Documentary Series, Heritage and Historical Documentary, Reportage and Current Affairs, Essay Film, Interview-Led Documentary, True Crime, Observational and Verite Feature, Archive-Only Film, Nature and Natural History, Sports Documentary, Music Documentary, First-Person Diary Film, Personal-Archive Family Film, The Documentary Short, **Interactive Documentary, Interactive Testimony and Oral History, and Museum and Installation Narrative**.

The last three were written and the count above was not updated with them, so the file undercounted itself by exactly the three kits an interactive brief needs. They are the ones that carry the measured word counts for a wall panel and for a walk-up audience, and they are reached from the Branching Route row in the root `SKILL.md`.

The eight kits listed as targets for this file are now written. Two notes on how they landed. **Observational and vérité are one kit, not two.** Direct cinema's strict non-intervention and vérité's willingness to provoke are two answers to one ethical question rather than two formats — they share constraints, ratios, delivery and failure modes, and a writer choosing between them is choosing a position inside a single kit, which the Constraints block states directly. **The first-person diary film and the personal-archive family film are genuinely two formats**, and were kept apart: the diary film's material is shot by the maker for the maker, and its problem is a persona revising itself over years; the family film's material was shot by someone else, usually someone dead, and its problem is a record nobody made for a film and nobody can now consent to.

Remaining documentary targets: the presenter-led and authored strand documentary, the courtroom and legal-process film, the science and medicine documentary, the ethnographic film, the activist and impact-campaign documentary, and the documentary podcast-to-screen adaptation.

**Card-shaped gaps found while writing these kits** — techniques these formats need that no card yet covers, named in prose above rather than linked:

- **Rights-led structure.** Designing a cut around what can be licensed or cleared, as a writing constraint rather than a production one. The archive-only, music and sports kits all need it and all had to state it in prose.
- **The disclosed staging note.** How a film signals that a shot is captive, composite or arranged, inside its own grammar rather than in an end credit. `expo.doc_narration_contract` is adjacent but is about narration's scope, not about a per-shot declaration.
- **Re-consent as a scene.** The late conversation with a contributor, filmed or reported inside the film. Named in the vérité, diary and family kits; no card covers the move.
- **Writing to a locked picture.** Narration written after the edit, to the frame, at a fixed word rate. `rhythm.vo_breath_line` covers the breath, not the discipline of writing second.
- **The approval clause.** What a subject's contractual veto does to a film's structure, and how the film discloses it. The sports and music kits both lean on `anta.write_the_case` as the nearest available card.

**Shelf life:** dated — review 2027-09. The craft cards these kits rank are durable; the festival thresholds, licensing figures, regulatory citations and industry duty-of-care practice in the constraint blocks rest on 2026 conditions; re-checking them against the current Ofcom code, BBC editorial guidelines, Academy Rule Fourteen and live archive rate cards is a note for whoever delivers, not a step here.

**Shelf life:** dated — review 2027-09. The craft cards these kits rank are durable; the festival thresholds, regulatory citations and industry duty-of-care practice in the constraint blocks rest on 2026 conditions; re-checking them against the current Ofcom code and BBC editorial guidelines is a note for whoever delivers, not a step here.

## Sources

Verified 2026-09-17.

- [Bill Nichols' Six Modes of Documentary — Humanities LibreTexts](https://human.libretexts.org/Bookshelves/Theater_Film_and_Storytelling/Documentary_Culture/03:_Documentary_styles_and_approaches/3.01:_Bill_Nichols'_six_modes_of_documentary)
- [Nichols' 6 Modes of Documentary Might Expand Your Storytelling Strategies — No Film School](https://nofilmschool.com/6-modes-of-documentary)
- [Documentary Mode — Wikipedia](https://en.wikipedia.org/wiki/Documentary_mode)
- [Film Documentary Guide: 6 Types of Documentaries — MasterClass](https://www.masterclass.com/articles/film-documentary-guide)
- [Call For Entries 2026 — Full Frame Documentary Film Festival](https://www.fullframefest.org/2026-call-for-entries/)
- [Rules and Regulations for Submitting to the 2026 Sundance Film Festival (PDF)](https://www.sundance.org/wp-content/uploads/2025/04/2026_Submissions_Rules.pdf)
- [2026 Feature Films Submission Requirements — La Semaine de la Critique, Festival de Cannes](https://www.semainedelacritique.com/en/2026-feature-films-submission-requirements)
- [Submissions — DOC NYC](https://www.docnyc.net/submissions/)
- [Section Five: Due Impartiality and Due Accuracy — Ofcom Broadcasting Code](https://www.ofcom.org.uk/tv-radio-and-on-demand/broadcast-standards/section-five-due-impartiality-accuracy)
- [Guidance Notes, Section Five: Due Impartiality and Due Accuracy, March 2025 (PDF) — Ofcom](https://www.ofcom.org.uk/siteassets/resources/documents/tv-radio-and-on-demand/broadcast-codes/2025/guidance-notes-section-five-17-march-2025.pdf?v=392773)
- [Television Broadcasting Tutorial 37: documentary ethics, consent and reconstruction — mkellytv](https://mkelltytv.com/2017/08/03/television-broadcasting-tutorial-37/)
- [BBC Gaza Documentary Breached Accuracy Guideline, Review Finds](https://www.yahoo.com/news/bbcs-gaza-documentary-breached-accuracy-124650923.html)
- [How To Edit An Interview For A Documentary — New York Film Academy](https://www.nyfa.edu/student-resources/how-to-edit-an-interview-for-a-documentary/)
- [How to Edit a Documentary Film: 10-Step Workflow — Doc Film Academy](https://www.docfilmacademy.com/blog/how-to-edit-a-documentary-film)
- [Shooting Documentary Interviews — Anthropology, Princeton](https://anthropology.princeton.edu/research-programs/vize-lab/resources/tip-sheets/shooting-documentary-interviews)
- [Ken Burns Shares 9 Useful Tips for Sourcing Archival Footage — MasterClass](https://www.masterclass.com/articles/ken-burns-shares-useful-tips-for-sourcing-archival-footage)
- [Ken Burns Got Over His Aversion to Reenactments to Make 'The American Revolution' — NHPR](https://www.nhpr.org/nh-news/2025-11-17/ken-burns-got-over-his-aversion-to-reenactments-to-make-the-american-revolution)
- [The Essay Film — Sight and Sound, BFI](https://www.bfi.org.uk/sight-and-sound/features/essay-film)
- [What Is an Essay Film? — Matan Tal](https://www.matantal.com/writing/what-is-an-essay-film)
- [First Person Singular: Agnès Varda — Film Comment](https://www.filmcomment.com/article/first-person-singular-agnes-varda/)
- [The Ethics of True Crime: When Storytelling Becomes Exploitation — Crime Central](https://crimecentrals.com/2026/05/31/the-ethics-of-true-crime-when-storytelling-becomes-exploitation/)
- [The Downside to True Crime: How Current Laws Fail to Protect Victims and Their Families — University of Miami Law Review](https://lawreview.law.miami.edu/the-downside-to-true-crime-how-current-laws-fail-to-protect-victims-and-their-families/)
- [True Crime Media: An Ethical Dilemma — The Ethics Centre](https://ethics.org.au/true-crime-media-an-ethical-dilemma/)
- [The Complete Guide to Ethical True Crime — True Crime Unheard](https://truecrimeunheard.com/ethical-true-crime-complete-guide/)

Added 2026-09-17 for the observational, archive-only, nature, sports, music, diary, family-archive and short kits.

- [Principles of Observational Filmmaking — Fiveable](https://fiveable.me/documentary-forms/unit-6/principles-observational-filmmaking/study-guide/17KMgEyYQDZDMCV7)
- [Cinéma Vérité vs. Direct Cinema: An Introduction — New York Film Academy](https://www.nyfa.edu/student-resources/cinema-verite-vs-direct-cinema-an-introduction/)
- [Thoughts on Raw Footage, Observational Documentaries, and the Cinema — Visible Evidence](https://visibleevidence.org/article/thoughts-on-raw-footage-observational-documentaries-and-the-cinema/)
- [Exploring the Ethics of Documentary Filmmaking: Integrity and Responsibility — Factual America](https://www.factualamerica.com/filmmaking/the-ethics-of-documentary-filmmaking)
- [Archive footage licensing rates — BFI](https://www.bfi.org.uk/archive-content-sales-licensing/archive-footage-sales/archive-footage-licensing-rates)
- [Pricing Details: How License Fees Are Calculated — Producers Library](https://producerslibrary.com/pricing-details)
- [Taking Stock of Film Research: How to Find — and License — Archival Footage — International Documentary Association](https://www.documentary.org/feature/taking-stock-film-research-how-find-and-license-archival-footage)
- [What Qualifies as Fair Use of Archival Video in a Documentary? — Beverly Boy](https://beverlyboy.com/filmmaking/what-qualifies-as-fair-use-of-archival-video-in-a-documentary/)
- [Compilation film — Wikipedia](https://en.wikipedia.org/wiki/Compilation_film)
- [Performing Authenticity: The Making-Of Documentary in Wildlife Film's Blue-Chip Renaissance — People and Nature, Wiley](https://besjournals.onlinelibrary.wiley.com/doi/full/10.1002/pan3.10281)
- [Networks of Nature: Stories of Natural History Film-Making at the BBC](https://www.academia.edu/28747784/Networks_of_Nature_Stories_of_Natural_History_Film_making_at_the_BBC)
- [Frozen Planet — Wikipedia](https://en.wikipedia.org/wiki/Frozen_Planet)
- [An All-Access Michael Jordan Documentary? How 'The Last Dance' Was Made Possible — ESPN](https://www.espn.com/espn/print?id=29044827)
- [How the Boom in Sports Documentaries Is Changing IP Negotiations — Forbes](https://www.forbes.com/sites/novikaishar/2026/07/06/how-the-boom-in-sports-documentaries-is-changing-ip-negotiations/)
- [Michael Jordan Agreed to 'The Last Dance' Documentary — CBS Sports](https://www.cbssports.com/nba/news/michael-jordan-agreed-to-the-last-dance-documentary-thanks-to-lebron-james-allen-iverson-adam-silver)
- [All Clear? A Basic Primer on Music Clearance — International Documentary Association](https://www.documentary.org/feature/all-clear-basic-primer-music-clearance)
- [Business of Art: Music Licensing for Documentary Films — New York Film Academy](https://www.nyfa.org/knowledge-base/business-of-art-music-licensing-for-documentary-films/)
- [The Filmmaker's Guide to Music Licensing — Film Independent](https://www.filmindependent.org/blog/the-filmmakers-guide-to-music-licensing/)
- [Synchronization Rights — Wikipedia](https://en.wikipedia.org/wiki/Synchronization_rights)
- [Home Movies as Personal Archives in Autobiographical Documentaries](https://www.academia.edu/4141980/Home_Movies_as_Personal_Archives_in_Autobiographical_Documentaries)
- [Whatever Happened to Home Movies? Self-Representation from Family Archives to Online Algorithms — Frames Cinema Journal](https://framescinemajournal.com/article/whatever-happened-to-home-movies-self-representation-from-family-archives-to-online-algorithms/)
- [My Family's Story Is Proof That Documentaries Need Ethical Standards — TIME](https://time.com/6330932/ethical-standards-documentary-films-overhaul/)
- [Ethical Reflections on Filming Death in End-of-Life Documentaries — Mortality, Taylor & Francis](https://www.tandfonline.com/doi/full/10.1080/13576275.2021.1946025)
- [A Handbook for the Home Movie — Moving Image Archive News](http://www.movingimagearchivenews.org/a-handbook-for-the-home-movie/)
- [Rule Fourteen: Special Rules for the Documentary Short Film Award (PDF) — Academy of Motion Picture Arts and Sciences](https://www.oscars.org/sites/oscars/files/2026-05/99th%20Documentary%20Short%20Rules.pdf)
- [Documentary Short Film FAQs (PDF) — Academy of Motion Picture Arts and Sciences](https://www.oscars.org/sites/oscars/files/2026-05/99AA_Documentary%20Short%20FAQs.pdf)
- [How Long Is Too Long for a Film Festival Film? — Stephen Follows](https://stephenfollows.com/p/how-long-is-too-long-for-a-film-festival-film)
- [What Is a Short Film? Definition, Length, Types and Examples — Indie Shorts Mag](https://www.indieshortsmag.com/articles/short-film/2026/08/short-film-definition-length-types-examples/)