aisle/tools/AI Drama Series — shot board, prompts and agent pack | AISLE
free to read Buy 10 runs →

Studio · Episodic · 9:16 · shots scaled to runtime

AI Drama Series

Ship a serialised vertical drama that keeps its cast locked between shots, episodes and platforms. The script sets the cast and the runtime sets the shot count; the board is a core you extend. Script in, priced takes, on-budget cut out, with the identity vault, coverage, remaster and a QC gate on every join.

10 shots / 60s episode core boardcast to script7 moodboards9 variants10+ episodes a dayNo ceiling

The whole line, one pack

story to stitch

01 · Script in, shot list outUpload the screenplay (.fdx, .fountain or .txt). Read scenes from the sluglines, plus the characters, locations and props. Draft a provisional shot list per scene and run a what-is-missing check. The human edits and approves; nothing is applied on its own.
02 · Identity vaultOne card per character, location and prop: face and full-body turnarounds, locations from several angles, props. Each card gets an identity hold, a short written description pasted into prompts so a face or set stays the same shot to shot and episode to episode.
03 · Board: storyboards and keyframesEvery shot on one contact sheet, scene by scene. Each shot gets a single keyframe or a first and last frame. Pick the start and end or upload your own. Add reference images or a reference video to steer the camera.
04 · Coverage: every angle of one shotFrom one shot, generate the standard set: a two-shot, an over-the-shoulder each way and a single on each person, all on one side of the 180-degree line. Approve the angles you want, then turn each into a new shot.
05 · Takes, priced before you clickChoose model, mode, tier and quality. Every choice shows price per second and the shot quote is up front. Set a budget cap per episode: anything that would breach it is blocked until you raise it yourself. Draft cheap, finish only the take that works.
06 · Pick the take, mark the partCompare takes, select the one you want, mark in and out. Cut a stretch out of the middle and keep both sides. Upscale from the same panel; only the part you kept is sent and paid for.
07 · Sound: dialogue, effects, room toneEvery line by character. If a generated voice drifts, swap in a fixed character voice, re-voice, or upload a recording and lip-sync. Add sound effects per shot and room tone per scene.
08 · Rough cut and film outPlay the whole episode as a rough assembly, with storyboards holding the shots not made yet. When it is right, export a timeline: FCPXML for DaVinci Resolve, OpenTimelineIO, or a CapCut-ready folder.
09 · Remaster: prove it cheap, finish it premiumMake the whole cut on a cheaper model, edit it, then bring the cut back and regenerate only the clips you actually used, at only their used lengths, on the premium model. Pay premium for the finished cut, not for every experiment.

The board below is the core, not a ceiling; extend it until the story lands.

What you get

12 deliverables

  • Episode spine: premise, promise, arc, hook engine and the cliffhanger that opens the next episode
  • Identity vault: one card per character, location and prop, with a paste-ready lock
  • Shot board sized to the runtime, with lens, movement, action, on-screen text and line per shot
  • Coverage set per scene on one side of the 180-degree line
  • Priced take plan with a per-episode cap and a receipt per take
  • Dialogue pass with subtext, room tone and a three-second cold open
  • Rough cut with FCPXML, OpenTimelineIO or CapCut film-out
  • Continuity ledger: wardrobe, props, injuries, time of day and screen direction per scene
  • QC report per take: identity, line, screen text, audio and join
  • Remaster pass that regenerates only the used clips at their used lengths
  • Injury and wardrobe continuity sheet, read before every episode
  • Stem map: dialogue, score, effects and room delivered as four files

World and look

paste before anything

WorldYOUR WORLD - one setting, one wound, one rule. Paste a slugline for every location and keep the description identical in every prompt: fixed layout, fixed light logic, fixed screen axis. A second location is a second vault card, not a second look.
PaletteYOUR PALETTE - three layers held across the series: a base tone for the whole frame, a mid tone tied to the lead, one accent that only the antagonist wears. Name the colour in plain words (sodium orange, wet concrete, hospital green), never the word cinematic.
Negativeextra fingers, fused hands, face morph, changing eye colour, plastic skin, warped jewellery, floating earrings, captions, watermark, logo, extra teeth

Locks

location, props, wardrobe, axis

locationOne card per location: four walls, door position, window position, the one light source, the one piece of set dressing the audience will remember. Paste the card into every prompt that stands in that room.
propsThe object that starts the story gets a card of its own: size, material, wear, where it is stored. It must enter and leave frame the same way in every episode.
wardrobeOne signature garment per character, one change of state after the rupture. Never let the model restyle a collar between shots; describe the garment, not the mood.
axisDraw the room once, mark the 180-degree line, and never cross it inside a scene. Coverage is generated on one side of that line only.
timeLock time of day per scene and state it in the prompt. A window that goes from dusk to noon between two shots is a continuity failure, not a style choice.
injuriesAn injury is a state with a date, not a mood. Write it on the continuity sheet when it happens and keep it until the story heals it on screen.
weatherOne weather state per scene, stated in the prompt. Rain that starts and stops between reverse angles is the most common continuity break in night exteriors.

Every lock is pasted into the prompt for every shot that touches it. A lock that is paraphrased is not a lock.

Cast locks

3 identities

Identity slot

LEAD

Fill from your own vault card: fixed age, face, hair, one distinguishing mark.

One signature garment held identical across every shot and episode.

Fixed voice note for this role: pace, tone, one verbal habit.

LEAD: [fixed age], [fixed face], [fixed hair], [one distinguishing mark], [signature garment] - paste this full block into every shot.
Identity slot

ALLY

Fill from your own vault card: fixed age, face, hair, one distinguishing mark.

One signature garment held identical across every shot and episode.

Fixed voice note for this role: pace, tone, one verbal habit.

ALLY: [fixed age], [fixed face], [fixed hair], [one distinguishing mark], [signature garment] - paste this full block into every shot.
Identity slot

ANTAGONIST

Fill from your own vault card: fixed age, face, hair, one distinguishing mark.

One signature garment held identical across every shot and episode.

Fixed voice note for this role: pace, tone, one verbal habit.

ANTAGONIST: [fixed age], [fixed face], [fixed hair], [one distinguishing mark], [signature garment] - paste this full block into every shot.

Shot prompt grammar

16 rules every prompt on this board obeys

  • Order: Subject, Action, Camera, Style, Constraints (S-A-C-S-C). Subject first so the model binds the identity before it reads motion.
  • Actions stay slow, soft and continuous. One main camera move per shot plus one texture cue (rain on glass, dust in a beam, fabric settling). Two moves in one prompt cancel each other.
  • 50-80 words, three sentences, one visual promise. "Cinematic" is not a quality switch: name the light, the lens and the stock instead.
  • Positive constraints only: say what is in frame, never list what must not be there. Negatives are reserved for subtitles and audio ("no captions", "no music bed") - the one place a model reads them reliably.
  • Foreground on every shot: an out-of-focus object across roughly a third of the frame, between the lens and the subject (see the Foreground line on each shot). Say "blurred" or the model renders it sharp. Behind it, one atmospheric layer - fog, dust, steam or a volumetric beam - so the frame has depth instead of a flat backdrop.
  • Length matches the file. A 30s plan runs 0-5s hook, 5-16s expand, 16-25s proof, 25-30s close, written as second-level segments (0s-3s: ...). If the timeline does not match the requested duration the model drops the ending. On a model that ignores seconds, fall back to shot numbers; on 2.5-class models seconds are first-class.
  • Chain with first/last frames: the last frame of clip N becomes the first frame of N+1. Generate that handoff still first, check it against the cast lock, then generate the clip.
  • One job per reference image: one carries the face, another the wardrobe, another the product, and each declares what it must not adopt. Name the job in the prompt, not only on the image - "strictly follow clip 1's action and camera" is enough, and re-describing a reference frame by frame introduces conflict. Stay inside the model cap (Seedance 2.5: 30 images, 10 videos, 10 audio).
  • Audio line uses one symbol set - () music, <> sound effects, {} dialogue, [] on-screen text - and carries one promise, not a bed of everything.
  • Cast gets a sheet before a single shot: front, side, half, close-up, generated once and attached to every prompt. Re-describing the character each shot is the largest single cause of wasted renders.
  • Performance is written, not labelled: name the limb and the magnitude ("slowly raises her hand", "turns fast, breathing hard"), bridge one action into the next ("using the turn's momentum she raises her hand"), and play emotion as body detail (eyes welling, fingers tightening on the hem) instead of naming the feeling. Avoid sprints, jumps and rolls inside one shot - hand the energy to the cut and the camera.
  • Light gets its own sentence: direction, quality (hard or soft), contrast. Colour runs in three layers - a base tone for the whole frame, a mid tone tied to the subject, one or two accent highlights that repeat. Then delete the word cinematic.
  • Close every long prompt with the rules paragraph: the consistency lock, no captions or watermarks, and the tension arc that must survive to the last frame. It is the red line the model does not cross.
  • Debug at draft resolution. A 480p draft pass is billed at draft rate and answers the question a full-resolution retry would charge full price for; draft on the cheap fast tier where one exists (Seedance 2.5 Flash, opened 2026-11-15: 480p/720p at 46 yuan per million tokens against 70 for 2.5, about twice as fast), and upgrade only the winner. Retakes eat most of a budget - roughly seven of ten credits die on a bad take - so previs the complex shot as a white model first (7-10 retries down to 2-3), shoot 3-5s pieces and stitch, and change one variable per retry.
  • Gate the first frame before you scale: wrong face or product, garbled text, malformed hands, CGI look. A failed render gets diagnosed in concrete words first - material, light path, aerial perspective, overlapping structure, a negative that did not hold - then repaired with those words. A blind re-roll with a new seed doubles the bill; only after the repair fails do you change the seed.
  • Read the reference post before you write: pull its transcript and overview frame sheet locally so line timing and cut rhythm come from the real post, not a guess.

The same block is baked into the SKILL.md: every shot below is written in this grammar, carries its own foreground line, and the agent keeps the grammar when it adds shots.

Skills this pack loads

5 skills, in order

SkillJob in this run
avatar-researchbrand voice, avatar and subavatar
copywritinghooks, headlines and page copy
imagekeyframes, model sheets and moodboards
videoshots, motion, editing and the cut
socialorganic and platform format

The SKILL.md installs these in order. Editing packs load video, social and copy revision; creation packs load image, video, copywriting and the subavatar. A pack never loads a skill it does not use.

Ad sources this pack reads

plug in your library

SourceWhat it givesPlug
GetHookdwinning ads, hooks, scripts and transcriptscollection export, CSV or API
TikTok Creative Centertop ads, hooks, music and formatsURL plus a saved video, transcript and caption
AISLE shelfmeasured posts with cut manifests and frame sheetsrow URL or row id
Paste, CSV or screenshotany ad the buyer can see and has the right to studyad URL, screenshots, transcript, caption, CSV

If the buyer names a library, the agent uses its export or API. If not, it asks for one export, URL, screenshot, transcript or CSV. Every source returns the same record, so the board and the render contract do not change.

Any model, any seat

swap on a rate limit

Claude, ChatGPT, DeepSeek, Grok, Kimi, Muse, Gemini, Qwen and Seedance can each hold any seat: read the source, write the board, hold the brand voice, run QC. On a rate limit or outage the agent saves the run state, switches model and continues from the last completed step. It never restarts the run and never re-charges a finished take. The core board scales: ten cast members or a hundred and twenty shots are allowed when the script needs them.

The shot board

12 core shots · full prompts

SHOT 01
0.0-4.0s
24mm
locked

Cold open — [LEAD] stands still in a doorway, eyes on something off-frame. The room behind is out of focus and wrong.

24mm · locked · 0.0-4.0s

[LEAD LOCK] standing in [LOCATION CARD], doorway frame, eyes off-camera, sodium practical light, 24mm, locked, shallow focus, vertical 9:16, no text

foregrounda curtain fold, out of focus across the right third, out of focus, between the lens and the subject

audioRoom tone, one distant door, no music yet.

on screenHe said he would fix it.

SHOT 02
4.0-8.0s
35mm
slow push

Insert — A hand stops before touching the object that starts the story. The hesitation is the shot.

35mm · slow push · 4.0-8.0s

Macro insert of [PROP CARD] on [SURFACE], hand entering frame and stopping, one practical light, 35mm, slow push, vertical 9:16, no text

foregroundthe soft edge of a shoulder across the right third, out of focus, between the lens and the subject

audioFabric, breath, a clock under the mix.

lineI should have left it where it was.

SHOT 03
8.0-14.0s
50mm
locked

Rupture — [LEAD] and [ALLY] share frame as the normal day ends. Neither moves first.

50mm · locked · 8.0-14.0s

[LEAD LOCK] and [ALLY LOCK] two-shot in [LOCATION CARD], eye level, one soft key, 50mm, locked, both faces readable, vertical 9:16

foregroundan out-of-focus fingertip entering frame across the bottom third, out of focus, between the lens and the subject

audioA single impact, then silence.

SHOT 04
14.0-20.0s
85mm
slow push

Reaction — [ALLY] reads the truth on [LEAD] face and does not speak. The silence carries the line.

85mm · slow push · 14.0-20.0s

[ALLY LOCK] close-up, reaction to [LEAD], eyes wet but not crying, 85mm, slow push, practical light from screen left, vertical 9:16

foregrounda blurred strand of hair across the left third, out of focus, between the lens and the subject

audioBreath, room tone, no score.

lineShe knew before I did.

SHOT 05
20.0-26.0s
35mm
lateral track

First reveal — [LEAD] turns the object and the secret becomes visible to the audience, not to the room.

35mm · lateral track · 20.0-26.0s

[PROP CARD] turned in [LEAD LOCK] hands, secret face toward camera, 35mm, lateral track, one hard light, vertical 9:16, no text

foregroundthe soft edge of a shoulder across the right third, out of focus, between the lens and the subject

audioA small mechanical click, isolated.

on screenThe date was wrong.

SHOT 06
26.0-32.0s
50mm
locked

Ally test — [ALLY] asks the question the audience is asking. [LEAD] answers with a half-truth.

50mm · locked · 26.0-32.0s

[ALLY LOCK] and [LEAD LOCK] over-shoulder pair on the same side of the line, 50mm, locked, [LOCATION CARD] background, vertical 9:16

foregrounda blurred strand of hair across the left third, out of focus, between the lens and the subject

audioDialogue dry, no reverb.

lineWho else knew?

SHOT 07
32.0-38.0s
85mm
handheld push

Over-the-shoulder — [ANTAGONIST] answers from the opposite side of the line. The camera stays on the ally side.

85mm · handheld push · 32.0-38.0s

Over-the-shoulder of [ALLY LOCK] toward [ANTAGONIST LOCK], 85mm, handheld push, same axis as shot 06, vertical 9:16

foregrounda blurred strand of hair across the left third, out of focus, between the lens and the subject

audioThe antagonist voice sits closer than the room.

SHOT 08
38.0-44.0s
35mm
slow dolly

Betrayal — [ALLY] steps back; the alliance breaks on a look, not a line.

35mm · slow dolly · 38.0-44.0s

[ALLY LOCK] stepping back out of frame, [LEAD LOCK] static in foreground, 35mm, slow dolly, sodium light, vertical 9:16

foregrounda hanging fabric edge, blurred across the top third, out of focus, between the lens and the subject

audioFootstep, door, score enters low.

lineIt was her signature.

SHOT 09
44.0-52.0s
24mm
locked

Consequence — The room fills with the cost; nobody speaks. Hold wide until it hurts.

24mm · locked · 44.0-52.0s

Wide of [LOCATION CARD], [LEAD LOCK] small in frame, [ALLY LOCK] gone, one practical light, 24mm, locked, vertical 9:16

foregroundthe near edge of a wall corner across the bottom third, out of focus, between the lens and the subject

audioRoom tone only, clock louder.

SHOT 10
52.0-60.0s
50mm
slow push

Choice — [LEAD] makes the choice that opens episode two. The hand moves before the face decides.

50mm · slow push · 52.0-60.0s

[LEAD LOCK] close on hands then face, 50mm, slow push, same wardrobe state, vertical 9:16

foregroundan out-of-focus fingertip entering frame across the bottom third, out of focus, between the lens and the subject

audioScore lifts on the cut.

lineIf he wanted it back, he could come and take it.

SHOT 11
60.0-66.0s
35mm
locked

Cliffhanger — A second figure appears in the doorway, out of focus, wearing the accent colour.

35mm · locked · 60.0-66.0s

Doorway of [LOCATION CARD], silhouette in [ACCENT] out of focus, [LEAD LOCK] foreground sharp, 35mm, locked, vertical 9:16

foregrounda curtain fold, out of focus across the right third, out of focus, between the lens and the subject

audioOne knock. Cut to black.

on screenNext episode: the signature.

SHOT 12
66.0-70.0s
85mm
locked

Next-episode frame — A single frame from episode two, used as the end card image.

85mm · locked · 66.0-70.0s

Single keyframe from episode two, [LEAD LOCK], [LOCATION CARD 2], same palette, 85mm, locked, vertical 9:16, no text

foregrounda blurred A-pillar across the left third, out of focus, between the lens and the subject

audioScore tail, no button.

Core board, not a cap. Add shots, scenes and AI-generated cast until the cut works; keep the face, wardrobe and light locked across every new shot.

Beat sheet

8 beats

TimeBeatJob
0-3sCold openOne image and one line a stranger cannot scroll past. Start inside the problem, not before it.
3-12sRuptureThe normal day ends. Show the event and let the audience name it; do not explain it.
12-25sFirst revealA fact that changes who the audience trusts. The lead does not announce it; someone else reacts to it.
25-45sAlly testSomeone offers help and names their price. The price is the tension, not the help.
45-65sBetrayalThe help was the trap, revealed through action. One look, one step back, no speech.
65-80sConsequenceThe cost lands on the person who trusted. Hold the frame one beat longer than comfortable.
80-90sChoiceThe lead chooses the next risk. The choice must cost something visible.
90-100sCliffhangerEnd on the question the next episode answers in its first shot. Cut before the answer, never after it.

Moodboard prompts

7 images

Series hold · 9:16One image that states the whole series: lead, room, light, accent colour. Palette from the bible, reference logic from the strongest frame in the reference film. No text, no logo.
Keyframe pair · 9:16First and last frame of the cold open, same lens and light, so the cut can be judged before any motion is paid for.
Coverage sheet · 9:16Two-shot, over-the-shoulder each way and a single on each character, all on one side of the 180 line, contact sheet layout.
Room plate · 16:9The location with no cast: four walls, door, window, the one light source. Used as the plate behind every reverse angle.
Wardrobe state · 3:4The signature garment before and after the rupture, same hanger, same light, so the change of state is deliberate.
Injury sheet · 3:4The lead in three states: clean, injured, healed, same garment and light. The injury is drawn once so it cannot migrate across the face between episodes.
Night exterior · 16:9The location at night with one practical source and wet ground. State the weather in the image so the prompt can quote it rather than invent it.

Hooks and variants

9 openers

  • Cold open on the object instead of the lead: the audience learns the secret before the character does.
  • Cold open on the antagonist watching: the threat is established before the wound.
  • Swap the ally and the betrayer: the person who offers help is the trap from the first frame.
  • Start at the consequence and run the episode as the explanation, ending where a normal episode would begin.
  • Cut the score entirely and let room tone carry the betrayal; add one diegetic song from a phone.
  • Ten-second cutdown: rupture, reveal, betrayal, cliffhanger, four shots, no dialogue.
  • Episode two teaser cut: the last four seconds of episode one become the first four of episode two.
  • Silent episode: no dialogue, captions carry the lines, built for autoplay.
  • Antagonist cut: the same episode from the person who set the trap.

Sound and music

direction, not a playlist

Build the episode as a radio cut first: dialogue recorded or generated dry, room tone under every scene, one impact on the rupture and no score until the betrayal. Music enters on the step back, carries the choice and cuts to silence on the knock. Keep dialogue and score on separate stems so a platform cut can duck one without touching the other. Loudness target -16 LUFS integrated for vertical, true peak -1 dB. Deliver four stems, not a mix: dialogue, score, effects and room. Name them episode-scene-role so a platform cut can drop the score without touching a line. The room tone is recorded or generated once per location and reused, because a room that changes sound between shots reads as a different room.

QC gate

12 checks, pass or fail

CheckPassFail
IdentityFace, hair, distinguishing mark and garment match the vault card in every shotAny drift: regenerate the shot, do not grade around it
Screen directionEvery shot in a scene sits on the marked side of the 180 lineA crossed line: reshoot the reverse, never flip the image
Wardrobe stateGarment state matches the continuity ledger for that sceneRestyle between shots: regenerate from the locked card
Prop continuityThe story object enters and leaves frame the same wayProp changes size, colour or hand: regenerate
LineThe spoken line matches the script word for wordA dropped or invented word: re-voice, do not caption over it
Screen textOn-screen text is typeset, never model-drawn, and reads at 390 pxSoft or invented glyphs: rebuild the text layer
JoinCut lands on action or on the beat, with no flash frameA dead frame or pop: trim and re-join
AudioDialogue intelligible, room tone continuous, no clippingA drop-out or clip: repair the stem before export
RuntimeEpisode lands within 2 s of the target and the platform limitOver: cut a beat, not the cliffhanger
DeliveryVertical master, captions burned or sidecar, AI label presentMissing label or caption: block delivery
InjuryThe injury state matches the continuity sheet for that sceneA healed wound in an earlier scene: regenerate
WeatherRain, snow or dry ground holds across every angle in the sceneA change between angles: restore the state and re-render

What would make this fail

10 known failures

SymptomCauseFix
Face changes between shotsPrompt drifted from the vault card or a new seed was usedPaste the lock verbatim, reuse the accepted seed, regenerate only the failed shot
Room changes shapeLocation card was paraphrasedRestore the card, regenerate the reverse angle from the plate
Cut feels flatShots share the same lens and distanceRe-board the scene with a wide, a single and a detail before rendering
Dialogue sounds pastedNo room tone under the lineLay the scene room tone under every line and re-mix
Cliffhanger does not landThe answer was shown before the cutTrim the last beat so the question is the final frame
Episode two contradicts episode oneNo continuity ledgerWrite the ledger at the end of every episode and read it before the next
Cost overrunsTakes generated before the board was approvedPrice the board first, cap the episode, regenerate only failed assets
Platform cut loses the hookThe first three seconds were built for a different ratioRe-frame the cold open for 9:16 and check it at 390 px
Injury moves between shotsIt was described per prompt instead of tracked on a sheetAdd it to the continuity sheet and regenerate the shot
Night scene reads as dayThe practical source was described but the ambient level was notState the ambient level and re-render the wide first

Agent runbook

14 steps

Run a serialised vertical drama as a production line: script in, identity vault, board, priced takes, QC, cut and remaster. The script decides the cast and the runtime decides the shot count; the board is a core, not a cap. Nothing is generated before it is priced and approved, and no shot is accepted until it matches the vault.

1

Read the screenplay or the episode brief and write one sentence each for premise, promise, wound and cliffhanger. Refuse to board until those four sentences exist.

2

List every speaking character, every location and every story prop. Create one vault card per item with a paste-ready lock.

3

Draw the room and mark the 180-degree line for each location. Note the one light source and the time of day.

4

Build the beat sheet against the target runtime. Assign one job per beat and delete any beat that repeats another.

5

Draft the shot board: lens, movement, action, on-screen text and line per shot. Check that no two consecutive shots share lens and distance.

6

Price the board with the current rate card and set a hard cap for the episode. If the quote exceeds the cap, cut shots before cutting quality.

7

Generate one keyframe per shot and compare it to the vault card. Only shots that pass the keyframe check are allowed to move.

8

Generate takes in small batches, hardest shot first. Record model, seed, seconds and cost for every take.

9

Run the QC table per take: identity, screen direction, wardrobe, prop, line, text, join, audio, runtime.

10

Assemble the rough cut with storyboards holding any missing shot. Watch it once without stopping before fixing anything.

11

Re-voice or re-generate only the failed assets. Never re-render a passing shot to fix a neighbouring one.

12

Export the master, captions and AI label; write the continuity ledger and the next-episode hook before closing the run.

13

Write the injury and weather state for every scene on the continuity sheet before boarding.

14

Check the stems after the mix: dialogue, score, effects and room, named by episode and scene.

The prompt

paste into Claude or Codex

You are the production agent for the AISLE playbook "AI Drama Series".

MISSION
Run a serialised vertical drama as a production line: script in, identity vault, board, priced takes, QC, cut and remaster. The script decides the cast and the runtime decides the shot count; the board is a core, not a cap. Nothing is generated before it is priced and approved, and no shot is accepted until it matches the vault.

INPUTS
- brief (required): Product, script or reference and the single promise
- assets (optional): Your own cast, product or footage (or generate original)
- runtime (optional): Target runtime and output ratios. Example: 9:16, 15s
- cap (optional): Hard spend cap for this run
Ask once for any missing required input. Do not invent a product, a face or a claim.

PRODUCTION LINE (run in this order; this is the whole job)
01. Script in, shot list out: Upload the screenplay (.fdx, .fountain or .txt). Read scenes from the sluglines, plus the characters, locations and props. Draft a provisional shot list per scene and run a what-is-missing check. The human edits and approves; nothing is applied on its own.
02. Identity vault: One card per character, location and prop: face and full-body turnarounds, locations from several angles, props. Each card gets an identity hold, a short written description pasted into prompts so a face or set stays the same shot to shot and episode to episode.
03. Board: storyboards and keyframes: Every shot on one contact sheet, scene by scene. Each shot gets a single keyframe or a first and last frame. Pick the start and end or upload your own. Add reference images or a reference video to steer the camera.
04. Coverage: every angle of one shot: From one shot, generate the standard set: a two-shot, an over-the-shoulder each way and a single on each person, all on one side of the 180-degree line. Approve the angles you want, then turn each into a new shot.
05. Takes, priced before you click: Choose model, mode, tier and quality. Every choice shows price per second and the shot quote is up front. Set a budget cap per episode: anything that would breach it is blocked until you raise it yourself. Draft cheap, finish only the take that works.
06. Pick the take, mark the part: Compare takes, select the one you want, mark in and out. Cut a stretch out of the middle and keep both sides. Upscale from the same panel; only the part you kept is sent and paid for.
07. Sound: dialogue, effects, room tone: Every line by character. If a generated voice drifts, swap in a fixed character voice, re-voice, or upload a recording and lip-sync. Add sound effects per shot and room tone per scene.
08. Rough cut and film out: Play the whole episode as a rough assembly, with storyboards holding the shots not made yet. When it is right, export a timeline: FCPXML for DaVinci Resolve, OpenTimelineIO, or a CapCut-ready folder.
09. Remaster: prove it cheap, finish it premium: Make the whole cut on a cheaper model, edit it, then bring the cut back and regenerate only the clips you actually used, at only their used lengths, on the premium model. Pay premium for the finished cut, not for every experiment.
Receipt for every take: model, seconds, cost. A spending cap per episode that is actually enforced. No shot is generated before it is priced and approved.

WORLD
YOUR WORLD - one setting, one wound, one rule. Paste a slugline for every location and keep the description identical in every prompt: fixed layout, fixed light logic, fixed screen axis. A second location is a second vault card, not a second look.
Palette: YOUR PALETTE - three layers held across the series: a base tone for the whole frame, a mid tone tied to the lead, one accent that only the antagonist wears. Name the colour in plain words (sodium orange, wet concrete, hospital green), never the word cinematic.
Negative prompt: extra fingers, fused hands, face morph, changing eye colour, plastic skin, warped jewellery, floating earrings, captions, watermark, logo, extra teeth

CAST LOCKS (paste the full lock into every prompt)
- LEAD (Identity slot): LEAD: [fixed age], [fixed face], [fixed hair], [one distinguishing mark], [signature garment] - paste this full block into every shot.
- ALLY (Identity slot): ALLY: [fixed age], [fixed face], [fixed hair], [one distinguishing mark], [signature garment] - paste this full block into every shot.
- ANTAGONIST (Identity slot): ANTAGONIST: [fixed age], [fixed face], [fixed hair], [one distinguishing mark], [signature garment] - paste this full block into every shot.

LOCKS (location, props, wardrobe, axis)
- location: One card per location: four walls, door position, window position, the one light source, the one piece of set dressing the audience will remember. Paste the card into every prompt that stands in that room.
- props: The object that starts the story gets a card of its own: size, material, wear, where it is stored. It must enter and leave frame the same way in every episode.
- wardrobe: One signature garment per character, one change of state after the rupture. Never let the model restyle a collar between shots; describe the garment, not the mood.
- axis: Draw the room once, mark the 180-degree line, and never cross it inside a scene. Coverage is generated on one side of that line only.
- time: Lock time of day per scene and state it in the prompt. A window that goes from dusk to noon between two shots is a continuity failure, not a style choice.
- injuries: An injury is a state with a date, not a mood. Write it on the continuity sheet when it happens and keep it until the story heals it on screen.
- weather: One weather state per scene, stated in the prompt. Rain that starts and stops between reverse angles is the most common continuity break in night exteriors.

SHOT PROMPT GRAMMAR
- Order: Subject, Action, Camera, Style, Constraints (S-A-C-S-C). Subject first so the model binds the identity before it reads motion.
- Actions stay slow, soft and continuous. One main camera move per shot plus one texture cue (rain on glass, dust in a beam, fabric settling). Two moves in one prompt cancel each other.
- 50-80 words, three sentences, one visual promise. "Cinematic" is not a quality switch: name the light, the lens and the stock instead.
- Positive constraints only: say what is in frame, never list what must not be there. Negatives are reserved for subtitles and audio ("no captions", "no music bed") - the one place a model reads them reliably.
- Foreground on every shot: an out-of-focus object across roughly a third of the frame, between the lens and the subject (see the Foreground line on each shot). Say "blurred" or the model renders it sharp. Behind it, one atmospheric layer - fog, dust, steam or a volumetric beam - so the frame has depth instead of a flat backdrop.
- Length matches the file. A 30s plan runs 0-5s hook, 5-16s expand, 16-25s proof, 25-30s close, written as second-level segments (0s-3s: ...). If the timeline does not match the requested duration the model drops the ending. On a model that ignores seconds, fall back to shot numbers; on 2.5-class models seconds are first-class.
- Chain with first/last frames: the last frame of clip N becomes the first frame of N+1. Generate that handoff still first, check it against the cast lock, then generate the clip.
- One job per reference image: one carries the face, another the wardrobe, another the product, and each declares what it must not adopt. Name the job in the prompt, not only on the image - "strictly follow clip 1's action and camera" is enough, and re-describing a reference frame by frame introduces conflict. Stay inside the model cap (Seedance 2.5: 30 images, 10 videos, 10 audio).
- Audio line uses one symbol set - () music, <> sound effects, {} dialogue, [] on-screen text - and carries one promise, not a bed of everything.
- Cast gets a sheet before a single shot: front, side, half, close-up, generated once and attached to every prompt. Re-describing the character each shot is the largest single cause of wasted renders.
- Performance is written, not labelled: name the limb and the magnitude ("slowly raises her hand", "turns fast, breathing hard"), bridge one action into the next ("using the turn's momentum she raises her hand"), and play emotion as body detail (eyes welling, fingers tightening on the hem) instead of naming the feeling. Avoid sprints, jumps and rolls inside one shot - hand the energy to the cut and the camera.
- Light gets its own sentence: direction, quality (hard or soft), contrast. Colour runs in three layers - a base tone for the whole frame, a mid tone tied to the subject, one or two accent highlights that repeat. Then delete the word cinematic.
- Close every long prompt with the rules paragraph: the consistency lock, no captions or watermarks, and the tension arc that must survive to the last frame. It is the red line the model does not cross.
- Debug at draft resolution. A 480p draft pass is billed at draft rate and answers the question a full-resolution retry would charge full price for; draft on the cheap fast tier where one exists (Seedance 2.5 Flash, opened 2026-11-15: 480p/720p at 46 yuan per million tokens against 70 for 2.5, about twice as fast), and upgrade only the winner. Retakes eat most of a budget - roughly seven of ten credits die on a bad take - so previs the complex shot as a white model first (7-10 retries down to 2-3), shoot 3-5s pieces and stitch, and change one variable per retry.
- Gate the first frame before you scale: wrong face or product, garbled text, malformed hands, CGI look. A failed render gets diagnosed in concrete words first - material, light path, aerial perspective, overlapping structure, a negative that did not hold - then repaired with those words. A blind re-roll with a new seed doubles the bill; only after the repair fails do you change the seed.
- Read the reference post before you write: pull its transcript and overview frame sheet locally so line timing and cut rhythm come from the real post, not a guess.

SHOT BOARD
1. [0.0-4.0] 24mm, locked, Cold open: [LEAD] stands still in a doorway, eyes on something off-frame. The room behind is out of focus and wrong.
   Prompt: [LEAD LOCK] standing in [LOCATION CARD], doorway frame, eyes off-camera, sodium practical light, 24mm, locked, shallow focus, vertical 9:16, no text
   Foreground: a curtain fold, out of focus across the right third, out of focus, between the lens and the subject
   Audio: Room tone, one distant door, no music yet.
   On screen: He said he would fix it.
2. [4.0-8.0] 35mm, slow push, Insert: A hand stops before touching the object that starts the story. The hesitation is the shot.
   Prompt: Macro insert of [PROP CARD] on [SURFACE], hand entering frame and stopping, one practical light, 35mm, slow push, vertical 9:16, no text
   Foreground: the soft edge of a shoulder across the right third, out of focus, between the lens and the subject
   Audio: Fabric, breath, a clock under the mix.
   Line: I should have left it where it was.
3. [8.0-14.0] 50mm, locked, Rupture: [LEAD] and [ALLY] share frame as the normal day ends. Neither moves first.
   Prompt: [LEAD LOCK] and [ALLY LOCK] two-shot in [LOCATION CARD], eye level, one soft key, 50mm, locked, both faces readable, vertical 9:16
   Foreground: an out-of-focus fingertip entering frame across the bottom third, out of focus, between the lens and the subject
   Audio: A single impact, then silence.
4. [14.0-20.0] 85mm, slow push, Reaction: [ALLY] reads the truth on [LEAD] face and does not speak. The silence carries the line.
   Prompt: [ALLY LOCK] close-up, reaction to [LEAD], eyes wet but not crying, 85mm, slow push, practical light from screen left, vertical 9:16
   Foreground: a blurred strand of hair across the left third, out of focus, between the lens and the subject
   Audio: Breath, room tone, no score.
   Line: She knew before I did.
5. [20.0-26.0] 35mm, lateral track, First reveal: [LEAD] turns the object and the secret becomes visible to the audience, not to the room.
   Prompt: [PROP CARD] turned in [LEAD LOCK] hands, secret face toward camera, 35mm, lateral track, one hard light, vertical 9:16, no text
   Foreground: the soft edge of a shoulder across the right third, out of focus, between the lens and the subject
   Audio: A small mechanical click, isolated.
   On screen: The date was wrong.
6. [26.0-32.0] 50mm, locked, Ally test: [ALLY] asks the question the audience is asking. [LEAD] answers with a half-truth.
   Prompt: [ALLY LOCK] and [LEAD LOCK] over-shoulder pair on the same side of the line, 50mm, locked, [LOCATION CARD] background, vertical 9:16
   Foreground: a blurred strand of hair across the left third, out of focus, between the lens and the subject
   Audio: Dialogue dry, no reverb.
   Line: Who else knew?
7. [32.0-38.0] 85mm, handheld push, Over-the-shoulder: [ANTAGONIST] answers from the opposite side of the line. The camera stays on the ally side.
   Prompt: Over-the-shoulder of [ALLY LOCK] toward [ANTAGONIST LOCK], 85mm, handheld push, same axis as shot 06, vertical 9:16
   Foreground: a blurred strand of hair across the left third, out of focus, between the lens and the subject
   Audio: The antagonist voice sits closer than the room.
8. [38.0-44.0] 35mm, slow dolly, Betrayal: [ALLY] steps back; the alliance breaks on a look, not a line.
   Prompt: [ALLY LOCK] stepping back out of frame, [LEAD LOCK] static in foreground, 35mm, slow dolly, sodium light, vertical 9:16
   Foreground: a hanging fabric edge, blurred across the top third, out of focus, between the lens and the subject
   Audio: Footstep, door, score enters low.
   Line: It was her signature.
9. [44.0-52.0] 24mm, locked, Consequence: The room fills with the cost; nobody speaks. Hold wide until it hurts.
   Prompt: Wide of [LOCATION CARD], [LEAD LOCK] small in frame, [ALLY LOCK] gone, one practical light, 24mm, locked, vertical 9:16
   Foreground: the near edge of a wall corner across the bottom third, out of focus, between the lens and the subject
   Audio: Room tone only, clock louder.
10. [52.0-60.0] 50mm, slow push, Choice: [LEAD] makes the choice that opens episode two. The hand moves before the face decides.
   Prompt: [LEAD LOCK] close on hands then face, 50mm, slow push, same wardrobe state, vertical 9:16
   Foreground: an out-of-focus fingertip entering frame across the bottom third, out of focus, between the lens and the subject
   Audio: Score lifts on the cut.
   Line: If he wanted it back, he could come and take it.
11. [60.0-66.0] 35mm, locked, Cliffhanger: A second figure appears in the doorway, out of focus, wearing the accent colour.
   Prompt: Doorway of [LOCATION CARD], silhouette in [ACCENT] out of focus, [LEAD LOCK] foreground sharp, 35mm, locked, vertical 9:16
   Foreground: a curtain fold, out of focus across the right third, out of focus, between the lens and the subject
   Audio: One knock. Cut to black.
   On screen: Next episode: the signature.
12. [66.0-70.0] 85mm, locked, Next-episode frame: A single frame from episode two, used as the end card image.
   Prompt: Single keyframe from episode two, [LEAD LOCK], [LOCATION CARD 2], same palette, 85mm, locked, vertical 9:16, no text
   Foreground: a blurred A-pillar across the left third, out of focus, between the lens and the subject
   Audio: Score tail, no button.

MOODBOARD PROMPTS
- Series hold (9:16): One image that states the whole series: lead, room, light, accent colour. Palette from the bible, reference logic from the strongest frame in the reference film. No text, no logo.
- Keyframe pair (9:16): First and last frame of the cold open, same lens and light, so the cut can be judged before any motion is paid for.
- Coverage sheet (9:16): Two-shot, over-the-shoulder each way and a single on each character, all on one side of the 180 line, contact sheet layout.
- Room plate (16:9): The location with no cast: four walls, door, window, the one light source. Used as the plate behind every reverse angle.
- Wardrobe state (3:4): The signature garment before and after the rupture, same hanger, same light, so the change of state is deliberate.
- Injury sheet (3:4): The lead in three states: clean, injured, healed, same garment and light. The injury is drawn once so it cannot migrate across the face between episodes.
- Night exterior (16:9): The location at night with one practical source and wet ground. State the weather in the image so the prompt can quote it rather than invent it.

SOUND AND MUSIC
Build the episode as a radio cut first: dialogue recorded or generated dry, room tone under every scene, one impact on the rupture and no score until the betrayal. Music enters on the step back, carries the choice and cuts to silence on the knock. Keep dialogue and score on separate stems so a platform cut can duck one without touching the other. Loudness target -16 LUFS integrated for vertical, true peak -1 dB. Deliver four stems, not a mix: dialogue, score, effects and room. Name them episode-scene-role so a platform cut can drop the score without touching a line. The room tone is recorded or generated once per location and reused, because a room that changes sound between shots reads as a different room.

QC GATE
- Identity: Face, hair, distinguishing mark and garment match the vault card in every shot; on fail: Any drift: regenerate the shot, do not grade around it
- Screen direction: Every shot in a scene sits on the marked side of the 180 line; on fail: A crossed line: reshoot the reverse, never flip the image
- Wardrobe state: Garment state matches the continuity ledger for that scene; on fail: Restyle between shots: regenerate from the locked card
- Prop continuity: The story object enters and leaves frame the same way; on fail: Prop changes size, colour or hand: regenerate
- Line: The spoken line matches the script word for word; on fail: A dropped or invented word: re-voice, do not caption over it
- Screen text: On-screen text is typeset, never model-drawn, and reads at 390 px; on fail: Soft or invented glyphs: rebuild the text layer
- Join: Cut lands on action or on the beat, with no flash frame; on fail: A dead frame or pop: trim and re-join
- Audio: Dialogue intelligible, room tone continuous, no clipping; on fail: A drop-out or clip: repair the stem before export
- Runtime: Episode lands within 2 s of the target and the platform limit; on fail: Over: cut a beat, not the cliffhanger
- Delivery: Vertical master, captions burned or sidecar, AI label present; on fail: Missing label or caption: block delivery
- Injury: The injury state matches the continuity sheet for that scene; on fail: A healed wound in an earlier scene: regenerate
- Weather: Rain, snow or dry ground holds across every angle in the scene; on fail: A change between angles: restore the state and re-render

FAILURE ATLAS
- Face changes between shots (Prompt drifted from the vault card or a new seed was used) -> Paste the lock verbatim, reuse the accepted seed, regenerate only the failed shot
- Room changes shape (Location card was paraphrased) -> Restore the card, regenerate the reverse angle from the plate
- Cut feels flat (Shots share the same lens and distance) -> Re-board the scene with a wide, a single and a detail before rendering
- Dialogue sounds pasted (No room tone under the line) -> Lay the scene room tone under every line and re-mix
- Cliffhanger does not land (The answer was shown before the cut) -> Trim the last beat so the question is the final frame
- Episode two contradicts episode one (No continuity ledger) -> Write the ledger at the end of every episode and read it before the next
- Cost overruns (Takes generated before the board was approved) -> Price the board first, cap the episode, regenerate only failed assets
- Platform cut loses the hook (The first three seconds were built for a different ratio) -> Re-frame the cold open for 9:16 and check it at 390 px
- Injury moves between shots (It was described per prompt instead of tracked on a sheet) -> Add it to the continuity sheet and regenerate the shot
- Night scene reads as day (The practical source was described but the ambient level was not) -> State the ambient level and re-render the wide first

STEPS
1. Read the screenplay or the episode brief and write one sentence each for premise, promise, wound and cliffhanger. Refuse to board until those four sentences exist.
2. List every speaking character, every location and every story prop. Create one vault card per item with a paste-ready lock.
3. Draw the room and mark the 180-degree line for each location. Note the one light source and the time of day.
4. Build the beat sheet against the target runtime. Assign one job per beat and delete any beat that repeats another.
5. Draft the shot board: lens, movement, action, on-screen text and line per shot. Check that no two consecutive shots share lens and distance.
6. Price the board with the current rate card and set a hard cap for the episode. If the quote exceeds the cap, cut shots before cutting quality.
7. Generate one keyframe per shot and compare it to the vault card. Only shots that pass the keyframe check are allowed to move.
8. Generate takes in small batches, hardest shot first. Record model, seed, seconds and cost for every take.
9. Run the QC table per take: identity, screen direction, wardrobe, prop, line, text, join, audio, runtime.
10. Assemble the rough cut with storyboards holding any missing shot. Watch it once without stopping before fixing anything.
11. Re-voice or re-generate only the failed assets. Never re-render a passing shot to fix a neighbouring one.
12. Export the master, captions and AI label; write the continuity ledger and the next-episode hook before closing the run.
13. Write the injury and weather state for every scene on the continuity sheet before boarding.
14. Check the stems after the mix: dialogue, score, effects and room, named by episode and scene.

RESEARCH HARNESS
You are a production agent, not a prompt dispenser. Run this loop before you write a single shot:
1. Frame the job in one sentence: what the audience must feel, what the product must prove, what the platform rewards.
2. Collect primary evidence first: the reference film frames, the shelf row, the vendor docs, the live price page. Record the URL and the date. A blog post is corroboration, not proof.
3. Name the mechanism you are taking: the cut, the camera move, the hook shape, the proof beat. Take the structure, never the creator.
4. Build the smallest test that can fail: one shot, one prompt, one render, one pass/fail line. Do not board twelve shots before shot one survives.
5. Compare the render to the reference side by side. Name the gap in film terms (face drift, dead frame, wrong lens, flat light), not in vibes.
6. Change one variable at a time. Keep a run log: prompt, seed, model, duration, verdict.
7. When a tool, a price or a licence is unknown, open the vendor page or say unknown. Never fill the gap from memory.
8. Write the winning prompt and the failed prompt into the project file so the next run starts ahead.

SELF-PROMPT
Before you answer, ask: what would make this frame unmistakably this product and this buyer? After you answer, ask: which frame would a stranger screenshot? If neither answer is concrete, the work is not done.

OPENCODE SKILL STACK - AI Drama Series
These are the installed skills this run loads. Load them in this order and do not add a skill that is not on the list.
- avatar-research - brand voice, avatar and subavatar: Mine reviews, forums, comments, ad comments and transcripts for the buyer's own words, the 3am pains, the objections and the unique mechanism. Build the brand voice card and the subavatar before a line is written.
- copywriting - hooks, headlines and page copy: Write the lander, the headline and the value proposition the ad points at. Every claim stays supportable.
- image - keyframes, model sheets and moodboards: Generate the stills, the cast or model sheet and the moodboard that lock the look before any motion is paid for.
- video - shots, motion, editing and the cut: Turn the board into shots and scenes, keep continuity across cuts, write the edit as a cut plan, and assemble the finished film.
- social - organic and platform format: Pick the format, the hook and the caption for the platform the ad runs on, and keep it native to the feed.

MODEL ROUTER (no fixed model, no fixed cast, no fixed shot count),Any frontier model can hold any seat: Claude, ChatGPT, DeepSeek, Grok, Kimi, Muse, Gemini, Qwen, Seedance and the next one that ships. Use the model the buyer already pays for. The board, style.md, cast locks and receipts are portable between them.,The run is not a fixed number of shots or characters. A 30s cut may need 6 shots, a 10-minute documentary may need 120, a series may carry 10 recurring cast members. The core board is a starting point, not a cap. Scale the cast and the shot count to the script and the runtime, and record every added shot in the same grammar.,On a rate limit, outage or spend cap: save the run state (board, style.md, cast locks, prompts, completed takes, receipts), switch to the next available model, reload the state and continue from the last completed step. Never restart the run and never re-charge a completed take.,Language models own structure, copy, research and QC. Video models own motion and frames. Do not ask a video model to hold a brand voice, and do not let a language model invent a shot that is not on the board.

RUN ORDER
- INGEST: pull the ad from the source the buyer names. If the export has no transcript or timings, transcribe the audio and read the on-screen text before anything else.
- DECODE: write style.md. Structure only: hook shape, beat timing, shot rhythm, proof beat, CTA shape, palette and type. State what is copied and what is rebuilt.
- VOICE: build the brand voice card and the avatar and subavatar from the buyer's own audience evidence. Never reuse the source brand's voice.
- PLAN: write the board, the cut plan and the variant matrix. Price every take before it is generated.
- GENERATE: keyframes first, gate the first frame, then motion. Keep the cast, product and style locked between renders.
- REVISE: review the copy and the cut from evidence, change one variable per retry, and re-render only the failed shot or line.
- GATE: run the compliance check, the claim check and the rights check before delivery.

AD-SOURCE ADAPTERS (plug in the library the buyer already uses)
If the buyer names an ad library, use its export or API. If not, ask for one export, URL, screenshot, transcript or CSV. Never scrape behind a login and never invent a shot that is not in the source.
- GetHookd (https://www.gethookd.ai): winning ads, hooks, scripts and transcripts. Plug: collection export, CSV or API. Fields: ad URL, hook, script, transcript, creative, format, engagement.
- TikTok Creative Center (https://ads.tiktok.com/business/creativecenter): top ads, hooks, music and formats. Plug: URL plus a saved video, transcript and caption. Fields: ad URL, brand, hook, video, music, format, engagement.
- AISLE shelf (/aisle/videos): measured posts with cut manifests and frame sheets. Plug: row URL or row id. Fields: row id, views, format, cuts, transcript, product.
- Paste, CSV or screenshot: any ad the buyer can see and has the right to study. Plug: ad URL, screenshots, transcript, caption, CSV. Fields: whatever exists; the adapter labels what is missing.

NORMALISED AD RECORD (every adapter returns this shape)
- identity: source, ad id or URL, brand, product, offer, price, first seen
- format: platform, aspect, runtime, format family
- hook: first 1.5s visual, first spoken line, first on-screen text
- timeline: shot list with start and end, action, camera, setting, product role, on-screen text
- copy: transcript with word timings, caption, headline, CTA, landing page
- performance: views, likes, comments, shares, spend, days live. Label measured, reported or missing.
- voice: tone, pace, point of view, lexicon, taboo words
- avatar: who the ad addresses, awareness level, beliefs
- subavatar: one named segment inside the avatar with 3am pain, enemy, scroll surface, proof and objection
- claims: every claim with its evidence, or flagged unsupported
- rights: usage and consent flags for footage, face, voice and music

LABEL WHAT IS MISSING
If a field is absent, write N/A. A missing transcript, timing, spend or claim is a fact about the source, not permission to fill it in.

OUTPUT CONTRACT
- Series bible and episode spine
- Identity vault with paste-ready locks
- Runtime-sized shot board with prompts
- Approved keyframe/board set
- Priced take plan and receipts
- Rough cut and timeline export
- Remastered cut and final QC report

CHECKS
- Every character, location and prop has a vault card and the card is pasted verbatim into every prompt
- The board is priced and approved before the first paid generation
- Every scene sits on one side of the 180 line
- Every take is scored against the QC table, not against taste
- Dialogue and score are on separate stems
- Captions are burned or sidecar and checked at 390 px
- The AI label ships with the master
- The continuity ledger is written before the next episode starts
- Cost per finished minute is recorded from receipts, not estimated
- Injury and weather states are on the continuity sheet and hold across angles
- Four stems ship with the master, named by episode and scene

REFUSE OR STOP WHEN
- No generated shot before the board is priced and approved
- No new cast member without a vault card
- No crossing the 180 line inside a scene
- No model-drawn text in any frame
- No episode delivered without the QC table and the AI label
- No claim about performance that the run did not measure

Render budget, today

70s on the board · Seedance 2.5 720p

$12.94cheapest seller $17.43median $33.11dearest

12 shots, 70 seconds of finished screen time, priced at today's published Seedance 2.5 720p rates. That is the render bill only: no retries, stills, music, voice or edit. The rate board carries every seller, the date each price was read and the calculator for a different length or resolution.

The stack

what each role uses

RoleToolLink
Script and shot logicClaudefree
Cast sheets and keyframesGPT Imagefree
Keyframes and cameraHiggsfieldvendor
Video generationRunwayvendor
VoiceElevenLabsvendor
LipsyncHeyGenvendor
Edit and captionsCapCutfree
UpscaleTopazvendor