Back to Token Sink
Real-time film · 19:17 · production notes · 8 October 2026

The Crown of Scotland

The Crown of Scotland, 19:17. Watch on YouTube.

A nineteen-minute toy-theatre history of Scotland’s crown, told as the question of who has the right to wear it, with Mary, Queen of Scots as the longest act. An Opus director, nine Sonnet animators and ten Haiku researchers made it in an afternoon and evening. This is where the time and the money went.

The session in four numbers

From a request to a published film

5 h 29 minfrom the first message at 16:49 to the final report at 22:19, with about 42 minutes of gaps
27 h 28 minof agent working time, run in parallel
20 agentsone Opus director, 9 Sonnet animators and 10 Haiku researchers and checkers
$206.21if every call had been billed at API list prices
19:17 film1376×768 at 3061 shots131 narrated lines50 plates141 sprites27 ambience and effects1 code-composed score
The afternoon and evening

16:49 to 22:19

Each bar is one agent, from its first instruction to its last report; each thin mark is a tool call. The row below marks every picture, voice line and sound the machine produced. Research ran first, then the animators in three waves, one act each, while the director built the score and mix. The gaps in the director’s bar are times with no activity.

16:0017:0018:0019:0020:0021:0022:0023:00Directorresearch: origins and Sconeresearch: Macbeth, Margaret, David Iresearch: Alexander III and successionresearch: Bruce and independenceresearch: Stewart kings to James Vresearch: Mary, Queen of Scotsresearch: James VI and the Honoursresearch: toy theatre and Stevensonresearch: traditional tunes in ABCfact check: five claimsAct I: The StoneAct II: Gold into LeadAct III part AAct III part BAct IV part AAct IV part BAct V part AAct V part BAct VILocal: plates (50)Local: sprites (141)Local: voice lines (131)Local: ambience and effects (27)context summarised 18:48MP4 written 22:18
Opus 5.5Sonnet 5.5Haiku 5.5Local machine, no API cost
The company

Who did what, for how long, at what price

Agent and assignmentRan (local)TimeAPI callsTool callsCost at list prices
Director · Opus 5.5
Director: orchestration, review and export16:49–22:195 h 29 min647702$91.94
Research · Haiku 5.5
Research: Stewart kings to James V17:00–17:3636 min32107$0.52
Research: traditional tunes in ABC17:06–17:5347 min2463$0.49
Research: Bruce and independence17:00–17:5353 min30122$0.87
Research: Mary, Queen of Scots17:00–17:4848 min3895$0.59
Research: toy theatre and Stevenson17:06–17:4842 min2166$0.45
Research: James VI and the Honours17:00–17:5454 min25114$1.24
Research: Alexander III and succession17:00–17:4343 min3091$0.46
Research: origins and Scone17:00–17:3434 min1359$0.66
Research: Macbeth, Margaret, David I17:00–17:3435 min1979$0.58
Fact check · Haiku 5.5
Fact-check five script claims18:02–18:2927 min1355$0.24
Animation, one agent per act · Sonnet 5.5
Animate Act III part B19:01–20:321 h 30 min161194$8.32
Animate Act VI19:29–21:091 h 39 min188204$13.02
Animate Act IV part B19:01–20:181 h 17 min128167$8.18
Animate Act I: The Stone18:24–20:302 h 06 min221247$17.56
Animate Act V part A19:29–20:481 h 18 min146171$9.71
Animate Act III part A19:01–20:251 h 24 min151169$10.54
Animate Act IV part A19:01–20:501 h 49 min177206$12.08
Animate Act II: Gold into Lead18:24–20:362 h 12 min177227$14.26
Animate Act V part B19:29–21:131 h 44 min179197$14.51
20 agents2,4203,335$206.21
The bill

Where $206.21 went

By role

Director · $91.94 · 45%Script, paper-theatre engine, plates, score, mix, the prologue and epilogue, review of every act, export and publishing.
9 Haiku researchers · $5.86 · 3%One note each, from Scone and Macbeth to James VI, the Honours and the toy theatre, plus the traditional tunes in ABC.
1 Haiku checker · $0.24 · 0.1%Five script claims read against primary texts.
9 act animators · $108.18 · 52%Acts I, II and VI one agent each; III, IV and V split into parts A and B.

By model

Opus 5.5 · 1 agent$91.94 · 45%
Sonnet 5.5 · 9 agents$108.18 · 52%
Haiku 5.5 · 10 agents$6.09 · 3%

By kind of token

Cache reads (1,016 million tokens)$129.57
Cache writes (22 million tokens)$47.70
Output, thinking included (2.3 million tokens)$24.16

The director, hour by hour

engine and film code1 h 05 min · 23%
waiting on agents53 min · 18%
review and checking35 min · 12%
score and sound35 min · 12%
plates and images (generation)34 min · 12%
research, script and brief21 min · 7%
export, video and publishing15 min · 5%
agents: brief, launch, message10 min · 4%
waiting on the user7 min · 2%
clean-up6 min · 2%
docs, memory and notes5 min · 2%
other and unclassified2 min · 1%

Each gap between two transcript events is given to the tool call or message that ended it, then sorted by tool name, file path and command. The buckets are keyword matches, so read them as proportions. 4 h 48 min in all.

The animators

How long each act took

Each ran from its first brief to its handback, with a polish pass on its own act after the director’s review.

Nine act animators

Act II: Gold into Lead2 h 12 min
Act I: The Stone2 h 06 min
Act IV part A1 h 49 min
Act V part B1 h 44 min
Act VI1 h 39 min
Act III part B1 h 30 min
Act III part A1 h 24 min
Act V part A1 h 18 min
Act IV part B1 h 17 min
The film

What the evening produced

Nine frames from the film: a red stage curtain in a gilded proscenium, Duncan I standing among bare trees and standing stones, Edward I on a city wall with a row of playing-card kings, black schiltrons against a gold sky, a mounted legend at Stirling, Mary and her women in a harbour fog at Leith, a red banner reading En ma fin est, a vault with an open chest, and a loch glen with a small crown above silhouetted figures.
Nine frames from 20 s to 1,140 s: the curtain, Duncan, Edward I, Bannockburn, Stirling, Leith, the motto, the chest and the glen.

The local machine

191 images

50 plates from 17:55 to 20:58 and 141 sprites, with 131 voice lines and 27 ambience and effects clips. Backcloths and cut-out pieces are generated; all motion, light, weather and type are drawn in code.

The engine

Paper theatre

A toy-theatre engine adapted from The Long Reach: flats slide on from the wings, backcloths drop from the flies, and dramatic moments turn the figures into black silhouettes. Palatino type, with TeX Gyre Pagella embedded as a fallback.

The sound

Code score, bf_emma

A Scottish bed composed in code (reed drone and chanter, clarsach, fiddle, drums, a recurring crown motif, traditional airs from ABC) under Kokoro bf_emma at speed 0.9, with three character voices for attributed quotations.

The MP4

19:17, 993 MB

1376 by 768 at 30 fps, 6.7 Mbit/s with AAC audio; the first export wrote 34,710 frames in 45 minutes. The single-file page is 124 MB.

What the checks caught

Eight findings from the fact check and the reviews

The director read each act from a whole-act contact sheet and sent a numbered polish list to its own animator; the fact check read five claims against primary texts.

FoundWhat went wrongHow it was fixed
Fact check, John DamianThe excuse for his failed flight from Stirling was paraphrased loosely, and 1507 is the modern date, not Lesley’sThe line now has him fall and break his thigh, blaming hens’ feathers that longed for the midden, not the sky, as Lesley tells it
Fact check, Carey’s ride“Galloped” overstated it, and his memoir gives no total durationThe narration says a rider spent three days in the saddle riding north to tell James he was King of England as well
Fact check, James V’s last words“It cam wi’ a lass” is not in Knox, who has “it came from a woman; and it will end in a woman”The quotation follows Knox: “It came from a woman, and it will end in a woman”
Fact check, the Coronation ChairIt was made to hold the Stone (about 1297 to 1300), but it was not Edward I’s own coronation chairThe narration says Edward had a coronation chair made to hold the Stone
Flat and ghostly piecesFlags drawn as flat filled shapes, Picts that read as “penny plain” ghosts, a flat bright-green mound at Scone, genealogy words that read as scratchesFlags engraved, the Picts repainted, the mound and the genealogy redrawn; rule 3 of the director’s notes came out of it
Black frames and wrong depthBlack frames between scenes, and pieces hidden behind a backcloth’s near layerScene changes made by stage machinery, and a depth rule written into the director’s notes
Sound levelsA ten-gun salvo, battles and 50 hammer blows set the bed’s peak 10 dB over the music, so the music dropped and the mixer would have lowered the whole filmOne-shot clips trimmed to their first strike and the effects bus soft-limited at −3 dB, 6 dB under the music
Left inA few first-draw hitches of under a second when a shot’s cards are built, and a code-drawn Langside clothListed as known small things in the director’s notes
What it taught

Six things the numbers say

Method

How these figures were measured

Measured from the session's own transcripts: the director's, and one for each of the 19 agents it launched. Assistant messages are deduplicated by message id and the last entry's usage is kept. Models come from each agent's metadata. The session ran in the crown-scotland-showcase worktree and continued once after its context was summarised; both halves are in the same transcript. Gaps of more than 20 minutes (41.7 minutes in all) are left out of the hour-by-hour split.

Cost is what the usage would have cost at API list prices in October 2026: Opus 5.5 at $4 per million input tokens and $20 per million output, Sonnet 5.5 at $2 and $10, Haiku 5.5 at $0.10 and $0.50 (higher rates above a 100,000-token prompt), cache reads at 0.05 times input, cache writes at 1.25 times (five minutes) or 2 times (one hour). A subscription session was not billed this way, so treat the dollar figures as a measure of work, not an invoice.

Times are UK local time (BST, UTC+1). Local-machine marks are the modification times of the generated files. The findings come from the project’s POLISH.md, DIRECTOR_NOTES.md and research/10_fact_check.md, and the film figures from the lock file and ffprobe of the published MP4. The hour-by-hour split sorts tool calls by keyword, and gaps of more than 20 minutes with no events are left out of it.