Five channels. One five-stage pipeline. Every episode is auditable end to end, the brief, the frames, the gate, the post. Three are live. Two are test-beds where the format is the work.
Dark history. The events, regimes and people history skipped over.
The first episode kept 65.3% of viewers past the midpoint on Musashi's duel with Kojirō. Retention is the format signal.
LIVE · EP_MUSASHI · 40.1s · 876 first-day views
View the 19-shot breakdown →Tested objects. Watches, engines, instruments. Everything claims it works. We put it through the test.
Episode one is built, gated, and ready to upload. The QA gate ran 9 of 9 frames against the reference bank at vs-seed and vs-hero.
READY · EP_SPEEDMASTER · 40.0s · 23 shots, 148 words
View the breakdown →Demon Slayer lore, vertical narration-driven. Hard cuts on beat boundaries, not the blur-whip.
13 episodes built and gated, proof the pipeline can ship a single franchise to format. Subject ratios measured against the channel median: Tengen 3.77x, Sanemi 1.91x, Upper Moons 1.47x.
READY · EP_OBANAI · 40.0s · 13 episodes built, awaiting a YouTube channel
Format research. 3.00-second grid, blur-whip transitions, one red word at 140px. The opposite of Zenpai, on purpose.
4 episode folders (Gojo, Madara, Sukuna, Ben 10), the format itself is the experiment. No subject catalogue yet; building it is the next move.
R&D · EP_GOJO · 48.0s · 4 experimental episodes
Money, economy, companies that failed. Money/economy measured 0.09x against the channel median, this needs its own scrape before it ships.
Not started. The pipeline is ready. The subject list is the work.
EMPTY · Awaiting scrape + subject list
Plan, select, build, gate, post. Every episode on every channel goes through these five stages. The gate is the load-bearing one, that's why it carries the only warm colour in the diagram.
The gate runs ResNet50 cosine similarity at three frame positions per scene, then asks a vision model for a modern/anachronism check. The pass/fail criterion is calibrated against the reverse-engineered luxury episode:
| Scene | t = 0% | t = 50% | t = 98% |
|---|---|---|---|
| s1_establish | 0.325 / 0.317 | 0.324 / 0.316 | 0.329 / 0.320 |
| s2_deepen | 0.335 / 0.327 | 0.353 / 0.346 | 0.369 / 0.362 |
| s3_exit | 0.231 / 0.230 | 0.228 / 0.226 | 0.211 / 0.213 |
The numbers are vs-seed / vs-hero. The third scene has the lowest scores and that is correct, a character walking away should diverge from the reference. The gate flags catastrophic divergence, not stylistic movement.
Each rule closes one class of failure that cost real time or money. Every rule is documented, fileable, and automatic, not convention.