Section index
Credits and provenance
Draft credit language for team review. No copyright license has been assigned by this document.
Primary creative roles
Artist / project lead
- Originated and directed the project through iterative review.
- Approved the world, image tradition, selections, voice system, evidence-film structure, native aspect ratio, typography, continuous-motion conform, website-slot language, and synchronized timeline experience.
- Supplied the artistic practice and decision framework within which the complete work was developed.
AI production agent
Hermes Agent by Nous Research
- Practice reconnaissance and concept expansion
- TITLES MCP orchestration
- Prompt and script drafting
- Asset/provenance management
- Local procedural sound generation
- FFmpeg/ImageMagick/Python production
- Website and timeline implementation
- Media/browser verification
- Hosting packaging and documentation drafting
The agent's output remained subject to direct artist review and repeated correction.
TITLES artist-model credit
Final canonical portrait tradition
Cut From Elsewhere by @juujuumama
Model ID: ffbe887a-4f38-415b-b45d-2ba4e51ad919
Adapter ID: 1a782b61-3561-411e-8a0d-e2d4170d826e
Architecture recorded by TITLES: flux-2-klein-9b-base
TITLES describes the model as trained exclusively on Sierra Juujuumama's original physical collage works from 2024–2026.
Suggested public credit:
Canonical portraits generated through TITLES using Cut From Elsewhere by @juujuumama.
Visual-tradition tests
The following artist models were tested before the final selection:
Collage Around Me by @sudutalien
Model ID: 7b5efa71-29a1-4976-8e73-f51784027af3
Adapter ID: ab441a44-cf30-4105-bf4d-1ecc2e9f2dfd
Bones and All by @antidote
Model ID: 88a9631b-f355-4631-a115-c892b5e31b82
Adapter ID: b1f5b030-5335-482c-b069-6e3cb29d52ed
These tests informed selection but are not among the sixty final canonical portraits.
Video model
Kling Video v2.6 Pro
Model ID: 249177de-df51-498e-bf9a-abab2ee6a6cf
Adapter ID: 1a102e60-0263-4b0d-88e0-0aaef4789d6d
Role: sixty reference-led five-second canonical manifestations.
Final evidence cast
Caller / private answering machine
- Model: ElevenLabs Turbo v2.5
- Model ID:
546bf826-ed1e-4901-b1a0-ae7403098401 - Adapter ID:
dde7df58-050e-4e65-b664-4d26db4c6fbc - Preset voice: River
Clerk / municipal intercom
- Model: MiniMax Speech 2.8 Turbo
- Model ID:
6b25d7b9-ca34-4caa-8107-a457716863aa - Adapter ID:
70c04ce5-771f-4b4b-bd34-5e377d9f4026 - Preset voice: Patient_Man
Witness One and Witness Two
- Model: Gemini 3.1 Flash TTS
- Model ID:
9abdcf50-5e37-4f5c-bc7f-d9d54babcdd6 - Adapter ID:
0058c2d2-274b-4ed9-a20e-7f290e7b476c - Preset voices: Charon and Aoede
Auditioned speech models
The wider voice study also used:
- MiniMax Speech 2.8 HD —
afa71476-5d33-424d-893e-432aec6c8f8c - Chatterbox HD —
759b9328-7658-459f-a22d-15a932e0b217
TITLES platform credit
Suggested technical credit:
Artist-model discovery, image generation, image-to-video transformation, and speech synthesis were executed through TITLES and the TITLES MCP tool surface.
Persistent session:
f2739b27-bd62-4a36-ad50-cccb6569c290
The Cabinet of Minor Gods — Launch Pantheon
No project asset was published to the public TITLES feed by the agent.
Local production tools
- Python — manifests, deterministic scores, audio relics, cue sheets, batch building, packaging
- FFmpeg / ffprobe — conversion, DSP, mixing, editing, mastering, probing, and full decode verification
- ImageMagick — typography, cards, borders, crops, splits, and contact sheets
- HTML/CSS/JavaScript — original Cabinet and synchronized Evidence Cabinet
- Caddy — byte-range-capable local static serving and deployment QA
- Browser DOM/console inspection — interaction, seeking, responsive layout, and error verification
Fonts
- Website: Arial / Helvetica fallback stack
- Definitive website-slot film: Helvetica Neue from the local macOS system font library
- Earlier superseded conform tests used Avenir Next and Avenir Next Condensed
Confirm font licensing and substitution strategy if editable source materials—not only rendered video and CSS fallback stacks—are distributed beyond the current team.
Music and sound
- Ten chapter scores were procedurally synthesized locally.
- Sixty nonverbal audio relics were procedurally synthesized locally.
- Raw speech was generated in TITLES and transformed locally through project-specific recording environments.
- No TITLES music generation was charged or used in the final work.
Credit questions to resolve before public release
- Final public name and credit order for the artist/project lead.
- Whether the AI production-agent credit appears publicly or only in process documentation.
- Whether @juujuumama receives a dedicated artist-model line on the main page, film credits, or both.
- Whether TITLES requests a specific platform/MCP credit format.
- Whether the generic model providers require trademark or attribution language.
- Whether the project will be released as all-rights-reserved, under a specific license, or privately to the team.
- Whether generated source assets may be redistributed directly under the applicable TITLES/model terms, as distinct from exhibiting the composed websites and film.
Rights note
This package does not grant a public license. Before public distribution, the project lead should review:
- TITLES terms applicable to generated outputs
- Artist-model creator attribution expectations
- Generic model/provider terms
- Rights to the final project title, texts, code, and sound
- Team access and redistribution scope