Open the live demo → · 60-second tour · Working case study · Workflow · Discuss
Proposed and system-architected by AI Creator BAO · Developed and operated by STUDIO GENESIS
LIGHTBRINGER is the production control layer between an approved screenplay and image/video generation. It keeps shots editable, identities synchronized, costs visible, and every approval or retry traceable.
The original public demo ORBITAL ECHO follows one 42-second scene through the real LIGHTBRINGER workflow. It uses isolated deterministic data, so no client work, private account, credential, or external model credit is exposed.
| Screenplay | Segments | Shots | Asset identities | Storyboard results | Failed stages |
|---|---|---|---|---|---|
| 42 sec | 2 | 6 | 5 | 3 | 0 |
Start with the 60-second visual tour, then inspect the complete English or Korean case study.
The repository now includes a deployable clean-room demo with original dummy data. Node.js 20 or newer is the only local requirement.
Launch the public LIGHTBRINGER demo
npm startOpen http://localhost:4173. The default deterministic Mock API costs nothing. To connect your own LLM, store its key in server-side secret environment variables—never in the browser—and select either the Anthropic or OpenAI-compatible adapter.
Self-hosting and BYO API guide →
approved screenplay
↓
segment explorer → shot design + timing
↓
continuity + versioned @asset identities
↓
Korean review master + English generation master
↓
storyboard / previz → human cost approval → video render
↓
result history, retry, version, recovery
| Production stage | Human decision | System output |
|---|---|---|
| Screenplay | Writers and reviewers approve the source | Versioned screenplay and review history |
| Shot design | Artists edit split, camera, direction, dialogue, and timing | Navigable shot records |
| Continuity | The team confirms visible characters and locations per shot | Locked @asset identities and state |
| Prompt master | Directors review intent before generation | Korean, English-video, and storyboard masters |
| Render | A producer reviews targets and estimated cost | Preserved outputs, attempts, failures, and lineage |
Generating a single AI shot is fast. Producing a coherent episode is an operations problem.
| Production failure | LIGHTBRINGER control |
|---|---|
| Characters and locations drift between shots | Versioned shot-level asset locks |
| Long scripts become hundreds of unmanageable cards | Segment explorer with one selected work unit |
| LLM output is truncated or times out | Fast batch path with targeted shot-level fallback |
| Korean creative intent diverges from English engine input | Parallel review and generation masters |
| Expensive renders are repeated too early | Human approval gate before paid generation |
| Completed work disappears during retry or deletion | Persistent history, trash, recovery, and lineage |
See the full workflow specification and reference architecture.
This repository is a public workflow specification, working-product case study, and clean-room runnable demo, not the commercial application source distribution.
Included:
- real application screenshots in Korean and English;
- an original neutral screenplay and editable shot records;
- bilingual prompt-master examples;
- a deployable dummy UI, deterministic Mock API, and server-side BYO API adapters;
- workflow, architecture, adoption, and evaluation documents;
- issue templates for production feedback.
Not included:
- customer or unreleased production data;
- reference media and private asset libraries;
- accounts, audit logs, admin routes, or API credentials;
- the commercial LIGHTBRINGER application source.
| If you want to… | Start here |
|---|---|
| Understand the product quickly | 60-second product tour |
| Inspect a complete run | Working demo case study |
| Read the production rules | Workflow specification |
| Evaluate system boundaries | Architecture |
| Adopt the stages in another stack | Adoption guide |
| Evaluate outcomes without inflated claims | Industry impact and metrics |
| Inspect machine-readable examples | examples/orbital-echo |
| Run safely or connect your own API | Self-hosted demo guide |
LIGHTBRINGER is a working internal production system under active development at STUDIO GENESIS. The public framework is currently preparing the bilingual working-demo release.
- Review the roadmap and changelog.
- Share a bottleneck through the workflow-feedback issue form.
- Start a broader conversation in GitHub Discussions.
- Read CONTRIBUTING.md before proposing a change.
AI Creator BAO is the proposer and system architect of LIGHTBRINGER. STUDIO GENESIS develops and operates the system, connecting human creative decisions with reproducible production data.
STUDIO GENESIS website · Citation · License
The public documentation and diagrams are available under the repository license. LIGHTBRINGER and STUDIO GENESIS brand identifiers are excluded from the license grant.