FLUX DOCUMENTATION SYSTEM Layer 8 — ROADMAP | roadmap flux.dantesisofo.com/wiki/roadmap/

ROADMAP

Where FLUX is going.

1. CURRENT PRIORITY

Infrastructure and corpus bootstrap phase.

Hardware has been ordered (Mac mini M4 Pro + Synology NAS). When it arrives, the priority shifts from publishing to archive infrastructure.

Establish NAS canonical folder structure
Migrate ~400,000 loose Ricoh JPEGs from SanDisk SSD into organized corpus
Integrate existing keeper archive (~15,000 selections) as labeled metadata
Build SQLite metadata layer
Begin autonomous ingest pipeline

The publishing system (portal, generator, catalog) is stable and deployed. The next phase is building the archive foundation that the publishing system will eventually draw from automatically.

See BOOTSTRAP for the full infrastructure plan.

2. SUMMER 2026

Hardware arrives. Infrastructure begins.

Setup NAS + Mac mini as permanent FLUX node
Migrate full SSD corpus onto NAS
Match keeper archive against full corpus
Build metadata database
Begin embedding generation
Continue growing the public catalog
Begin first FLUX Node institutional proposal

This is infrastructure mode, not launch mode.

3. V1 — CURRENT STATE

Personal FLUX generator: operational (423+ issues)
FLUX archive: live at flux.dantesisofo.com
Collaborative projects: Broad Street In Flux published
Protocol page: locked
Zine format: locked
Wiki: complete
Public generator: live — browser-based, no account, 36 JPEGs → full issue + submission
Public catalog: live — 18 entries (CAT_001–CAT_018), auto-generated at 36 unassigned frames
Portal: live — private command center for photo upload, approval, issue monitoring
Auto-generation: fully autonomous — portal approvals trigger issue build automatically when 36 frames accumulate

4. V2 — PUBLIC GENERATOR + CATALOG

Status: Complete. Live at https://flux.dantesisofo.com/generator/

The public generator is browser-based, requires no account, accepts exactly 36 JPEG photographs, and produces a complete 44-page FLUX issue identical in structure to a personal volume.

After generating, photographers submit directly to the public FLUX catalog. Every submission is reviewed before publication. Approved issues receive a permanent catalog number (CAT_NNN) and a public catalog page at flux.dantesisofo.com/catalog/CAT_NNN/.

The protocol QR code in every FLUX issue — personal and public — points here. Anyone who scans it can run the protocol.

See GENERATOR and PUBLIC CATALOG for details.

5. V2 — FLUX NODE (PROTOTYPE)

Physical archive terminal in one public institution.

Start extremely small. One successful node is more important than scale.

Target: a library or camera store in Philadelphia.

6. V3 — INFRASTRUCTURE

Hardware ordered (arriving):

Apple Mac mini M4 Pro — permanent FLUX compute node
Synology NAS + 2x mirrored IronWolf drives — canonical archive storage

Architecture:

NAS      = canonical archive (all originals, metadata, embeddings)
Mac mini = compute node (generation, processing, automation)
S3       = public distribution layer (continues unchanged)

Canonical NAS structure:

/FLUX_ARCHIVE/       ← every photograph ever made
/FLUX_SYSTEM/        ← scripts and tools
/FLUX_METADATA/      ← SQLite databases
/FLUX_EMBEDDINGS/    ← vector databases
/FLUX_ISSUES/        ← generated PDFs and manifests
/FLUX_INBOX/         ← watch folder for autonomous ingest

Ideal flow (target state):

Photograph with Ricoh GR
Transfer via iPhone / SD card reader → /FLUX_INBOX/
FLUX node detects new files
Automatic EXIF extraction, organization, renaming, hashing
Embeddings generated
Keeper scores computed
Issue draft generated when threshold reached
PDF + assets uploaded to S3
Archive updated
FLUX Node sync (if applicable)

See BOOTSTRAP for the full phased plan.

7. V3 — COLLABORATOR ONBOARDING

The public generator exists. The catalog is live. The first community issues are published.

The next phase is community development:

Students and photographers begin making their own issues regularly
The catalog grows as a curated public record of photographers running the protocol
The community catalog and the personal archive remain separate namespaces — CAT_NNN vs FLUX_NNN
Each community photographer develops their own sequential catalog of issues over time

The naming convention for community issues is resolved: the catalog assigns CAT_NNN in approval order. Future infrastructure may support per-photographer sequential numbering — a photographer's own FLUX_001, FLUX_002 — without conflicting with Dante Sisofo's personal sequence.

8. V4 — AI SELECTION

Train a keeper model on:

Full corpus (~400,000 frames)
Keeper selections (~15,000 labeled positives)
Rejected frames (implicit negatives)
Chronological sequences
Metadata and visual embeddings

Goal:

Learn what the photographer keeps
Learn what the photographer rejects
Learn sequencing rhythm
Automate culling
Generate issue drafts automatically

The keeper model answers:

"What would Dante probably keep?"

Not: "What is objectively great photography?"

This distinction is critical. The model learns personal visual taste, not universal aesthetics. The keeper archive is training data. The full corpus is the test set.

The dream:

Go out and photograph. FLUX does the rest.

This is not the immediate priority. This is V4 or V5. The prerequisite is a fully organized corpus with keeper metadata attached — Phase 3 of the Bootstrap plan.

See INTELLIGENCE for the full intelligence layer specification.

9. OPEN SOURCE

FLUX should become open and upgradeable.

Not a closed social platform.
Not the next Instagram.

More like: - open protocol - open publishing system - open archive format - open photographic workflow - open-source cultural infrastructure

People should be able to use it, fork it, improve it, generate their own zines, publish their own archives, contribute fixes, build on top of it.

The generator code, the protocol specification, and the archive format should all be publicly documented and reproducible.

10. WHAT NOT TO BUILD

No social feed
No likes or engagement metrics
No algorithmic ranking
No advertising layer
No subscription paywall for the archive
No proprietary file formats
No vendor lock-in

11. FINAL PRINCIPLE

The stronger the archive system becomes:

the less explanation is needed
the less manifesto is needed
the more self-evident the concept becomes

The archive itself creates belief.

Document	Layer	Relationship
BOOTSTRAP	Layer 4 — Infrastructure	The phased infrastructure plan this roadmap references
INTELLIGENCE	Layer 5 — Intelligence	The AI selection layer described in V4
PRESERVATION	Layer 7 — Preservation	The verification infrastructure on the roadmap
NODES	Layer 4 — Infrastructure	The node prototype on the V2 roadmap
MANIFESTO	Layer 1 — Foundation	The open protocol vision the roadmap is implementing