Thoughts

UFO UAP RV YT Pipeline - Phase 0 Complete (2026-03-16) Project: Automated YouTube Shorts pipeline. Takes content from a NotebookLM knowledge base (75+ UFO/UAP/remote viewing books) and produces finished YouTube Shorts with minimal human intervention. Architecture decided: Railway Postgres + Cloudflare R2 (not Supabase). Chose this over Supabase to avoid the 2-active-project limit and because pipeline orchestration involves long-running async jobs that would exceed Supabase edge function timeouts. Infrastructure set up: - Railway project: ufo-uap-yt-pipeline (ID: e86c55d4-f102-4c34-8b13-c21be10d5e4d) - Railway Postgres: public proxy at switchyard.proxy.rlwy.net:58688 - Cloudflare R2: subscribed, bucket ufo-uap-pipeline-assets created - 5 database tables created: characters, character_outfits, shorts, short_assets, pipeline_runs - Google Sheet production manifest schema defined (columns for script, character, visual plan, publishing metadata) Project location: ~/Library/Mobile Documents/com~apple~CloudDocs/Projects/UFO UAP/UFO UAP RV YT Pipeline/ Next: Phase 1 is character design (LoRA selection for art style, portrait generation for 7 alien characters, ElevenLabs voice assignment). Requires Dave's creative input.

UFO UAP RV YouTube Pipeline - Pre-Build Complete (2026-03-16) Project: Automated faceless YouTube Shorts channel mining a NotebookLM knowledge base (75+ UFO/UAP/remote viewing books) for content. Pipeline produces finished shorts with minimal human intervention. Location: /Users/davetedder/Library/Mobile Documents/com~apple~CloudDocs/Projects/UFO UAP/UFO UAP RV YT Pipeline ARCHITECTURE DECIDED: Manual touchpoint (weekly, 30-40 min): Gemini + NotebookLM session. Multiple Gems handle extraction, scripting, grading/culling, visual planning, and character casting. Output is a complete production manifest in a Google Sheet with finished scripts, visual asset plans, character assignments, outfit selections, and environment choices. This is the only manual step. Automated pipeline (runs itself): 1. Google Sheets API reads new production manifest rows into Supabase 2. ElevenLabs API (Creator plan, $22/mo) generates voiceover per character voice 3. Whisper API generates word-level caption timestamps 4. Flux API (Kie.ai/BFL) generates 2D still images for Ken Burns shots 5. Luma Dream Machine API generates 3-10 sec video b-roll clips 6. D-ID API generates lip-synced talking head avatar segments from character portraits + audio 7. MiniMax Hailuo API generates character-in-scene clips using subject_reference for consistency 8. Remotion on Railway assembles all assets into finished vertical shorts (React-based, no per-render cost) 9. YouTube Data API v3 publishes with metadata, thumbnails, scheduled timing CHARACTER SYSTEM: 7 initial characters (Reptilian, Mantis, Grey, Little Green, Shadow Person, Tall White/Nordic). All 2D animated, consistent art style enforced via Flux-compatible LoRA. Each character has multiple outfits (space suit, priest, MIB suit, etc.). Unique ElevenLabs voice per character. Character registry in Supabase table. Extensible by adding portrait + voice + DB row. Estimated monthly cost at 25 shorts/week: $100-175 (excluding visual generation API costs which are the biggest variable, roughly $75-200/mo additional depending on asset density per short). APPROACH: Approach C selected. Remotion from day one (no Creatomate). Full API stack, no self-hosted GPU workloads. No Make/Zapier, orchestration owned via Supabase edge functions. BUILD PHASES (10 phases): - Phase 0: Scaffolding + DB schema (partially done, directory structure and git committed) - Phase 1: Character design (LoRA selection, portraits, voices) - needs Dave's creative input - Phase 2: Gemini prompt chain dev (Gems vs Claude skills comparison testing) - Phase 3: ElevenLabs + Whisper integration - Phase 4: Visual asset generation (Flux, Luma, D-ID, Hailuo) - Phase 5: Remotion templates + Railway deployment - Phase 6: Pipeline orchestration - Phase 7: YouTube publishing - Phase 8: End-to-end testing - Phase 9: Intelligence loop (competitor research, performance analytics) CURRENT STATUS: Phase 0 partially complete. Remaining: Supabase tables, RLS, Google Sheet template. Phase 1 next. KEY DECISIONS: - Gemini does ALL creative heavy lifting (extraction, scripting, grading, visual planning, casting) - LoRA enforces consistent 2D animation style across all characters - Luma over Runway for video clips (cheaper, quality sufficient for Shorts) - Both D-ID (talking heads) and Hailuo (character scenes) in stack for different purposes - Gems to be tested against Claude Code skills to find best tool per stage