Visiowave AI Studio (AISAS)

Production-focused AI cinematography studio for image/video generation, shot orchestration, sequence building, and export.

What We Have Built So Far

Core Product

• Authenticated dashboard with Projects -> Scenes -> Shots workflow.
• Shot Builder with structured cinematic controls (shot size, lens, movement, lighting, mood, etc.).
• Fast Track Video Studio for quick direct-to-video generation.
• Reusable element/reference upload flow for scene consistency.
• Shot list with generation options, approval workflow, compare/pin, and animate-to-video actions.
• Sequence builder and sequence export pipeline.
• Inner Circle waitlist capture flow.

Studio AD (AI Assistant Director)

• Assistant Director panel embedded in Shot Builder and Fast Track.
• /api/ad/direct-shot endpoint for shot direction packets.
• Structured packet output:
  • strategy
  • shot strategy (framing/lens/movement/lighting/mood/composition)
  • master prompt + negative prompt
  • variants
  • technical metadata
  • production scores
  • director notes
• Critic/Refiner pass:
  • Detects contradictions and low-readiness outputs.
  • Auto-refines packet before returning when needed.
• Auth + throttling on AD route.
• AD packet persistence/history:
  • Saved packet metadata + scores in database.
  • Recent directs surfaced in UI.

UI/UX Upgrades (Current)

• Premium design system pass:
  • new button variants (studio, studioSecondary, studioGhost)
  • new card/field/chip utility styles (studio-card, studio-subcard, studio-field, studio-chip)
• Shot Builder + Assistant Director layout cleanup for less crowded composition.
• Shot generation/approval cards redesigned for clearer hierarchy and presentation readiness.

Stack

• Next.js 16 (App Router)
• React 19
• TypeScript
• Tailwind CSS
• Supabase (Auth, Postgres, Storage, RLS)
• OpenAI SDK

Local Setup

1. Install dependencies:

npm install

2. Copy env template:

cp .env.local.example .env.local

3. Fill required variables in .env.local:

• NEXT_PUBLIC_SUPABASE_URL
• NEXT_PUBLIC_SUPABASE_ANON_KEY
• ENCRYPTION_KEY

Recommended/optional:

• OPENAI_API_KEY (required for Studio AD/OpenAI generation features)
• KIE_AI_API_KEY
• TYPESAFE_API_KEY (optional; enables Jev for server-side typed decisions)
• EXPORT_WORKER_SECRET or CRON_SECRET
• SUPABASE_SERVICE_ROLE_KEY (required for admin/server-only flows such as waitlist/admin operations)
• STUDIO_AD_MODEL (optional override; default in code)
• STUDIO_AD_REFINER_MODEL (optional override; defaults to STUDIO_AD_MODEL)

4. Run dev server:

npm run dev

Database Migrations

Migrations live in:

• src/infrastructure/supabase/migrations

Important recent migration:

• 0016_studio_ad_packets.sql (Studio AD packet history table + RLS policies)

Apply migrations in your Supabase workflow before testing new AD history features.

Build

npm run build
npm run start

API Surface (Key Routes)

• POST /api/ad/direct-shot - Generate Studio AD packet.
• GET /api/ad/direct-shot?limit=... - Fetch recent Studio AD directs for signed-in user.
• POST /api/inner-circle - Submit Inner Circle waitlist lead.
• POST /api/exports/worker - Export queue worker execution.

Jev decision layer

Jev is wired as a server-only client at src/core/services/jev.ts. Use it for bounded decisions—routing, scoring, risk gates, and human escalation—not prose generation or direct execution. Add TYPESAFE_API_KEY to .env.local, then call decideWithJev with one shared state object and all independent questions in the same request.

import { decideWithJev, requiresHumanReview } from "@/core/services/jev"

const result = await decideWithJev({ job, render }, {
  next_action: {
    type: "choice",
    instructions: "Which safe next step should this production job take?",
    criteria: {
      continue: "All prerequisites are present and the render may proceed.",
      repair: "The job needs a prompt or continuity revision before rendering.",
      review: "The job has a material ambiguity that needs a human decision.",
    },
  },
  continuity_risk: {
    type: "score",
    instructions: "Rate the risk that the supplied production plan loses visual continuity.",
    criteria: ["Low risk", "Moderate risk", "High risk"],
  },
})

const action = result.answers.next_action
if (requiresHumanReview(action) || action.choice === "review") {
  // Queue a human review. Code, not the model, controls execution.
}

Deploy (Vercel)

• Production branch: main
• Project root: repo root
• vercel.json includes export worker cron:
  • */5 * * * * -> /api/exports/worker?limit=3
• Set environment variables in Vercel:
  • Supabase keys
  • model/provider keys
  • CRON_SECRET or EXPORT_WORKER_SECRET

Notes

• API keys are encrypted at rest.
• Provider connection tests are available in Dashboard Settings.
• Export queue supports manual run and background worker processing.
• Studio AD now saves prompt packets and readiness scores for continuity and reuse.