FORGE AI Platform

FORGE AI is an advanced, unified generative AI platform designed for creative professionals. It integrates state-of-the-art AI models for video generation, image upscaling, background removal, and audio processing into a single, cohesive interface.

🚀 Key Features

🎬 Video Generation

• Runway Integration:
  • Gen-4 Turbo (Image-to-Video): High-fidelity generation with native auto-cropping and advanced camera controls.
  • Veo 3 & 3.1 (Runway): Generation using text or image inputs with native 720p support.
• Google Veo Integration (Native): Access Google's Veo models directly via Vertex AI.
• Smart Processing: Automatic aspect ratio handling and image resizing to meet strict model requirements.

🖼️ Image Tools

• Upscaling: Professional-grade upscaling using Topaz Photo AI integration (Face Recovery, Denoising).
• Background Removal: Multi-provider support (Clipping Magic, Bria AI) for precise subject isolation.
• Generation: Multi-model image generation (OpenAI DALL-E 3, Stable Diffusion, etc.).

🔊 Audio & Utilities

• Voice-to-Text: Transcription using OpenAI Whisper.
• Text-to-Speech: High-quality voice synthesis via ElevenLabs.
• Subtitle Processor: Automatic subtitle generation and burning for videos.
• Prompt Studio: AI-powered prompt enhancement and management.

---

🏗️ Architecture

FORGE AI is built as a containerized microservices application using Docker Compose.

Tech Stack

• Frontend: Next.js 14 (React), TypeScript, Tailwind CSS. Served via forge-frontend.
• Backend: FastAPI (Python 3.11). Handles API orchestration, job management, and third-party integrations. Served via forge-backend.
• Database: PostgreSQL 16. Stores Jobs, Assets, Users, and Projects.
• Cache/Queue: Redis. Manages Celery background tasks and caching.
• Reverse Proxy: Nginx. Routes traffic and handles static assets.

Data Flow

1. User Request: User interacts with the Next.js UI.
2. API Call: Frontend sends request to forge-backend (FastAPI).
3. Job Creation: Backend validates input (Pydantic) and creates a Job record in PostgreSQL.
4. Async Processing: complex tasks (Video Gen, Upscaling) are queued in Redis/Celery.
5. External APIs: Worker nodes call APIs (Runway, Google, Topaz, etc.).
6. Asset Storage: Resulting files are stored in the assets/ volume and indexed in the DB.
7. Notification: Frontend polls or receives socket updates (planned) for job completion.

---

🔒 Security & Configuration

• Environment Variables: extensive configuration via .env files.
• Database Security: User/Password authentication for Postgres.
• Volume Management: Persistent storage for Database (postgres_data) and Assets (assets_data).

---

📚 Documentation

• Installation Guide - How to set up and run FORGE AI.
• API Documentation - Details on backend endpoints.
• Frontend Guide - UI development/components.

---

© 2025 BTG Unified Platform