A cost-optimized AI-driven platform that autonomously generates high-conversion, brand-ready promotional videos. The platform leverages advanced AI to automate the entire video creation pipeline from ideation to final delivery, producing professional-quality 720p advertisement videos at $1.2-1.6 per video with minimal human intervention.
Relicon is an enterprise-grade AI-powered video generation platform that creates 15-20 second promotional videos using modern commercial ad structure (Hook → Problem/Solution → Resolution/CTA). The system processes brand information through product-focused visuals, ElevenLabs voiceover and background music at 720p resolution for cost optimization.
- Cost-Optimized Architecture: Generates 12-16 videos per $20 budget (vs 4-5 previously)
- Modern Ad Structure: Hook → Problem/Solution → Resolution/CTA format with 3 scenes, 6 seconds each
- Product-Focused Visuals: Modern commercial advertisements prioritizing product demonstration over talking heads
- Professional Audio: ElevenLabs TTS with mandatory background music at 15% volume
- Enterprise Logging: ECS-compliant JSON logging with trace correlation across all services
- Real-Time Progress: Enhanced progress tracking with 4-step visual indicators and percentage updates
- Provider Resilience: Hailuo-02 (MiniMax) primary, Luma AI fallback for 720p video generation
| Area | Technology |
|---|---|
| Frontend | Next.js, React, TypeScript, Tailwind CSS |
| Backend | FastAPI, Python, PostgreSQL, Drizzle ORM |
| AI & ML | GPT-4o, LangChain |
| Video | Hailuo-02 (MiniMax), Luma AI, FFmpeg |
| Audio | ElevenLabs, OpenAI TTS |
| Music | ElevenLabs Music Generation |
| Task Queue | Celery, Redis |
| Logging | Structured JSON, ECS-compliant |
relicon/ [v0.9.0]
.
├── assets/
│ ├── relicon-full-logo.png
│ └── relicon-logo.png
├── backend/
│ ├── api/
│ └── core/
├── config/
│ ├── env.example
│ ├── __pycache__/
│ └── settings.py
├── core/
│ ├── assembly_service.py
│ ├── audio_service.py
│ ├── cost_tracker.py
│ ├── __init__.py
│ ├── logger.py
│ ├── music_service.py
│ ├── orchestrator.py
│ ├── planning_service.py
│ ├── provider_manager.py
│ ├── __pycache__/
│ └── video_service.py
├── frontend/
│ ├── index.html
│ ├── package.json
│ └── src/
├── __init__.py
├── interfaces/
│ ├── audio_generator.py
│ ├── __init__.py
│ ├── __pycache__
│ ├── text_generator.py
│ └── video_generator.py
├── outputs/
├── package.json
├── providers/
│ ├── elevenlabs.py
│ ├── hailuo.py
│ ├── __init__.py
│ ├── luma.py
│ ├── openai.py
│ └── __pycache__/
├── README.md
├── requirements.txt
├── test/
│ └── test_logging.py
└── utils/
└── __pycache__/
- Clone the repository:
git clone https://github.com/aegix-ai/relicon.git
cd relicon- Setup environment:
pip install -r requirements.txt
cp config/env.example .env
# Add your API keys in .env- Launch Development Server:
npm run dev
# Server running at http://localhost:5000| Version | Release Date | Major Features |
|---|---|---|
| 1.0.0 | 2025-08-18 | Cost-Optimized AI Ad Creator |
| 2.0.0 | TBD | Advanced Feedback Loop & Performance Tracking |
| 3.0.0 | TBD | Multi-Provider Optimization & Self-Improvement |
| 4.0.0 | TBD | Enterprise Analytics Dashboard & AI Chat Interface |
- Cost Optimization: 720p resolution targeting $1.2-1.6 per 18-second video
- Modern Ad Structure: Hook → Problem/Solution → Resolution/CTA with product focus
- AI Planning: Autonomous creative decision-making using GPT-4o AutonomousVideoArchitect
- Video Generation: Dual-service approach with Hailuo-02 primary, Luma AI fallback
- Audio Processing: ElevenLabs TTS + mandatory background music at optimized volume levels
- Enterprise Logging: Single-line JSON objects with trace correlation and ECS compliance
- Real-Time Progress: 4-step visual indicators with accurate percentage tracking
- Video Assembly: FFmpeg-based assembly with precise 18-second duration control
- Single Ad Focus: ALWAYS creates exactly ONE advertisement with 3 scenes of 6 seconds each
- No Branching: No multiple variations or alternative versions - one cohesive 18-second story
- Cost Controls: $3 maximum, $2 warning threshold with budget validation
- Progress Tracking: Real-time updates every 2 seconds with smooth transitions
- Service Redundancy: Primary/fallback pattern for video generation resilience
- Structured Logging: Domain-specific fields (video., audio., cost., job., orchestration.*)
- Website: www.relicon.co
- Email: [email protected]
- Twitter: @reliconAI
© 2025 Relicon by Aegix Group. All rights reserved.