PROJECT-HELIX

🧬

"The MP3 for Images" - TechCrunch (Hypothetical) "Up to 10x semantic compression" - Benchmark Results "Preserves identity perfectly" - AI Weekly

HELIX is a semantic compression system that converts media into compact blueprints (.hlx) storing only identity-critical information. Unlike traditional compression (JPEG, MP4), HELIX separates "High Entropy" (faces, text, unique features) from "Low Entropy" (sky, walls, noise).

"Store the instructions for regeneration, not the pixels themselves."

Two Killer Features

Feature	Description	Target
📱 Cross-Platform	HLX files are valid PNGs viewable by ANY app	Adoption
🧠 HELIX SDK	Up to 10x semantic compression for AI training	Enterprise

Quick Install

pip install helix-sdk

🚀 Quick Start

Prerequisites

• Python 3.10+
• Node.js 18+
• Gemini API Key

Installation

# Clone
git clone https://github.com/DB0SZ1/PROJECT-HELIX.git
cd PROJECT-HELIX

# Backend
pip install -r requirements.txt

# Frontend
cd ui/helix-app
npm install

Environment Setup

Create .env file:

GEMINI_API_KEY=your_api_key_here

Run

# Terminal 1: Backend
python -m uvicorn src.api.server:app --port 8001 --reload

# Terminal 2: Frontend
cd ui/helix-app
npm run dev

Visit http://localhost:3000

📱 Cross-Platform Compatibility (HLX v2)

HELIX v2 files are valid JPEGs that any app can display!

┌─────────────────────────────┐
│  JPEG Image Data            │  ← Any app sees this
│  (FF D8 ... FF D9)          │
├─────────────────────────────┤
│  HELIX_DATA_START           │  ← Our marker
├─────────────────────────────┤
│  Encrypted Blueprint        │  ← Only HELIX can read
└─────────────────────────────┘

API Endpoints

Endpoint	Method	Description
/api/encode/v2	POST	Encode image to cross-platform HLX
/api/hlx/preview	POST	Extract JPEG preview from HLX v2
/api/hlx/format-info	GET	Get format version details

📦 HELIX SDK for AI Training

Compress your training data up to 10x while maintaining identity:

Remote Mode (Lightweight - Recommended)

from helix_sdk import HelixSDK

# Connect to HELIX API - no heavy local dependencies!
sdk = HelixSDK(base_url="https://api.helix-codec.dev")

# Compress any image
sdk.compress("photo.jpg", "photo.hlx")

# Materialize at any resolution
sdk.materialize("photo.hlx", "photo_4k.png", resolution="4K")

Batch Processing for Data Centers

from helix_sdk import HelixDataset, HelixLoader, BatchCompressor

# Compress your dataset (one-time)
compressor = BatchCompressor(workers=8)
stats = compressor.compress_directory(
    input_dir="/data/imagenet/train/",   # 150GB
    output_dir="/data/imagenet_hlx/"     # ~15GB
)
print(f"Saved {stats.space_saved_percent}%!")

# Use in training (PyTorch compatible)
dataset = HelixDataset("/data/imagenet_hlx/", target_resolution="512p")
loader = HelixLoader(dataset, batch_size=64, num_workers=4)

for batch in loader:
    model.train(batch)

SDK Benefits

• 📦 2-5x semantic compression
• 🌐 Remote or Local mode
• 🚀 PyTorch/TensorFlow compatible
• 🔄 On-demand materialization at any resolution
• ♾️ Infinite variant generation (free augmentation)
• 🔍 Semantic search across dataset

🔌 API Endpoints

Endpoint	Method	Description
/	GET	Health check with feature status
/api/encode	POST	Encode image to HLX
/api/encode/v2	POST	Encode to cross-platform HLX
/api/materialize	POST	Reconstruct image from HLX
/api/sdk/info	GET	SDK documentation

🏗️ Architecture

graph TD
    User["📱 User / Client"] -->|Upload Image| API["⚡ HELIX API (FastAPI)"]
    API -->|Extract Identity| GeminiExtract["🧠 Gemini 3 Pro (Vision)"]
    GeminiExtract -->|Anchors + Metadata| Blueprinter["📝 Blueprint Generator"]
    Blueprinter -->|Encrypted .hlx| Storage["💾 Storage (10x Compressed)"]
    
    Storage -->|Load .hlx| Materializer["🏗️ Materializer Engine"]
    Materializer -->|Stitch Anchors| Stitcher["🧵 Deterministic Stitcher"]
    Stitcher -->|Baseline Image| Output["🖼️ High-Fidelity Image"]
    Stitcher -.->|Optional Upgrade| GeminiRefine["✨ Gemini 3 Pro / Stealth"]
    GeminiRefine -->|Enhanced Image| Output

Loading

📊 Benchmark Results

Metric	Value
Extreme Compression	Up to 10x smaller than JPEG (V1 format)
Identity Preservation	0.85+
Materialization Time	~3s

🛡️ Security

• AES-256-GCM encryption for blueprints
• HMAC-SHA256 tamper detection
• Secure key derivation
• AES-256 encrypted anchors keep faces secure

☁️ Production Deployment

Backend (Render / Railway)

1. Deploy to Render:
   • Connect GitHub repo
   • Select render.yaml if prompted, or choose "Web Service"
   • Runtime: Python 3
   • Build Command: pip install -r requirements.txt
   • Start Command: uvicorn src.api.server:app --host 0.0.0.0 --port $PORT
   • Environment Variables:
     • GEMINI_API_KEY: Your key
     • ELEVENLABS_API_KEY: Your key (optional)
     • CORS_ORIGINS: https://your-frontend.vercel.app (comma separated)
     • HELIX_SECRET_KEY: Random string for encryption
   • Persistence: Add a Disk named helix-data mounted at /data for SQLite persistence.

Frontend (Vercel)

1. Deploy to Vercel:

   • Connect GitHub repo
   • Framework Preset: Next.js
   • Root Directory: ui/helix-app
   • Environment Variables:
     • NEXT_PUBLIC_API_URL: Your backend URL (e.g., https://helix-backend.onrender.com)

2. Custom Domain:

   • Go to Vercel Project Settings > Domains
   • Add your custom domain (e.g., helix.yourdomain.com)
   • Update your DNS nameservers or A records as instructed by Vercel

👥 Team

Built By the Helix Team

---

HELIX: Your memories, compressed for the future.