🎮 Jellyfin AI Upscaler Plugin v1.5.0.2

Caution

🧪 TEST PHASE - v1.5.0.2

This is an EXPERIMENTAL version with the new Docker AI microservice architecture! AI upscaling now runs in a separate Docker container instead of directly in Jellyfin.

🐳 Docker Image: kuscheltier/jellyfin-ai-upscaler

Please report bugs: GitHub Issues

---

🐳 New Architecture: Docker AI Service

Why Docker? (Problem with v1.4.9.x)

Jellyfin's plugin system tries to load ALL .dll files as .NET assemblies. Native C++ libraries (ONNX Runtime, CUDA, OpenCV) caused:

System.BadImageFormatException: Bad IL format
Failed to load assembly "onnxruntime_providers_shared.dll"

Result: Plugin was disabled, no AI upscaling possible.

The Solution: Microservice Architecture

┌──────────────────────────────────────────┐
│  Jellyfin Server                         │
│  ┌────────────────────────────────────┐  │
│  │  AI Upscaler Plugin v1.5.0.0       │  │
│  │  ✅ Only ~1.6 MB (instead of 417MB)│  │
│  │  ✅ No native DLLs                 │  │
│  │  ✅ Sends frames via HTTP          │  │
│  └──────────────┬─────────────────────┘  │
└─────────────────┼────────────────────────┘
                  │ HTTP POST /upscale
                  ▼
┌──────────────────────────────────────────┐
│  AI Upscaler Docker Container            │
│  ┌────────────────────────────────────┐  │
│  │  Python + FastAPI + OpenCV DNN     │  │
│  │  ✅ CUDA / GPU Acceleration        │  │
│  │  ✅ FSRCNN, ESPCN, LapSRN, EDSR    │  │
│  │  ✅ Web UI for Model Management    │  │
│  └────────────────────────────────────┘  │
└──────────────────────────────────────────┘

Benefits

Feature	Old (v1.4.9.x)	New (v1.5.0.0)
ZIP Size	417 MB	~1.6 MB
Native DLLs	In plugin → Crashes	In Docker → Isolated
GPU Support	Issues with Jellyfin	Full CUDA support
Updates	Rebuild plugin	Pull Docker image

---

📥 Installation (2 Steps)

Step 1: Start Docker AI Service

Option A - Docker Hub (easiest):

docker run -d --name jellyfin-ai-upscaler \
  -p 5000:5000 \
  -v ai-models:/app/models \
  kuscheltier/jellyfin-ai-upscaler:latest

Option B - Build locally:

cd docker-ai-service
docker-compose up -d --build

Open http://YOUR_SERVER_IP:5000 to see the Web UI.

Step 2: Install Jellyfin Plugin

1. Open Jellyfin Dashboard → Plugins → Repositories → Add
2. Enter URL:

   https://raw.githubusercontent.com/Kuschel-code/JellyfinUpscalerPlugin/main/manifest.json
   

3. Go to Catalog, find "AI Upscaler", install v1.5.0.0
4. Restart Jellyfin
5. In Plugin Settings: Set AI Service URL to http://YOUR_SERVER_IP:5000

---

🚀 Features

• Docker Microservice: AI runs isolated in a container (no DLL conflicts!)
• Multiple AI Models: FSRCNN, ESPCN, LapSRN, EDSR (2x, 3x, 4x upscaling)
• Web UI: Manage models at http://YOUR_SERVER_IP:5000
• Hardware Detection: Automatic GPU/CPU detection
• Dashboard: Job monitoring in Jellyfin sidebar
• FFmpeg Integration: Automatic filter injection

---

⚙️ Configuration

After installation, find settings under Dashboard → Plugins → AI Upscaler Plugin.

Setting	Description
AI Service URL	URL to Docker container (e.g., http://nas:5000)
Enable Plugin	Global switch
Scaling Factor	2x, 3x, or 4x
Quality Level	low / medium / high

---

📋 Changelog

v1.5.0.1 (Hotfix)

• 🔧 Fixed #34: Plugin initialization error (HardwareBenchmarkService DI)
• 🔧 Fixed #33: Checksum mismatch during installation
• 🔷 Added #32: Intel GPU/iGPU support via OpenVINO (Dockerfile.intel)

v1.5.0.0 (TEST PHASE)

• 🐳 Docker Microservice Architecture: AI processing in separate container
• 📦 ~1.6 MB instead of 417 MB: No more native DLLs in plugin
• 🔧 OpenCV DNN Models: FSRCNN, ESPCN, LapSRN, EDSR from public sources
• 🌐 Web UI: Model management at http://localhost:5000
• ✅ Fixed version format: 4-part version for Jellyfin compatibility

v1.4.9.4

• Settings Page Fix
• Cross-Platform Support
• Complete DI Registration

---

🔧 Troubleshooting

Plugin shows "Not Supported"

1. Make sure you uninstalled old versions (1.4.9.x)
2. Delete old plugin folder from Jellyfin plugins directory
3. Restart Jellyfin
4. Install v1.5.0.0 fresh from repository

Plugin won't start

# Check Docker container
docker ps --filter name=jellyfin-ai-upscaler

# View logs
docker logs jellyfin-ai-upscaler

Upscaling not working

1. Check if Docker is running: curl http://YOUR_IP:5000/status
2. Check Plugin Settings: AI Service URL correct?
3. Check if model is loaded: http://YOUR_IP:5000 → Web UI

---

📖 Support

• Project Website
• GitHub Issues
• GitHub Wiki

---

📜 License

MIT License - See LICENSE for details.