Complete guide to optimizing AI upscaling performance

{
  "ai_method": "DLSS 3.0",
  "scale_factor": 4.0,
  "quality_preset": "ultra",
  "hdr_enhancement": true,
  "frame_interpolation": true,
  "thermal_protection": true
}

{
  "ai_method": "DLSS 2.4",
  "scale_factor": 2.5,
  "quality_preset": "high",
  "hdr_enhancement": false,
  "frame_interpolation": false,
  "adaptive_quality": true
}

{
  "ai_method": "FSR 2.1",
  "scale_factor": 1.5,
  "quality_preset": "balanced",
  "performance_monitoring": true,
  "thermal_protection": true
}

Anime/Animation:

• Use Waifu2x-cunet for best results
• Higher saturation values (1.2-1.4)
• Sharp enhancement enabled
• Scale factors up to 4x work well

Live Action Movies:

• DLSS 3.0 or Real-ESRGAN
• Conservative sharpness (0.3-0.5)
• Natural color enhancement
• 2-2.5x scaling for best quality

Documentaries/Text:

• Conservative enhancement mode
• Focus on text clarity
• Lower scale factors (1.5-2x)
• Minimal color adjustments

NVIDIA RTX Cards:

• Enable DLSS when available
• Use RTX HDR for compatible content
• Enable hardware decode acceleration
• Monitor thermal throttling

AMD Radeon Cards:

• FidelityFX Super Resolution preferred
• Enable AMD hardware decode
• Use conservative thermal limits
• Avoid frame interpolation on older cards

Intel Arc Cards:

• XeSS method recommended
• Enable Intel hardware acceleration
• Monitor driver compatibility
• Use balanced settings for stability

The plugin provides real-time performance monitoring:

{
  "gpu_usage": 75,        // GPU utilization %
  "vram_usage": 6.2,      // VRAM usage in GB
  "temperature": 68,      // GPU temperature °C
  "fps": 55,             // Current FPS
  "frame_time": 18.2,    // Frame time in ms
  "upscale_time": 4.5,   // AI processing time in ms
  "quality_score": 8.7   // Estimated quality (1-10)
}

• 🟢 Optimal: < 70°C, > 90% target FPS
• 🟡 Warning: 70-80°C, 70-90% target FPS
• 🔴 Critical: > 80°C, < 70% target FPS

1. Reduce Scale Factor

   4.0x → 2.5x → 2.0x → 1.5x
   

2. Change AI Method

   DLSS 3.0 → DLSS 2.4 → FSR 3.0 → FSR 2.1
   

3. Adjust Quality Preset

   Ultra → High → Balanced → Performance
   

1. Enable Thermal Protection

   {
     "thermal_protection": true,
     "thermal_limit": 75,
     "thermal_action": "reduce_quality"
   }

2. Reduce Processing Load

   {
     "frame_interpolation": false,
     "hdr_enhancement": false,
     "real_time_processing": false
   }

1. Monitor VRAM Usage

   // Check available VRAM
   const vramInfo = upscaler.getVRAMInfo();
   if (vramInfo.available < 2.0) {
     // Reduce settings
   }

2. Optimize Buffer Sizes

   {
     "buffer_size": "auto",
     "max_concurrent_frames": 2,
     "memory_limit": "80%"
   }

{
  "priority": "latency",
  "buffer_frames": 1,
  "predict_frames": false,
  "immediate_processing": true,
  "quality_preset": "performance"
}

{
  "priority": "quality",
  "buffer_frames": 3,
  "predict_frames": true,
  "batch_processing": true,
  "quality_preset": "ultra"
}

// Create custom profile
const customProfile = {
  name: "Home Theater 4K",
  settings: {
    ai_method: "Real-ESRGAN",
    scale_factor: 2.0,
    quality_preset: "high",
    hdr_enhancement: true,
    color_space: "rec2020",
    bit_depth: 10,
    thermal_limit: 70
  },
  conditions: {
    resolution: ">=1080p",
    gpu_memory: ">=8GB",
    content_type: "movie"
  }
};

// Enable adaptive quality
{
  "adaptive_quality": {
    "enabled": true,
    "target_fps": 60,
    "min_quality": 0.5,
    "max_quality": 1.0,
    "adjustment_speed": "medium",
    "monitor_interval": 1000
  }
}

# Run performance test
curl -O https://github.com/Kuschel-code/JellyfinUpscalerPlugin/raw/main/tools/performance-test.js
node performance-test.js --duration=60 --method=DLSS --scale=2.0

1. Visual Quality Tests

   • A/B comparison with original
   • Artifact detection
   • Color accuracy validation
   • Sharpness evaluation

2. Objective Metrics

   • PSNR (Peak Signal-to-Noise Ratio)
   • SSIM (Structural Similarity Index)
   • VMAF (Video Multimethod Assessment Fusion)
   • LPIPS (Learned Perceptual Image Patch Similarity)

Access performance history at:

http://your-jellyfin-server/plugins/upscaler/performance

Available Metrics:

• Average FPS over time
• Quality improvements tracked
• Thermal performance history
• GPU utilization trends
• VRAM usage patterns

// Export performance logs
upscaler.exportPerformanceData({
  format: 'csv',
  timeRange: '7days',
  includeSystemInfo: true
});

🎉 Ready to optimize your setup? Check our Configuration Guide for detailed settings!