Blue Iris – Best Settings (4K + AI Optimized)
🔹 Overview
These settings are based on real-world deployment of multi-camera Blue Iris systems using 4K cameras, AI detection, and direct-to-disk recording.
The goal is:
- Stable performance
- Clean alerts (low false positives)
- Efficient storage usage
- Reliable long-term operation
🔹 Core Recording Settings
- Direct-to-disk: Enabled
- Codec: H.264 (preferred for stability)
- FPS: ~15 FPS
- Keyframe interval: Match FPS
Why:
- Reduces CPU usage
- Keeps playback smooth
- Avoids re-encoding overhead
🔹 Motion Detection (Critical)
- Use zones (A–H) to limit detection area
- Set Min Object Size to ignore noise
- Adjust Min Contrast to reduce false triggers
- Use Make Time ~1.0s
Why:
- Eliminates bugs, rain, and shadows
- Improves AI accuracy downstream
🔹 AI Settings (CodeProject / ONNX)
- Confirm alerts with AI: Enabled
- Minimum confidence: ~60%
- Primary detection: Person / Vehicle
- Use main stream images if available
Why:
- Filters false positives
- Uses higher-quality frames for detection
🔹 Alert Settings
- Trigger: New triggers only
- Add to alerts list: Database only
- Require multiple alerts: Optional for noisy areas
Why:
- Keeps alerts meaningful
- Avoids spam
🔹 Storage Strategy
Typical setup:
- New: 3–5 days
- Stored: Long-term (size-based, e.g. 2TB)
- Alerts: Separate folder (short retention)
Why:
- Fast recent access
- Efficient long-term storage
- Clean alert management
🔹 Performance Optimization
- Use hardware decode (Intel / GPU)
- Limit unnecessary re-encoding
- Disable unused features (audio, overlays if not needed)
🔹 Recommended Baseline
If you want a starting point:
- 15 FPS
- H.264
- Direct-to-disk
- AI confirmation ON
- Confidence ~60%
- Motion zones configured
🔹 Common Mistakes
- Using H.265 everywhere → causes lag
- No zones → constant false triggers
- AI confidence too low → noisy alerts
- Too many FPS → wasted resources
🔹 Final Thought
The best Blue Iris system is not the most aggressive —
it is the most stable and predictable over time.