| Runs on | Windows, macOS, and Linux, as a desktop app. | Linux in Docker, including the Home Assistant add-on. |
| Price | Free for two cameras. €29 + VAT a year or €99 + VAT once for unlimited cameras. | Free. Optional Frigate+ detection models cost $50 a year. |
| Recording | H.264 and H.265 copied to disk exactly as the camera sends them. Continuous, or on motion with pre-roll and post-roll. | Camera stream written without re-encoding. Continuous, motion, or around tracked objects. |
| Motion detection | Uses each camera’s own ONVIF motion events, so the recorder never analyses video. | Analyses a decoded detection stream on the server for every camera. |
| Object detection | Records on the person and vehicle events of cameras that report them. No AI runs on the recorder. | Built-in real-time detection, its main strength. A Coral, GPU, or NPU accelerator is recommended. |
| Server load | About 0.1% of one CPU core and 19 MB of memory in our measurement. Video is only decoded for snapshots or on request. | Decodes the detection stream of every camera. Hardware decoding is recommended. |
| Retention | By age and by a hard storage cap across all cameras, plus a free-disk floor. Shows the days of footage the cap allows. | By days, per recording type. Oldest footage is deleted early only when about an hour of space remains. |
| Setup | Install and add cameras in the browser. Stream probing and an ONVIF port test check each camera. | Docker plus a YAML configuration file, editable from the web interface. |
| Phones and remote access | With a licence, any browser on your network, behind a password. Away from home, use your own VPN. No native app yet. | Web app you can add to your home screen. Remote access through your own VPN, proxy, or Home Assistant. |
| Integrations | Local HTTP API with an MJPEG stream and a still image for every camera, ready for Home Assistant and other home automation. | API, MQTT, and an official Home Assistant integration. |