What Is RTSP? How IP Camera Streaming Works

What RTSP does

Think of RTSP as the session-control conversation between a camera and a viewer or NVR. The client asks what media is available, negotiates how each track will be transported, starts delivery, keeps the session alive, and eventually closes it. The camera's H.264 or H.265 video remains separately encoded media.

Many deployed IP cameras still implement RTSP 1.0 from RFC 2326. RFC 7826 defines RTSP 2.0 as a replacement that is not backward compatible, so a client must support the version a camera actually exposes. In either version, “supports RTSP” is only the beginning of a compatibility check: authentication, transport, codec profile, timestamps, stream path, and recording behavior can still differ.

How an RTSP camera session starts

1. The client connects to the host and configured RTSP port.
2. It requests a media description for the selected camera path.
3. The camera describes one or more video and audio tracks.
4. The client and camera set up transport for those tracks.
5. The client starts playback and maintains the session until it is stopped.

The exact request sequence can vary by RTSP version and implementation. A successful network connection proves only that one endpoint answered; it does not prove that the path, credentials, media tracks, or codec are usable.

RTSP, RTP, codecs, ONVIF, and an NVR are different layers

Read RTSP vs ONVIF for the protocol boundary, or what a software NVR does for the recording layer.

RTSP URLs, paths, and port 554

A typical address has a scheme, host, optional port, and model-specific path: rtsp://CAMERA_IP:554/vendor-path. When an rtsp URI omits its port, the RTSP 2.0 specification says to use port 554. Cameras can still be configured to listen elsewhere, and vendors define their own channel and main/sub-stream paths.

Use the RTSP URL format list for documented brand patterns, the RTSP URL discovery workflow for an exact model, and the RTSP port 554 guide for TCP, UDP, alternate ports, firewalls, and safe testing.

Where a viewer and an NVR fit

An RTSP viewer opens the session and displays current video. NVR software must keep that session stable, write media segments, index timestamps, enforce retention, recover after interruptions, and play or export recorded time. Live video working for a minute does not validate the recording path.

Before relying on a camera, run the Mac RTSP stream test: inspect the media, keep it open, make a recording, play the timeline, interrupt the network, and confirm reconnect behavior.

Use RTSP only on networks you control

RTSP URLs can expose camera addresses, paths, and credentials through logs, screenshots, notes, and shell history. Prefer separate protected credential fields, create a camera-local account with limited permissions, and use only cameras and streams you are authorized to access.

Do not publish port 554 or a camera administration service directly to the internet. For remote access, use a maintained VPN or another authenticated private path and restrict it to the devices that need access.