
A live video may use several streaming protocols before it reaches a viewer. The connection carrying a camera feed into a studio is not necessarily the one delivering the finished show to a phone. This guide explains where the main protocols fit, shows three practical workflows, and identifies the settings you need to connect each stage.
Streaming protocols define how software and devices exchange live or on-demand media, or control the sessions used to deliver it. Different protocols solve different parts of the video journey.
A protocol is different from a codec or file format. H.264 compresses video, MP4 stores media in a container, and protocols govern its transfer or playback session. Supporting the same protocol does not automatically mean two devices support the same video and audio formats.

Start with what needs to connect to what. The table shows common uses.
| Protocol | Common Connection | What It Does There |
|---|---|---|
| RTMP / RTMPS | Encoder to platform ingest server | Uploads an encoded live stream |
| SRT | Field encoder to studio or cloud receiver | Transports media with packet recovery and configurable buffering |
| RTSP | IP camera or media server to a compatible client | Controls the media session; RTP commonly carries the media |
| HLS | HTTP server or CDN to a compatible player | Delivers playlists and media segments, supporting adaptive playback |
| MPEG-DASH | HTTP server or CDN to a compatible player | Delivers a manifest and media segments, supporting adaptive playback |
| WebRTC | Browser or app to another endpoint or media server | Supports real-time audio and video interaction |
RTMP and its TLS-protected form, RTMPS, commonly carry a finished stream from an encoder to a platform. SRT is useful when a production feed must travel between compatible systems over a network with packet loss or varying delivery times.
SRT's recovery needs time and bandwidth. It cannot repair every network problem, and its buffering must balance resilience against delay.
HLS and DASH support segmented delivery through web servers and CDNs. When multiple quality versions are available, a compatible player can switch between them as conditions change.
WebRTC is a collection of technologies and protocols for real-time communication. It is relevant when participants need to respond to each other, such as a remote guest speaking with a host. It can also support interactive viewing through suitable infrastructure.
Latency is not a fixed protocol specification. Encoding, network conditions, buffering, server processing, and player settings all contribute to the delay viewers experience.
These examples show how protocols can work together. They are possible architectures, not mandatory configurations for every platform.
A presenter uses production software to combine a camera, microphone, and slides. The software sends the finished program to a streaming platform, which prepares it for viewer playback.

Example path: Encoder → RTMPS → streaming platform → HLS or DASH → viewers.
The creator configures the platform's ingest connection. The platform manages its playback formats, so viewers do not need to receive RTMPS directly. Other delivery methods may be used depending on the service and viewing mode.
An event team sends a live feed from a venue to a studio. The studio adds commentary, graphics, and other sources before publishing the finished program.

Example path: Field encoder → SRT → studio receiver and production system → RTMP → streaming platform.
SRT handles the remote contribution link. RTMP handles the later platform upload. The receiving and production functions may run on separate devices or within one compatible system.
A production operator wants to include a network camera in a public broadcast. Compatible software accesses the camera feed, then prepares an output the destination accepts.

Example path: IP camera → RTSP-controlled session with RTP media → encoder or gateway → RTMP → streaming platform.
The gateway bridges the camera-side workflow and the platform-side workflow. Depending on codec compatibility, it may repackage the media or decode and re-encode it. An RTSP address is not interchangeable with a platform's ingest URL.
For a closer comparison of contribution and camera-access options, see RTMP vs SRT vs RTSP.
Once you identify the relevant workflow, configure one connection at a time. Start with the requirements of the device or service receiving that connection.
If you manage your own viewer delivery, there is an additional server-side task: package HLS or DASH and use a compatible player, or deploy WebRTC infrastructure. Selecting an encoder output does not create those delivery systems automatically.
When a presenter moves between a lectern and a demonstration area, keeping the shot framed becomes a separate task from sending the stream. OBSBOT Tail 2 suits productions that need remotely controlled framing and tracking without a dedicated operator beside the camera.
RTP carries real-time media packets and timing information. RTSP manages a media session, including operations such as setup and playback. They often work together rather than competing for the same job.
Neither is always better. TCP provides ordered, reliable delivery, but recovering missing data can delay later data. UDP does not provide that recovery itself; protocols built on it can add their own recovery, timing, and congestion behavior. Choose the supported application protocol for the workflow, not the transport name alone.
Ordinary HTML video playback in modern browsers does not generally accept an RTSP URL. Use compatible software, or a media gateway that turns the source into a browser-supported delivery method.
Not necessarily. RTMPS protects its connection using TLS, while SRT encryption must be configured appropriately. Other links, recordings, access controls, and viewer delivery need their own protection. Securing one connection does not secure the whole workflow.
Choose protocols by connection: accessing a source, transporting a production feed, uploading a program, or delivering playback. Then configure the sender to match the receiver and test the complete path. A working broadcast may use several protocols, with each handling a different job.



