What Is Simulcast? How It Works in WebRTC Video Calls
One person on fiber, another on hotel Wi-Fi, same group call. Simulcast is how each of them gets video that fits their connection without anyone re-encoding it.
Simulcast vs SVC (scalable video coding)
|
Simulcast |
SVC | |
|---|---|---|
|
What the sender uploads |
Two or three independent encodings of the same video |
One layered stream; higher layers build on lower ones |
|
How the server adapts |
Forwards a different encoding to each receiver |
Drops the layers a receiver can't use |
|
Extra upload cost |
Higher; every layer is a full encode |
Lower; layers share data |
|
Sender CPU |
Several encoders running at once |
One encoder, more complex |
|
Codec support |
VP8, H.264, VP9 and AV1 in modern browsers |
Mainly VP9 and AV1 |
|
Switching between qualities |
Needs a keyframe on the new encoding |
Smoother; drop or add layers |
|
Best for |
Broad browser and device support today |
Saving upload bandwidth where codecs allow it |
What is simulcast?
In a group video call, participants rarely have the same connection. One is on office fiber, one is on a train. If the sender uploads a single 720p stream, the server has two bad choices: send 720p to everyone and stall the weak connections, or ask the sender to drop quality and make everyone suffer.
Simulcast avoids that. The sender's browser encodes the same camera feed two or three times at different resolutions and bitrates, and sends all of them. Each version is called a layer or encoding, and each gets an ID (a RID) so the server can tell them apart. A media server, almost always an SFU, then forwards one layer to each receiver.
Simulcast also has an older meaning in broadcasting: airing the same program on two channels at once, such as a concert on TV and radio. This page covers the WebRTC meaning.
How simulcast works
1. The sender offers several encodings
When the call is set up, the client asks the browser for multiple encodings of one video track. A common setup is full, half and quarter resolution, for example 720p, 360p and 180p. The SDP offer lists them with RIDs, as defined in RFC 8853.
2. The browser encodes every layer
The browser runs one encoder per layer and sends all of them to the server. The low layers are small, so they add less upload than the top layer alone, but they do add some.
3. The SFU picks a layer per receiver
For each receiver, the SFU estimates available bandwidth and checks how the video is shown. The active speaker in a large tile gets the high layer. A thumbnail or a struggling connection gets the low one. The SFU only forwards packets; it never decodes the video.
4. The SFU switches as conditions change
When a receiver's bandwidth drops, the SFU switches it to a lower layer. A clean switch needs a keyframe on the new layer, so the SFU requests one from the sender.
Simulcast example in JavaScript
In the browser you turn simulcast on with sendEncodings when you add the video track. The SFU must support simulcast for this to help.
const stream = await navigator.mediaDevices.getUserMedia({ video: { width: 1280, height: 720 } });
const [track] = stream.getVideoTracks();
pc.addTransceiver(track, {
direction: 'sendonly',
sendEncodings: [
{ rid: 'q', scaleResolutionDownBy: 4, maxBitrate: 150_000 }, // 320x180
{ rid: 'h', scaleResolutionDownBy: 2, maxBitrate: 500_000 }, // 640x360
{ rid: 'f', maxBitrate: 1_500_000 } // 1280x720
]
});
The bitrates are starting points to tune, and browsers may switch off the top layer on their own when CPU or upload bandwidth runs short.
Simulcast vs SVC
Scalable video coding (SVC) reaches the same goal in a different way. The sender produces one stream built in layers: a base layer, plus layers that add resolution or frame rate on top. The SFU drops the upper layers for receivers who can't use them.
SVC saves upload bandwidth because layers share data, and switching is smoother. The catch is codec support: in browsers, SVC is practical with VP9 and AV1, while simulcast also works with VP8 and H.264. Many SFUs support both and pick per call.
Simulcast is also different from adaptive bitrate streaming, which serves pre-encoded video segments over HTTP for on-demand and live streams. Simulcast works in real time, with sub-second latency.
Rule of thumb: use simulcast when your SFU and your users' browsers support it and your calls have three or more people. For one-to-one calls, adaptive bitrate on a single stream does the job.
Where simulcast is used
Group video calls: the main use. Mixed networks stop dragging everyone to the lowest quality.
Speaker and gallery layouts: the active speaker gets the high layer, everyone in the grid gets the low one.
Webinars and large meetings: one presenter, hundreds of viewers on very different connections.
In-app calling: a telehealth or marketplace app where a doctor on clinic Wi-Fi and a patient on mobile data join the same call.
Limits and pitfalls
It costs the sender more. Extra upload and extra encoding CPU. On a weak uplink or an old laptop, the browser may drop layers.
It does nothing without an SFU. In a one-to-one or mesh call there is only one receiver per stream, so there is nothing to choose between.
Codec support varies. Hardware H.264 encoders on some devices handle simulcast differently. Test the browsers and devices your users actually have.
Layer switches can be visible. Each switch waits on a keyframe, so a receiver may see a short freeze or a jump in sharpness.
Screen sharing is a special case. Text needs full resolution, so many apps send screen share as a single high-resolution, low-frame-rate stream.
Group video in CometChat
Simulcast is one of the things you have to get right if you run your own media servers. If you add calling with an SDK, the provider handles the media path.
Group calls without running media servers
CometChat Calls is built on WebRTC. Its docs say group calls use a Selective Forwarding Unit (SFU), with media routed through CometChat's servers, and that the SDK adjusts bitrate and quality to network conditions. Recording happens on CometChat's media servers, not on participants' devices. You get 1:1 and group voice and video, screen sharing and pre-built call UI for JavaScript, iOS, Android, React Native and Flutter.
Frequently asked questions
What is simulcast in WebRTC?
A sender encodes the same video at several qualities and uploads all of them to an SFU. The SFU forwards each receiver the quality that fits its connection and screen.
Does simulcast use more bandwidth?
On the sender's upload, yes, because it sends two or three encodings. The lower layers are small, so the extra is a fraction of the top layer. Receivers often use less bandwidth, since many get a lower layer.
Do I need simulcast for one-to-one calls?
No. With one receiver there is nothing to choose between. Adaptive bitrate on a single stream handles changing network conditions.
What is the difference between simulcast and SVC?
Simulcast sends several independent encodings. SVC sends one layered stream where the server drops layers. SVC uses less upload but needs codecs like VP9 or AV1.
What does simulcast mean in broadcasting?
Airing the same program on more than one channel or medium at the same time, such as TV and radio. It is unrelated to the WebRTC technique apart from the idea of one source sent several ways.