Video latency is the time from camera capture to the moment the corresponding frame appears for the viewer. Measure that full glass-to-glass path; a network ping alone excludes encoding, packaging, buffering, decoding and display delay. This page owns the video-specific definition and causes. Use the general low-latency guide for network and application concepts, or the streaming architecture guide for protocol selection.
Knowing how to deal with latency is most important for those who deal with real-time video content: in the field of entertainment, e-sports, media and other OTT content providers.
What is video latency?
In the world of video, latency is a by-product of stream processing and delivery. It is called through, and often referred to using the term "glass-to-glass". For production ingest and monitoring, see continuous streaming. For playback integration and analytics, use video api.
Latency accounts for the period of time between the capture of what is happening on the camera and the moment it is displayed on the screen of the viewer.
In the context of live streaming, video latency depends on technical factors, including how streaming is set up on the platform.
A latency is considered undesirable since it prevents your viewers from receiving information directly in real time. However, sometimes it occurs unintentionally.
But there is also another type of “latency” – in order to synchronize streaming sources and add cinematic elements, producers need a small amount of time between capturing and transmitting video. The main thing is not to confuse them.
If you're designing a system for video content, whether it's a live streaming platform or a drone piloting program, your goal will be to minimize latency.
In general, the lower the latency, the better for your product and your users.
In case you do not address this short period of time, you will lose your end users eventually, as they will get a subpar experience.
For example, you watch football through an OTT streaming platform. And your neighbor is watching the same match on classic TV. He will already know who scored the goal while you’d be still waiting for it to happen, if the latency is not addressed properly.
What is the acceptable latency for video?
Latency is usually measured in units of time, such as seconds or milliseconds (ms).
The waiting time can vary from minutes to milliseconds.
A latency of 300 milliseconds or less is considered ultra-low.
If your product is an interactive application, you’ll need to get to 200 milliseconds and try to get as close to real time as possible.
There are 4 levels of unwanted latency (all indicators are approximate and may vary slightly in different sources):
1.Typical latency. The time interval is from 18 to 30 seconds. This is okay, if your broadcast is not interactive.
2. Broadcast latency can occur on both traditional TV and OTT platforms and ranges from 5 to 18 seconds. If you are broadcasting news or broadcasting from a football field or other sporting event, this is normal.
3. Ultra-low latency - ideal for live video conferencing. These are indicators below 300 milliseconds.
The peculiarity of low latency is that it does not have an absolute value. But everyone who publishes interactive content strives for it to be less than 1 second.
What causes video delay?
As part of the video processing, when temporary storage is required, buffering may occur to ensure the video reaches the user's screen without distortion.
An example of this is the translation of pixels captured by the camera into an image displayed on the display. The time it takes to convert a short piece of a video also depends on the frequency of the frames.
For example, take a one-frame delay in a 30fps video.
An HD 720p video frame has 720 horizontal lines, so a single line delay at 30 fps is 1/(30*720) = 0.046 ms latency. At 1080p at 30fps, the same single-line delay is less than 0.030ms.
Latency also affects video quality.
Due to adaptive streaming, platforms will degrade the picture quality to balance the speed and quality of playback.
Video latency is affected by three main parameters :
-
internet connection speed
-
video protocols
You can learn more about the encoder settings in the article : What is video encoding?
How to reduce latency if you are a gamer streamer
To prevent lags from affecting the gameplay, you can do a few simple steps:
1. Check your internet speed and bandwidth.
2. Make sure your hardware is set up correctly. To do this, find out the location and connection of the remote server, as well as diagnose the equipment.
3. To cut the latency down, you need to shorten the route that data packets travel from the source to you. For example, if your router is located far from the console, this also slightly affects the quality of the connection.
Make sure there are no walls, floors, furniture, or other physical objects between the computer and the router that could interfere with the wireless signal.
Few people know, but household appliances also affect the signal: wireless speakers, a microwave oven, a mobile phone, a charger, etc.
4. Close all background websites and programs.
5. Connect your device to the router with an Ethernet cable.
6. Play on a local server.
7. If all of the above did not help, replace the router and possibly upgrade your broadband package.
How else can you reduce video latency without sacrificing image quality?
Compare encoders and decoders using the format and settings you will actually run. A component's low-delay specification does not describe the whole path: capture, encoding, transport, decoding and display each add time. Measure glass-to-glass delay before and after changing one buffer or processing setting.
SRT uses retransmission requests to recover missing packets when the network, available bandwidth and configured latency window allow them to arrive in time. It cannot prevent the network from dropping packets. A shorter recovery window leaves less time for retries; the transport setting is only part of the end-to-end delay. See the SRT latency documentation for the protocol boundary.
For a contribution link, continue with setting SRT latency from RTT and a controlled path test. Keep the encoder and playback settings unchanged during that test so you can see what the transport adjustment changed.
How to find a balance between latency and bandwidth and video transmission quality?
Start with the delay the task can tolerate, then measure the complete path. An interactive conversation and a one-way event stream have different needs. Preserve enough transport and playback headroom for the network you measured, and check motion quality and audio alongside delay.
If the source already reaches its destination in a suitable format, an additional transcode is not automatically useful. Add processing when the receiver needs a different codec, resolution or bitrate, and measure the resulting delay again.
Quick answers
Is high or low latency better?
Lower latency is better when viewers or operators need to react quickly, such as live interaction, monitoring, or auctions. Higher latency can still be acceptable when buffering improves playback stability and real-time response is not required.
How do you get low latency?
Choose a protocol and player mode built for the target interaction, shorten encoder and player buffers, use a well-connected ingest path, and measure end-to-end delay. Smaller buffers trade resilience for faster response.
Is low latency always good?
Low latency is useful when someone must react to the live picture, but it is not automatically the best setting. End-to-end delay accumulates in capture, encoding, transport, receiver buffering, decoding, packaging, and the player buffer. Reducing one component does not remove the others.
Choose the delay target from the job: interaction, return monitoring, and time-sensitive operations may justify shorter buffers; a one-way event can accept more delay for steadier playback. Measure by showing a clock or changing visual at the source and comparing it with the viewer display, then record the protocol, encoder, receiver, and player settings. With SRT, the latency setting is a resilience buffer for network variation, not a promise of fixed ultra-low delay.
What low latency means in a video path
Low latency means keeping the end-to-end delay between an event at the source and its appearance at the viewer as small as the workflow can safely support. That delay is the sum of capture, encoding, transport, receiver buffering, decoding, packaging, and player buffering; it is not a single protocol number. A shorter buffer can improve responsiveness but leaves less margin for network variation.
Measure the whole path by showing a clock, timer, or deliberate visual change at the source and comparing it with the viewer display. Record the encoder, transport, receiver, packaging, and player settings with the result, because changing only one stage may not materially change the total. With SRT, the latency setting is a resilience buffer for variation on that transport leg; it is not a promise of a fixed ultra-low end-to-end delay.
Define video latency across the complete path
Set the acceptable source-to-viewer delay from the job, then agree on the viewer device and connection used for acceptance and the fallback if that target is missed. A one-way event can use a different acceptance criterion from a return-monitoring or interactive path.
Next steps
Use the low-latency workflow guide to separate contribution from viewer delivery. If you need to receive and route an SRT contribution, Callaba SRT Server explains that part of the path. Callaba runs in your selected infrastructure; it does not set the delay of an external platform or player.
