Video streaming is the delivery of video data over a network in a way that allows playback to begin before the entire video has been downloaded and, in many systems, allows data to continue arriving while the viewer watches. Instead of requiring a complete video file to be transferred before playback, a streaming client receives a sequence of media segments or a continuous media stream and maintains a buffer from which the player decodes and displays video.
Streaming can occur over the internet, local networks or specialised communications systems. On-demand services, live television over the internet, video conferencing, gaming broadcasts and many social-media video systems use streaming techniques. The fundamental challenge is to deliver enough encoded media in time for playback despite variable network bandwidth, latency, packet loss, congestion and device limitations.
Modern internet video streaming commonly uses adaptive bitrate delivery. A service creates multiple encoded representations of the same content at different resolutions and bitrates, such as low-, medium- and high-quality versions. The client measures network conditions and device performance and requests suitable segments. If bandwidth falls, the player can switch to a lower-bitrate representation to reduce the likelihood of buffering. If conditions improve, it can request higher-quality segments. Common adaptive streaming approaches include HTTP Live Streaming and MPEG-DASH. These systems typically divide media into relatively short segments and provide a manifest describing the available representations. The player uses the manifest and current playback conditions to decide which segment to request next.
Video streaming requires a complete chain extending from content production to the viewer's display. The source video is captured or generated, edited and encoded into one or more representations. A streaming platform may transcode the source into multiple resolutions, frame rates, bitrates and codecs. Encoded media are then stored on servers and distributed through content delivery networks, or CDNs, which place copies or caches closer to users.
When a viewer starts playback, an application retrieves the manifest and requests media segments over network protocols, commonly using HTTP-based delivery. The client buffers incoming segments, decodes them using hardware or software, synchronises audio and video and presents frames at the correct times. Failures at any stage can affect the viewing experience.
Buffering is a key part of streaming playback. The player normally maintains a quantity of media data ahead of the current playback position so that short-term network fluctuations do not immediately interrupt playback. A larger buffer can provide more resilience against temporary bandwidth reductions but can increase startup delay and make live interaction less immediate. Low-latency streaming systems use shorter segments, smaller buffers and specialised protocols or techniques to reduce the delay between capture and display.
Live sports, auctions, interactive broadcasts and remote events may require lower latency than ordinary on-demand films. Video conferencing imposes even stricter latency requirements because participants must interact in near real time, so its architecture differs from conventional on-demand streaming even though both involve networked video.
Streaming quality depends on bitrate, resolution, frame rate, codec efficiency, network capacity and device capabilities. A 4K stream generally contains more image information than a 1080p stream and therefore requires more processing and typically more network capacity for comparable quality. Efficient codecs such as H.264, H.265 and AV1 can reduce the bitrate needed for a given visual quality, although encoding and decoding complexity differ. The service may choose different codec representations for different devices. A modern television might receive hardware-decodable HEVC or AV1, while an older device might receive H.264. Adaptive streaming can also respond to screen size, available bandwidth and playback conditions rather than delivering a fixed representation to every viewer.
Streaming is not synonymous with downloading. In ordinary progressive download, a client may retrieve a conventional media file and begin playback before the entire file has arrived, while streaming architectures typically use segmented or continuous delivery and are designed around ongoing playback. The distinction can blur because modern HTTP-based streaming uses ordinary web requests to retrieve media segments. Streaming also does not necessarily mean that no data are stored locally. A player buffers data in memory or temporary storage, and some services permit deliberate offline downloads. The key distinction is the delivery and playback model rather than whether bits temporarily exist on the user's device.
Video streaming introduces technical issues involving reliability, congestion, caching, security and content protection. Encryption can protect media during transport, while digital rights management systems may control access to protected content. Content delivery networks reduce latency and origin-server load by serving cached media from geographically distributed infrastructure. Network protocols handle packet delivery, congestion control and retransmission, while application-layer streaming logic determines representation selection and buffering. If network throughput falls below the video's sustained bitrate, the buffer eventually empties unless the player switches to a lower-quality representation. This is why adaptive bitrate streaming can preserve uninterrupted playback at the cost of temporarily reducing resolution or visual quality.
Video streaming has transformed how television, films, live events, education, music video, gaming and user-generated content are distributed. Instead of requiring users to obtain a physical medium or a complete media file before viewing, network infrastructure can deliver content on demand to enormous numbers of devices. The technology depends on compression because raw high-resolution video would require impractical network capacity for most consumer connections. It also depends on scalable storage, encoding infrastructure, CDNs, media players and reliable networking.
Modern streaming systems increasingly combine multiple codecs, resolutions, dynamic adaptation and low-latency techniques. The resulting experience is therefore the product of an integrated system in which video compression, network engineering, software playback and content-distribution infrastructure work together to deliver moving images continuously to the viewer.
