Mux video glossary

Mux Video is an API-first platform, powered by data and designed by video experts to make beautiful video possible for every development team.

Check out Mux video

Back to Glossary

GOP (Group of Pictures)

In video compression, a Group of Pictures (GOP) is a fundamental unit that organizes and arranges video frames for efficient encoding. Rather than encoding each video frame sequentially, multiple frames are grouped into a GOP, allowing for more effective compression and decompression. Understanding GOP structure is crucial for video editing, streaming, and optimizing video compression in codecs such as H.264/AVC, HEVC, and AV1.

Structure of a GOP

A GOP consists of three types of frames: I frames, P frames, and B frames, each serving a unique purpose in the video compression process.

I Frames (Intra-coded frames)

An I frame is a complete image that can be encoded and decoded independently. It serves as a reference for the surrounding frames and is essential for the accurate display of the video. Each I frame is divided into 8x8 pixel blocks, grouped into 16x16 pixel macroblocks, and then compressed using various techniques. I frames are generated as needed, especially when there is a significant change in the image content, typically about twice per second in a standard video stream.

P Frames (Predicted frames)

P frames contain data describing the differences (motion vectors) between themselves and the previous frame, often an I frame. These frames significantly reduce file size by capitalizing on temporal redundancy, where consecutive frames share similar content. P frames cannot stand alone as they rely on previous frames for decoding. If there is a substantial change in the image, new pixel blocks are created for the changed portion, using the same encoding method as I frames.

B Frames (Bi-predictive frames)

B frames achieve the highest compression ratio by utilizing information from both preceding and following frames within the GOP. They contain motion vectors and pixel blocks, similar to P frames, but can reference frames both before and after them. This bidirectional prediction allows for greater compression but also means that B frames are highly dependent on surrounding frames for accurate decoding.