InterFrames

Interframes are a category of video frames encoded using temporal prediction from other frames, rather than being encoded independently like intra-frames (I-frames). They exploit similarities between successive frames to reduce the amount of data that must be stored or transmitted.

The primary forms of interframes are P-frames and B-frames. P-frames are predicted from one or more previously decoded frames (I-frames or P-frames) using motion compensation. The encoder estimates motion vectors that map blocks in the current frame to corresponding blocks in the reference frame and encodes only the residual differences. B-frames are bidirectionally predicted from past and future reference frames, often achieving higher compression by leveraging information from frames on both sides of the current frame. B-frames typically require decoding of future frames, which can affect decoding order and latency.

Interframes are organized within a group of pictures (GOP). A GOP usually starts with an I-frame, which serves as a self-contained reference, followed by several P- and/or B-frames. Although B-frames can offer the greatest compression, they introduce additional complexity and buffering requirements because they depend on both preceding and following frames for reconstruction. Furthermore, error in a reference frame can propagate to dependent interframes, impacting quality if data is lost or corrupted.

Applications and trade-offs: interframes are central to most modern video codecs (such as MPEG-2, H.264/AVC, HEVC/H.265, and AV1) and drive substantial bitrate reduction compared with truly intra-coded sequences. The trade-offs include increased latency, higher computational complexity, and potential vulnerability to error propagation, balanced against improved compression efficiency and bandwidth savings.