Pframes

P-frames, or predictive frames, are a type of video frame used in inter-frame compression. They encode only the changes from previously decoded frames rather than storing a complete image. To achieve this, the encoder analyzes the current frame against one or more reference frames and records motion vectors that describe how blocks in the current frame have moved relative to the reference, along with a residual image containing the difference between the predicted and actual pixels. The decoder uses the motion vectors to reconstruct the predicted blocks from the reference frame and then adds the residual to recover the original frame.

P-frames typically reference past frames, in contrast to B-frames, which may reference both past and future frames. They are positioned after an I-frame within a Group of Pictures (GOP) and contribute to lower bitrate by exploiting temporal redundancy. The use of P-frames reduces the amount of data needed to represent a video sequence, especially for scenes with motion or gradual changes.

Limitations and considerations include dependency on earlier frames, which means errors can propagate if a reference frame is corrupted or lost. This also adds a layer of decoding complexity. In modern codecs, P-frames can reference multiple previous frames to improve prediction accuracy, increasing efficiency at the expense of encoder/decoder complexity.

In common usage, a GOP starts with an I-frame and contains a mixture of P-frames (and sometimes B-frames), balancing random access, compression efficiency, and latency for a given application.