CTUs

CTU stands for Coding Tree Unit, the fundamental processing unit in the High Efficiency Video Coding (HEVC, also known as H.265) standard. A CTU represents the top-level block that encodes a region of a video frame, and frames are divided into CTUs in raster order. The maximum CTU size is 64 by 64 pixels, allowing the encoder to choose appropriate block sizes for different content areas.

Within each CTU, the image data is partitioned into Coding Units (CUs) using a quadtree structure. CUs determine how the region is predicted, either intra-predicted from adjacent pixels or inter-predicted from other frames. Each CU can itself be partitioned further, down to smaller sizes, to adapt to local detail. For residuals, Transform Units (TUs) within CUs carry the transform and quantization steps, with transform sizes typically ranging from 4x4 up to 32x32.

Encoding and decoding revolve around selecting CTU, CU, PU (Prediction Units), and TU partitions to minimize a rate-distortion objective while preserving visual quality. The use of CTUs, rather than fixed-size blocks, enables more flexible and efficient coding, particularly for high-resolution video and content with varying texture.

Compared to previous standards such as H.264/AVC, HEVC's CTU-based structure reduces overhead and improves compression efficiency for similar quality. In practical terms, CTUs are the basic units processed by encoders and decoders, with the partitioning decisions largely governing encoding bitrates and latency in streaming, broadcasting, and storage applications.