How to understand video encoding 444, 422, 420?

Understanding video encoding formats like 4:4:4, 4:2:2, and 4:2:0 requires focusing on chroma subsampling, a core compression technique that reduces file size by selectively discarding color information while prioritizing luminance (brightness) detail. The notation itself, often confusing, refers to the sampling pattern of chroma (color) components relative to luma in a conceptual block of pixels. In a 4:4:4 scheme, there is no chroma subsampling; for every block of four luminance samples, there are four corresponding samples for each of the two color difference channels (typically Cb and Cr). This preserves full color resolution, making it critical for tasks requiring pristine color fidelity, such as high-end visual effects compositing, green screen keying, and mastering for cinema or archival. Formats like ProRes 4444 and some high-bitrate RGB codecs operate this way, but they come with a significant data rate penalty, often requiring double the bandwidth of subsampled equivalents.

The widely used professional standard is 4:2:2 subsampling, where chroma resolution is halved horizontally but maintained fully vertically. In our conceptual block, for every four luma samples, there are two chroma samples for each color channel. This strikes a practical balance, dramatically reducing data rates—often by about one-third compared to 4:4:4—while retaining excellent color detail for most broadcast, editing, and acquisition workflows. It is the backbone of major professional codecs like ProRes HQ, DNxHR, and the intra-frame compression used in cameras like the ARRI Alexa. The visual compromise is minimal for most content, though very fine, sharp color transitions (like between colored text and a contrasting background) can exhibit mild color bleeding or fringing that would be absent in 4:4:4.

The most common format for consumer delivery and many acquisition contexts is 4:2:0 subsampling. Here, chroma resolution is halved in both the horizontal and vertical dimensions. For every 2x2 block of four luma pixels, only one set of chroma samples is shared for the entire block. This offers the greatest data efficiency, reducing the chroma channel's data to one-quarter of its original size, which is why it is universal for Blu-ray, streaming (H.264, HEVC, AV1), and most consumer cameras. The trade-off is a potential loss of fine color detail, which can manifest as color smearing or artifacts in areas of high chroma spatial frequency, such as intricate patterns on clothing. The human visual system's relative insensitivity to high-frequency color changes makes this an acceptable compromise for final distribution, but 4:2:0 is less ideal for multi-generation editing or heavy color grading, as repeated processing can amplify these artifacts.

The choice among these schemes is therefore a direct engineering trade-off between bandwidth/storage and color fidelity, dictated by the specific stage of the video pipeline. Acquisition and post-production benefit from 4:4:4 or 4:2:2 to preserve a high-quality master, while distribution almost universally defaults to 4:2:0 for efficiency. Understanding this hierarchy is essential for selecting appropriate codecs, anticipating potential quality limitations in a workflow, and diagnosing visual artifacts that stem not from the codec's bitrate but from its fundamental color sampling structure.