How Video Compression Works

The principles of video compression are based on reducing the amount of data while maintaining the highest possible image quality. This is done in two main ways: spatial and temporal reduction of information.

Basic Principles

Spatial data reduction is related to analyzing each individual frame. Inside a frame, there are many repetitive or similar elements — for example, large solid-colored areas or smooth color transitions. The codec uses mathematical methods (most often discrete cosine transform or, in modern variants, wavelet transform) to represent such areas more compactly.

Essentially, this is similar to JPEG compression for images: we remove redundant details that the human eye barely notices, thereby reducing the frame size.

Temporal data reduction works with a sequence of frames. In video, neighboring frames are often very similar to each other: the background remains unchanged, and only part of the objects moves. The codec doesn't store each frame entirely but describes changes relative to the previous one.
This allows for significantly reducing the amount of information, as instead of a complete frame, it's enough to record only what has changed.

This is where two key types of frames emerge:

• Intra-frame (I-frame) — a frame encoded completely, without relying on others. It serves as a reference point and can be played separately.
• Inter-frame (P-frame, B-frame) — frames encoded based on others. P-frames use data from previous frames, while B-frames use both previous and subsequent frames simultaneously. They contain only information about differences, making them much lighter in size.

Thus, the combination of spatial and temporal data reduction allows the codec to effectively balance quality and compression ratio: inside frames, excessive detail is removed, and between frames, only changes are preserved.

The Difference Between Compressed and Uncompressed Video

Data compression is divided into two major approaches — lossy and lossless, and they are applied in different situations depending on what's more important: quality or space savings.

Lossless Compression

Preserves all source information. After decoding, the data completely matches the original, without the slightest changes. This method is used where accuracy is critical: for example, in text documents, program code, databases, or archive formats like ZIP and RAR.

In video and audio, lossless compression also occurs — for example, FLAC format for music or Apple ProRes for video — but they take up much more space and require more resources for storage and transmission.

The limitation here is obvious: files remain large, so lossless compression is applied only when quality is paramount.

Lossy Compression

Lossy compression works differently: it removes part of the information considered insignificant for human perception.
In audio, these might be sounds beyond the audible range; in images — subtle color nuances; in video — details that the eye barely distinguishes.
This approach allows for significantly reducing file sizes, but with repeated re-compression or excessive compression, quality noticeably decreases.

Lossy compression is used everywhere: H.264 or HEVC codecs are the most common.
The limitation here is that restoring the original is impossible — part of the data is lost forever.

Ultimately, the choice between these methods depends on the task: if you need to archive master recordings or store legally important documents, use lossless compression; if the priority is compactness and convenience of transmission, use lossy compression.

Efficiency of Modern Codecs

Modern codecs evolve from H.264 to more efficient H.265, VP9, AV1, and VVC. Each new step provides better quality at lower bitrate but requires more computational resources and support from devices and platforms.

Codecs

Compression Comparison

Full HD movie (two hours)

Short 4K clip (ten minutes)

Online Full-HD streaming (one hour)

The Role of Format in Video Compression

The difference in compression results from the container format is practically non-existent. The container is just a "wrapper" that combines video stream, audio stream, subtitles, and service data inside itself. It doesn't handle the compression process itself but only determines how these elements will be stored and synchronized.

Compression and video quality are determined precisely by the codec, which works with frames and sequences, reducing the amount of data. The container only "packages" the codec's output. For example, the same video stream encoded with H.265 can be saved in MP4, MKV, or MOV — and the quality and compression ratio will be identical because they depend on the codec, not the container.

The difference may only manifest in usability:

• some containers support more features (e.g., MKV easily stores multiple subtitle and audio tracks)
• others are better compatible with devices and programs (MP4 plays almost everywhere)

In other words, the container affects compatibility and functionality but not compression efficiency.