User:Kmenzel/Helvetica

Video Compression is a process which reduces the amount of information required to store or transmits video. Software or hardware which performs the compression, and the decompression, is known as a compressor-decompressor (or codec). The effectiveness of codecs is given by how well it can preserve image fidelity at given bitrates. The complexity of a codec determines the hardware requirements to encode or decode the information in a reasonable amount of time, thus the complexity of a codec is important in determining how appropriate it is for a given situation.

Video compression is used to deliver digital video through a variety of methods: digital television systems, video communication services, streaming internet video services, and physical storage formats (for instance DVDs, or the tapes used by digital camcorders).

The Moving Picture Experts Group (or MPEG) has defined several popular standards with which to encode video and audio information.

Compression Process

Video compression can be either lossless, or lossy. Lossless compression is incredibly rare – it is typically used only in archival settings, because storage requires a significant amount of space. Lossy compression is more common because the smaller bitrates require less storage space, and less bandwidth to transmit, which is more economically viable for most businesses.

Lossless Compression

Lossless compression techniques for video focus on frame-per-frame image compression techniques, especially wavelet compression. Wavelet compression utilizes multiresolutional analysis to present a given frame as a series of real coefficients. The coefficients are typically quite close to zero for most images, which allows images to be represented by a small number of large coefficients. An implementation of this process can be designed using the JPEG-2000 schema.

Lossy Compression

Lossy video compression is achieved in two ways – Intraframe, and Interframe. Intraframe compression reduces the overall amount of data transmitted by reducing the amount of information contained within individual frames of video. This is typically achieved by an implementation of intraframe subsampling.

Interframe compression is achieved by examining the temporal redundancy of visual information. Most video contains very similar information from one frame to the next.

Intraframe Compression

The human eye is less sensitive to movement in color than it is to movement in brightness. Intraframe subsampling takes advantage of this by reducing the resolution of the color information. This can be described by the following format of ratio – Y:Cr:Cb. The Y value is the luminance, and Cr and Cb represent the red and blue chroma values which make up the color information. Both the Cr and Cb values are relative to the Y value. 4:4:4 would represent no chroma subsampling, whereas 4:2:2 would represent that both the Cr and Cb values are sampled at half the horizontal resolution. Ratios with a third value of zero are a special case – 0 indicates that the subsampling is the same for both Cr and Cb values, and is both horizontal and vertical. Thus 4:2:0 would indicate that Cr and Cb values are sampled at half the resolution of the luminance both horizontally and vertically (effectively a quarter of the resolution). This theoretically places the chroma pixel between the rows and columns of luminance information.

• 4:2:2 is commonly used in high-end digital video formats and interfaces, such as HDVPRO 50/60 or Digital Betacam.
• 4:1:1 is used in low-end and consumer applications, such as DVPRO, or NTSC DV.
• 4:2:0 is used in MPEG compression, PAL DV, JPEG compression, amongst others. .

Interframe Compression

Interframe compression compresses the information within sequences of frames, by specifying only the differences between given frames. Video information is typically divided into blocks, some of which remain the same (for instance background information in a non-moving shot), some of which move (moving objects or panned backgrounds). Instead of recoding this information, motion vectors encode the movement of macro blocks.

Sequences of frames are referred to as groups of pictures (or “GOP”). Frames within each group are indicated as I-frames, P-frames, and B-frames. I-frames are intracoded frames. Motion-prediction and Motion-compensation are applied to generate P-frames (or predicted frames), and B-frames (or in-between frames) are extrapolated from I- and P- frames. Formats that use these techniques are referred to as GOPPER formats, and are not typically used for professional video editing because they do not supply unique information for every frame.

Popular Compression Standards

There are several competing video compression standards. Some of the more popular standards are:

• Apple Video – used in early versions of QuickTime
• MPEG-1 – used in original digital video broadcasts and Video CD.
• MPEG-2 – used in terrestrial over the air digital video broadcasts, Super Video CD, and DVD. It is also the basis for HDV compression (the High-Definition successor to DV)
• MPEG-4 – DivX, 3Vix, XVid, Nero Digital and QuickTime 6 implement MPEG-4 part 2; x264, Nero Digital AVC, Quicktime 7, HD-DVD and BluRay implement MPEG-4 part 10.
• Real Video – Proprietary codec used by RealPlayer
• Windows Media Video – Proprietary codec by Microsoft originally based on MPEG-4 part 2; recent versions are the basis for the VC-1 codec supported in the HD-DVD and BluRay standards.

Containers

Containers are formats in which audio and video can be collected in a single file. Container types like Microsoft’s Audio Video Interleave container (or AVI) or the Matroska container (MKV) are often confused for video compression formats. To play a video file, one must have support for both the container (usually a “splitter” is involved – this separates the video and audio components and connects them to the appropriate filter), and the codec (i.e. MPEG-2).