10.07.2015 Views

Cinematography-Theory-And-Practice

Cinematography-Theory-And-Practice

Cinematography-Theory-And-Practice

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Lossy and Lossless CompressionA lossless compression system is one that is capable of reducing thesize of digital data in a way that allows the original data to be completelyrestored, byte for byte. This is done by removing redundantinformation. For example, if an area of the image is all pure white,where normally the digital code would be a long list of zeros, thecompression might write a code that means “there is a row of 5000zeros,” which will use only a few bits of data instead of thousands.Much higher compression ratios (lower data rates) can be achievedwith lossy compression. With lossy compression, information is discardedto create a simpler signal. These methods take into accountthe limits of human perception: they try to lose only informationthat won’t be missed by the eye — at least that’s the theory. Lossycompression may be invisible to the eye but can have a negative effecton post-production and later generations of duplication.Chroma SubsamplingMany cameras use chroma subsampling as a basic form of compression.In this technology, the luminance (black-and-white brightness) issampled at a different rate than the chrominance (color). It is based onthe idea that a real RGB signal (such as you might get from gatheringindependent signals from the red, green, and blue sensors in a threechipcamera), contains redundant information: in essence, each ofthe channels contains a duplicate black-and-white image. An RGBsignal has potentially the richest color depth and highest resolution,but requires enormous bandwidth and processing power.With chroma subsampling, there might be twice as many samplesof the luminance as for the chrominance. This would be expressed as4:2:2, where the first digit is the luminance channel and the next twodigits are the chroma channels — they are sampled at half the rateof the luminance channel. Video that is 4:4:4, has the same chromasampling for color channels as for luminance. There are other variations— for example, Sony’s HDCam cameras sample at 3:1:1. Youmay occasionally see a fourth digit, such as 4:4:4:4; in this case thefourth number is the alpha channel, which contains transparencyinformation.Because the human visual system is much more sensitive to luminancethan to color, lower-resolution color information can be overlaidwith higher-resolution luma (brightness) information, to createan image that looks very similar to one in which both color and lumainformation are sampled at full resolution.MPEG and JPEGOther types of compression are MPEG and JPEG. MPEG standsfor Motion Picture Experts Group, and JPEG is an acronym for JointPhotographic Experts Group. Various forms of MPEG are used for bothvideo and audio. Both MPEG and JPEG are types of codecs: whichmeans compressor—decompressor.Some compression systems compress footage one frame at a time,interpreting the video as a series of still frames; this is called intraframecompression. Interframe compression systems can further compressdata by eliminating redundancy between frames. This leads tohigher compression ratios, but can sometimes put a bigger processingload on the editing system. As with HD cameras, codecs andediting software and hardware are constantly improving. It is usuallypossible to transcode older formats into whatever codec you want, butthere is almost always some loss of quality in the process.1080P HD2K HD2048 Pixels1080ScanLinesFigure 9.6. Different methods fordefining the resolution of an HDformat. Resolutions such as 720 or1080 refers to the number if scanlines measured vertically. 2K and 4Krefer to the number of pixels measuredhorizontally. In this case a 2Kframe is 2048 pixels across.HD cinematography153

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!