This is just arithmetic. You can work all these figures out yourself but it’s really useful having some of the key numbers already to hand. Using the ITU-R BT.601 4:2:2 digital coding standard for SD, each picture occupies a large amount of storage space – especially when related to computer storage devices such as DRAM and disks. So much so that the numbers can become confusing unless a few benchmark statistics are remembered. Fortunately the units of mega, giga and tera make it easy to express the vast numbers involved; ‘one gig’ trips off the tongue far more easily than ‘one thousand million’ and sounds much less intimidating.
Storage capacities for SD video can all be worked out directly from the 601 standard. Bearing in mind that sync words and blanking can be re-generated and added at the output, only the active picture area need be stored on disks. In line with the modern trend of many disk drive manufacturers, kilobyte, megabyte and gigabyte are taken here to represent 103, 106 and 109 respectively.
Every line of a 625/50 or 525/60 TV picture has 720 luminance (Y) samples and 360 each of two chrominance samples (Cr and Cb), making a total of 1,440 samples per line.
There are 576 active lines per picture creating 1440 x 576 = 829,440 pixels per picture.
Sampled at 8 bits per pixel (10 bits can also be used) a picture is made up of 6,635,520 bits or 829,440 8-bit bytes – generally written as 830 kB.
With 25 pictures a second there are 830 x 25 = 20,750 kbytes or 21 Mbytes per second.
There are 480 active lines and so 1,440 x 480 = 691,200 pixels per picture.
With each pixel sampled at 8-bit resolution this format creates 5,529,600 bits, or 691.2 kbytes per frame. At 30 frames per second this creates a total of 21,039 kbytes, or 20.7 Mbytes per second.
Note that both 625 and 525 line systems require approximately the same amount of storage for a given time – 21 Mbytes for every second. To store one hour takes 76 Gbytes. Looked at another way each gigabyte (GB) of storage will hold 47 seconds of non-compressed video. 10-bit sampling uses 25% more storage.
If compression is used, and assuming the sampling structure remains the same, simply divide the numbers by the compression ratio. For example, with 5:1 compression 1 GB will hold 47 x 5 = 235 seconds, and 1 hour takes 76/5 = 18 GB (approx). The storage requirement for VBR compression cannot be precisely calculated but there is usually some target average compression ratio or data rate figure quoted.
All media are limited by the bandwidth available in the transmission/delivery channel. There is a wide choice of services and screens. In the most restricted cases some wireless and mobile applications are supported with a variety of small screens, shapes and resolutions ranging from VGA (480×640) and some 3 or 4G phones with up to 320×240, or 176×144 pixels and frame rates down to 15Hz. Many modern smart phones boast 1920 x 1080 HD screens.
There are many video formats for HD but the 1920 x 1080 format is popular. Using 4:2:2 sampling, each line has 1920 Y samples and 960 each of Cr and Cb = 3840 samples per line. So each picture has 3840 x 1080 = 4.147 M samples. For 10-bit sampling each picture has the equivalent data of 5.18 M (8-bit) bytes. Assuming 30 pictures (60 fields) per second these produce 155 M bytes/s – 7.4 times that of SD. An hour of storage now needs to accommodate 560 GB.
Ultra High Definition has two sizes of picture – 4K and 8K. 4K is 2160 x 3840 twice the length and breadth of 1080 HD. If using 4:2:2 10-bit sampling then each picture is 16.588 M samples, equivalent data of 20.735 MB. At 30 f/s that amounts to 622.05 MB/s, .2.24TB/h.
8K at 4320 x 7680 is twice the size, and four times the area of 4K. One frame is 66.355 Msamples, or 82.94 MB. At 30 f/s this produces 2.488 GB/s, making an hour nearly 9 TB of data.
2K and 4K DCI
2K is a format used in digital movie production that uses 4:4:4 10-bit sampling and RGB colorspace with an image size of 2048 x 1536, and has 24 frames per second. This makes one frame 11.80 MB, and an hour of storage 1.04TB. Note that, applied to digital cinema exhibition, the 2K pixel size is 2048 x 1080, and the color space is X´Y´Z´ and uses 12-bit 4:4:4 sampling, as defined by the DCI. The 4K image size is increasingly being used for. It is a 2×2 version of 2K, making x4 the number of pixels.
Here are some popular TV and digital film formats showing the volume of their uncompressed data. Compression of up to 100:1 is applied to MPEG-2 TV transmissions – over 100:1 may be used with more advanced codecs such as MPEG-4 and VC-1. DCI have given a maximum data rate for replay in digital cinemas is 250 Mb/s. Here JPEG 2000 compression is used and there is no inter-frame compression; this works out at a compression of about 6.4:1 for 2K and 25.5:1 for 4K.
|Format (H x V)||Sampling (MB)||Image size Mb/s||One Frame (GB)||Data rate||One Hour|
|320/15P||4:1:1 8-bit||320 x 240||0.12||14.4||6.5 (3G phone)|
|525/60I||4:2:2 8-bit||720 x 480||0.69||166||76|
|625/50I||4:2:2 8-bit||720 x 576||0.83||166||76|
|720/60P||4:2:2 10-bit||1280 x 720||2.3||1104||500|
|1080/60I||4:2:2 10-bit||1920 x 1080||5.2||1248||560|
|1080/25P||4:4:4 10-bit||1920 x 1080||7.8||1560||700 (RGB)|
|1080/60P||4:4:4 10-bit||1920 x 1080||7.8||3744||1680 (RGB)|
|2K/24P||4:4:4 12-bit||2048 x 1080||10||1913||860 (DCI cinema)|
|2K/24P||4:4:4 10-bit||2048 x 1536||12||2304||1036 (cine production)|
|4K/60P||4:2:2 10-bit||3840 x 2160||20.7||4976||2240 (4K UHD)|
|4K/24P||4:4:4 12-bit||4096 x 2160||39.8||7650||3442 (DCI cinema)|
|4K/24P||4:4:4 10-bit||4096 x 3072||48||9216||4144 (cine production)|
|8K/60P||4:2:2 10-bit||7680 x 4320||82.9||19904||8960 (8K UHD)|