Monday, October 6, 2014

Wither HEVC Part 3

In our previous blog, we raised the problem of the relative rates of:

1. bitrate reduction, due to new generations of compression technology
vs.
2. bitrate proliferation, due to the introduction of higher-resolution broadcast standards

We can easily quantify this problem with a little simple math. Let’s start in 1994, with the introduction of MPEG-2, the original digital compression standard developed for broadcast technology. Let’s set 1994 MPEG-2 compression technology to 1 and, likewise, set 1994 720 x 480 SD resolution formats at 1. For present purposes, we can assume these two forces are roughly balanced: that is to say, MPEG-2 compression technology successfully reduces the bits generated by digital 720 x 480 SD formats to manageable levels for practical purposes of transmission and storage.
Given this 1 to 1 parity between SD formats and MPEG-2 compression, as long as SD formats continued to dominate TV broadcasting, there was no great practical urgency about developing better encoding technology. And, in fact, although better H.264/AVC (MPEG-4) compression technology became available as early as 2003, there was little interest among broadcasters in the new technology over the next several years – despite its ability to cut SD bitrates in half.

Widespread interest in better H.264/MPEG-4 compression technology only began to develop among broadcasters after 2007, when the replacement of 720 x 480 SD formats by 6X larger 1920 x 1080 HD formats first become common. But, while the flood of bits generated by HD formats made the inefficiency of older MPEG-2 compression patently obvious, even after broadcasters switched to next-generation H.264 compression, the bottom line was not a return to the old 1 to 1 (SD to MPEG-2) parity of 1994. Instead, with a 6X increase in bits due to new HD formats, balanced by a 2X reduction in bitrates from H.264 compression, the new HD to MPEG-4 parity level was reestablished at 3 to 1.

The industry now faces the prospect of a second transition to a new 2X better level of compression technology with H.265/HEVC (MPEG-5).  Although adoption of MPEG-5 for HD formats would practically restore the old 1994 parity level (1.5 to 1 vs 1 to 1), just as adoption of MPEG-4 technology waited on the spread of new, higher resolution HD picture formats, significant take-up of MPEG-5 compression is likely to wait on widespread adoption of the new 4X larger 3940 x 2160 4K picture format. With 4K formats, the bottom line will not be something closer to 1994 parity levels, but rather something substantially worse than current levels, with a 6 to 1 ratio of 4K bits to MPEG-5 compression capabilities.

Future developments seem more likely to continue this progressively worsening trend than to succeed in reversing or even slowing it. Projecting forward to a new 2X better level of compression technology with a future H.266/MPEG-6 step, this advance in bitrate reduction seems certain to be more than offset by a yet another 4X larger resolution step: the 7880 x 4320 8K picture format.   In the 2020s, then, the bottom line is likely to be a 12 to 1 ratio of bits to compression capabilities, measured by 1994 standards.

Is this decade-by-decade slide in bit ratio, from 1 to 1 in the 1990s, to 3 to 1 in the 2000s, to 6 to 1 in the 2010s, to 12 to 1 in the 2020s, a worry? And if it is not something we should worry about, then why not? That will be the subject of our next blog.