Very good question Steve
.
H.265 or its marketing name, HEVC is indeed the next generation video compression technology. Its goal is to double the efficiency of H.264 which is the current standard used in Blu-ray among many other applications.
As a way of history, traditionally two standards organizations produced somewhat competing standards. One was MPEG which produced such standards as MPEG-2 (for video) and MP3 (for audio). ITU which is a European based organization created its own set of standards which go by the designation H.26X with X being different numbers. A few years ago, ITU got way ahead. MPEG produce something called MPEG-4 which was pretty lame at just 30% better than MPEG-2 whereas ITU created H.264 which was twice as good (at lower bit rates). Clever people at MPEG decided to join them if they can't beat them and decided to adopt H.264 as their own. Both it and HEVC/H.267 are now "joint" projects between the two organizations. You can read more about this historical perspective in this article I wrote a while back:
http://www.madronadigital.com/Library/Video Compression.html
OK, back to H.265. It is very challenging to keep improving the efficiency of video codecs. There is no new technique that all of a sudden gives you 2X improvement. Instead, people come up with an idea here, and idea there. Each one may gain you 5 to 10% depending the image being compressed and cleverness of the encoder. Put a lot of these together and you get your 2X target. Alas, as I just noted, not every image will benefit from all the tools. And not every encoder will implement every feature. In practice, compression gains will be varied with one video for example gaining 30% efficiency and the other 60%.
Standardization also takes a long time. To wit, HEVC spec work was just finished. So the real work of implementing it starts now. As with previous standards, some implementations will be incomplete. That is allowed by the way. All of these audio and video standard define how you *decode* the compressed stream. They never say how you encode them! They do that as to allow room for differentiation between encoder companies. This means as a person writing an encoder, I can choose to implement a subset and still call it the same name as another encoder implementing the full set. Even if I implemented the full set, I may take shortcuts on how clever my implementation may be.
As an example, these video codecs allow the encoder to select different rectangles/square set of pixels to encode. Which size is better requires a lot of analysis and iterations to determine but nothing in the standard stops you from quickly picking something up based on shallow analysis.
For all of these reasons, how much better a codec is, is hard to determine. Encoders tend to get better with time as people get more experience with the tools available, and computation power available increases. On that front, encoders that have to run real-time such as used for cable and satellite tend to have a harder time getting there since everything has to happen right now. So they tend to take more shortcuts than those that can take their time to produce a file that then gets viewed on demand.
I should also note that the percent improvement is determined using machine analysis and not a person. A metric called PSNR (power signal to noise ratio) is used to rate the quality of the video. By keeping PSNR the same, you can then vary the bit rate to see where they hand for the same PSNR. The problem with this is that our eyes don't work like PSNR meters (similar problem to audio
). Correlation varies from good to poor between PSNR and human quality assessments. So take numbers given on efficiency improvements with a grain of salt.
I say for now, it is safe to say that you get 1.5 times efficiency relative to best H.264. That is, if you are using 10 mbit/sec, you can use 7 mbit/sec for H.265/HEVC. The claimed performance would mean 5 mbit/sec or half as much. We may get there and on some clips but for now, I remain conservative.
Back to your question, "4K" video (or UHD as its is called now) has four times the pixels of 1080p: twice in each dimension. The hope is that by using HEVC one gets to encode 4K at the same bit rates as 1080p is done today. You already notice that the math does not work. Even at claimed 2X improvement for HEVC, you don't get to 4X. But that math is partly faulty. To need 4K would mean that all of those extra pixels are different than when used in 1080p. That tends to not be the case. Think of blue sky in a scene. Whether you encode it using 1920 horizontal pixels or 3840, it will be roughly the same since the pixels are all shades of blue. What is the real multiplier? We don't know since it will be picture dependent. As an average though, we probably need 2X efficiency improvement to get good performance in 4K vs 1080p at the same bit rate. Since we are not going to get that out of HEVC in all sources, likely the bit rate needs to go up.
Long story short, yes, HEVC is thought to be the key enabler or bringing 4K to us. Without it, we have a tough situation where we are barely getting there with H.264 and 1080p especially online.
BTW the demos I saw of HEVC at NAB (largest US broadcast/professional video) were disappointing. Vendors were attempting to show the 2:1 advantage and failing there. There were artifacts that need to be taken out and perhaps backing off from aggressive targets of 2X before we get good quality.
Next year would be the time to look as far as availability of encoders and decoders. What is there is prototype software decoders. We need hardware decoding to pump out 4K efficiency and at low power. Fortunately computing power has become plentiful unlike previous generations. Video encoding/decoding can use multiple cores which is nice as that is what we get more of every month.
BTW, I expect the real launch of 4K to be online and not from physical media. While there has been some whispers of Blu-ray doing that, I think it is unlikely. There just isn't much support behind Blu-ray association advancing the formats anymore.
