News - H.264 video may soon step aside for twice-as-efficient H.265
Quote:
MPEG issues a draft specification, hopes for commercial products as early as 2013.
Read more.
Re: News - H.264 video may soon step aside for twice-as-efficient H.265
So, how many patents will there be in this "standard"?
Re: News - H.264 video may soon step aside for twice-as-efficient H.265
Harumph!
Why couldn't they have figured this out before all the broadcasters chose a spec? :( If it's twice as efficient, that would mean 1080p broadcasts in the same space as we currently have 1080i (I know the beeb occasionally broadcast 1080p on DVB-T2 but it seems problematic for a lot of people)........but are we likely to see it in the next 10-20 years? The best chance would be Sky.....and even that seems unlikely to happen until they adopt the 4k standards.......so don't hold your breath.........
Re: News - H.264 video may soon step aside for twice-as-efficient H.265
Quote:
Originally Posted by
Steve
So, how many patents will there be in this "standard"?
At least a dozen or two, I'm sure.
Re: News - H.264 video may soon step aside for twice-as-efficient H.265
http://en.wikipedia.org/wiki/High_Efficiency_Video_Coding
In short, the biggest compression gains are due to more efficient and adaptive coding blocks sizes organized in a coding tree (quadtree) and better/extended temporal prediction support. There are other optimizations but these two seem to be most responsible for x.265 (or HEVC, "MPEG-5"?) comparing - favorably so - with MPEG-4 AVC in terms of quality per bitrate, even though it will require more processing power to encode/decode. I suspect most set-top-boxes out there would be too slow in decoding high profile HEVC (level 4 and up, or 1080p and beyond) and will, at the very least, require new firmwares to decode any. In the mean time, the "8x8 Transform" (also featured among many Handbrake's advanced settings) should enable similar additional compression without sacrificing quality and/or playback support with current decoders and a small computational overhead.