December 21, 2012

The Future of Encoding is Almost Here: HEVC (H.265) is Coming to 1 Billion Devices

The advancements in lossy video encoding have been both consistent and amazing. H.264 (or AVC), that much maligned DSLR de-facto codec, sought to yield improved quality over its predecessors such as MPEG-2, all the while using half the bitrate, or lower, than such earlier codecs. Now, High Efficiency Video Coding or HEVC -- likely to earn the alternate title H.265 -- seeks to do the same compared to H.264, once more halving the bit rates necessary for equivalent, or even higher, quality. As it turns out, the tech world is already saturated with devices set to support HEVC playback.

The general model of functionality that makes this sort of thing possible is basically 'smarter' compression, which produces media with a smaller storage footprint at a higher quality -- though this methodology means comparatively heavier processing intensiveness to decode in real time. Basically you sacrifice 'weight' in bits for harder 'thinking' in order to translate the more complex compression techniques. Everybody from the ISO/IEC's MPEG to the ITU-T's VCEGamong others, are poised to declare HEVC a jointly and universally accepted standard.

For what this palpably -- and visually -- means for the future of delivery encoding is demonstrated in the video below. Some may find a little irony in the fact that the clip's 1080p version is very likely encoded in H.264 (but the differences are still pretty clear).

http://www.youtube.com/watch?v=fmFo1asK1Sw

In terms of efficiency, HEVC Main Profile (or MP, one of the codec's sub-specs) shows over 35% improved efficiency over H.264 High Profile (the sub spec favored for broadcast and Blu-ray) in terms of peak signal to noise ratio. To accomplish this, HEVC uses a number of compression 'tools,' some of which are advancements over H.264 principles, others of which are new -- the major example of the latter is HEVC's optimization for parallel processing, which basically means you can decode multiple parts of the media at the same time. Additionally, one of the mainstays of this type of codec, the macroblock, has expanded from H.264's maximum of 16x16 pixels to a potential 64x64 -- other key facets such as prediction (spatially and temporally, in all kinds of directions) is also present.

One of the wildest features of this codec is something called internal bit depth increase. Here's an excerpt from Wikipedia, with the source citations left inline in case you case want to read the various documentation floating around:

Internal bit depth increase (IBDI) allows for pictures to be internally processed at a bit depth that is higher than the bit depth they are encoded at.[48][49][50] IBDI can be done at up to 14-bits and is processed at that bit depth up until the point where the pictures are fed into the loop filters.[50]

It's tough to say right now whether upcoming DSLR models will be switching to HEVC for video encoding. If they do, that may not actually be a good thing for DSLR filmmakers, because even though it means you save storage space while improving quality, more complex compression tactics mean potentially longer transcode times, or perhaps a messier situation for pushing the grade. That said, it's always nice to raise the quality ceiling of web video, where this codec will likely be seeing most of its action.

HEVC should also play a role in broadcast as well. While much of the TV world has switched to MPEG-4 encoding, they will surely be looking forward to a new standard that decreases bandwidth requirements, especially since this means they can add more channels and increase overall quality as they lower bitrates. HEVC will factor in to the adoption of 4K broadcasts and streaming as well. Much of the talk has been about the massive increase in bitrate for 4K, but the new H.265 standard will likely halve this requirement, and make broadcasting Ultra High Definition far more feasible within current limitations.

Either way, we're going to see plenty of devices that already support the codec when it finally hits primetime. According to Videonet, "iPads are among the 1 billion CE devices shipped in 2012 capable of decoding HEVC." We'll have to wait for all those tongue-twistingly-titled standards organizations to settle things next year before the codec starts to roll out.

Link: HEVC-capable devices hit 1B in 2012, ahead of standard -- Videonet

Your Comment

16 Comments

"because even though it means you save storage space while improving quality, more complex compression tactics mean potentially longer transcode times"

Why would this signify longer transcode times, especially in an era in which NLEs like Premiere will most likely support this codec natively?

December 21, 2012

-1
Reply
Kenny Stevenson

I would imagine so, at least in comparison to transcoding H.264 video.

December 21, 2012

1
Reply
Blah

It has less to do with the actual editing times and more to do with the final render from the native raw file format. The harder "thinking" would take place when you're exporting to H.265 and if you're working from a high resolution file stock then your machine will have to figure out what information to get rid of since in order to get to the final high image quality low/er bit rate output. More compression means longer processing on your export.

December 22, 2012

1
Reply

Simply, more complexity requires more math which requires more computing power. Just because an NLE supports a codec has nothing to do with transcode times. H264 & H265 are delivery codecs and while fine for the average consumer who will do simple cuts or no editing at all, they are terrible for acquisition as they provide for little manipulation in post. That's why raw is still ideal but has it costs in storage and post.

December 21, 2012

0
Reply
Space Captain

Guys, this bit could be very important:

"the major example of the latter is HEVC’s optimization for parallel processing, which basically means you can decode multiple parts of the media at the same time"

I'm unclear what this means, but what I HOPE it means is that the codec will do what Pro Res does, and leverage multi-core processors for encoding and decoding (H264 doesn't do this - one of the main reasons it takes so bloody long to encode versus proper post codecs).

December 21, 2012

0
Reply
Will S

I think you find Adobe media encoder used multiple cores for encoding h264

December 22, 2012

-1
Reply
sam

I'm hoping this will bring 4K DSLRs to the market in 2013

December 21, 2012

0
Reply
Dimitri

Who created that image? It doesn't bode well when you misspell efficiency in your product logo, and it's already on sites around the internet (see http://goo.gl/xUV5B).

December 22, 2012

1
Reply

Haha! EPIC Fial ;)

December 22, 2012

-3
Reply
Kraig

I understand that H.265 is a "heavier" compression than h.264 but exactly why would that make it harder to color grade in post? Or is that not what was meant by "...messier situation for pushing the grade."

I thought HEVC would be easier because it appears to hold more details, kinda like AVC with IPB frame compressions.

Is IPB footage harder to grade than IPP frames? I just assumed they were both equally bad.

December 22, 2012

0
Reply
P337

P337: It's heavier because when editing in that codec applying an effect means encoding a new "layer" of that particularly piece of video. So the harder a codec is to en- and decode, the harder it will be to edit, particularly with all kind of effects.

December 28, 2012

-1
Reply
Simon Falkentorp

Right, but wouldn't that only apply to editing straight on H.265? If we transcode to ProRes, edit on that then export to H.265 again would that not be just like using H.264 except with more image detail being retained per bit-rate?

We'll have to wait and see but I think overall it'll be better, I think I'm just really sick of working with AVC now lol.

December 30, 2012

1
Reply
P337

I feel like I've seen H.264 looking way better than what it looks like in this video... and H.265 will probably be just as annoying to work with as H.264. meh, we'll see

December 22, 2012

0
Reply
taz

They don't got over 2.2 MBPS. So it looks that compressed.

The bigger problem is that since it's on Youtube, the entire video goes through Youtube's compression pass as well.

December 22, 2012

0
Reply
Daniel

The good thing is, that camera processors become faster and faster. So if for example the new codec doubles the quality, and the processor at the same time is say twice as fast, the bitrate can be doubled (I presume). With less compression, an more than double the overall quality?

December 28, 2012

-2
Reply
Simon Falkentorp