High Efficiency Video

coding (HEVC):
The Next-Generation
Compression Technology
Matthew Goldman
Senior Vice President TV Compression Technology
Ericsson
High Efficiency Video Coding
Compression bitrate targets

MPEG-2 VIDEO
1994 50% bitrate
AVC
saving target
2003
Bitrate

50% bitrate HEVC

saving target
2013

1995 2005 2015

© Ericsson AB 2013 | 2013-03-25 | Page 2

 Video compression evolution
From AVC TO HEVC
› High Efficiency Video Coding (HEVC): A new standardized compression algorithm
– An evolution of AVC (H.264 | MPEG-4 Part 10)

› HEVC standardization
– A Joint Collaborative Team on Video Coding (JCT-VC) of MPEG & VCEG
– Aim: To deliver same picture quality for half the bitrate of AVC
› Up to 10x more computational complexity to encode and 2x-3x to decode
– Nomenclature: ISO/IEC 23008-2 MPEG-H Part 2 and ITU-T Rec. H.265

› Key dates
– March 2013 International Standard (Version 1)
Main Profile, Main 10 Profile, and Main Still Profile approved
– January 2014 FDAM – Range extensions (Contribution applications) & Multi-view extensions
– July 2014 FDIS – Scalable HEVC (SHVC)

© Ericsson AB 2013 | 2013-03-25 | Page 3

HEVC encoder
Headers
ENCODER CONTROL

ENTROPY CODER
Ref.
ALF SAO DF
Buffer
Bitstream
Source
Intra Recon T-1 Q-1

Prediction
Mode
ME/MC T Q
Dec

Residuals

© Ericsson AB 2013 | 2013-03-25 | Page 4

High level tool comparison
AVC HEVC
16X16 block size 64x64 block size

Hierarchical quad-tree
Various Inter partitions partitioning down to 8x8
down to 4x4 + 4x4 Transform Units

9 intra modes 35 intra modes

8x8 and 4x4 32x32, 16x16, 8x8 and

transform sizes 4x4 transform sizes

© Ericsson AB 2013 | 2013-03-25 | Page 5

Coding Tree

• Coding Tree is collection of

Coding Units (CU) – CU size
64x64 to 8x8
• CUs can have independent
coding modes
• Further partitioning using
Prediction Units (Motion Vectors)
• Independent Transform Tree
partitioning from 32x32 to 4x4

© Ericsson AB 2013 | 2013-03-25 | Page 6

Coding Tree

• Coding Tree is collection of

Coding Units (CU) – CU size
64x64 to 8x8
• CUs can have independent
coding modes
• Further partitioning using
Prediction Units (Motion Vectors)
• Independent Transform Tree
partitioning from 32x32 to 4x4

© Ericsson AB 2013 | 2013-03-25 | Page 7

Coding Tree

• Coding Tree is collection of

Coding Units (CU) – CU size
64x64 to 8x8
• CUs can have independent
coding modes
• Further partitioning using
Prediction Units (Motion
Vectors)
• Independent Transform Tree
partitioning from 32x32 to 4x4
© Ericsson AB 2013 | 2013-03-25 | Page 8
Coding Tree

• Coding Tree is collection of

Coding Units (CU) – CU size
64x64 to 8x8
• CUs can have independent
coding modes
• Further partitioning using
Prediction Units (Motion
Vectors)
• Independent Transform Tree
partitioning from 32x32 to 4x4
© Ericsson AB 2013 | 2013-03-25 | Page 9
 Coding Units: Prediction &
Transform Units

Source: JCTVC-A124

Separation of prediction and transform structures allows more flexible

and efficient coding of video under various conditions and resolutions
© Ericsson AB 2013 | 2013-03-25 | Page 10
HEVC tools – Intra Prediction

AVC HEVC
DC + DC + Planar +
8 directional modes 33 directional modes

© Ericsson AB 2013 | 2013-03-25 | Page 11

HEVC tools – In-loop filters

Deblocking filter
› Deblocking Filter No In-loop filters + SAO filter
– Similar to AVC deblocking filter but
does not filter 4x4 block edges

› Sample Adaptive Offset (SAO)

Filter
– Calculates edge and band offsets
signaled to decoder
– Offsets added to reconstructed pixels
– SAO is not restricted to block
boundaries

© Ericsson AB 2013 | 2013-03-25 | Page 12

Tool comparison:
AVC HP vs. HEVC MP
AVC HEVC
High Profile Main Profile
16x16 Macroblock Coding Unit quadtree structure, 64x64 down to 8x8
Partitions 16x16 to 4x4 Prediction Units, 64x64 to 8x8
square + non-square (inter) square + non-square (inter) + asymmetric (inter)
8x8 and 4x4 transforms Transform Units, 32x32 to 4x4, 4x4 skip
Intra prediction (8 directions + DC) Intra prediction (33 directions + DC + planar)
Inter prediction luma 6-tap + 2-tap, to ¼ pel Inter prediction luma 8-tap, to ¼ pel
Inter prediction chroma bi-linear interpolation Inter prediction chroma 4-tap. to 1/8 pel
Motion vector prediction Advanced motion vector prediction (spatial + temporal)
In-loop deblocking filter In-loop deblocking filter &
Sample Adaptive Offset (SAO) filter

CABAC or CAVLC CABAC using parallel operations

CABAC = Context Adaptive Binary Arithmetic Coding

CAVLC = Context Adaptive Variable Length Coding
© Ericsson AB 2013 | 2013-03-25 | Page 13
 HEVC potential
› For DTH, HEVC will give 40-50% saving over best AVC encoder today
– SD: AVC 1.5 - 2.5 Mbps  HEVC 0.8 - 1.5 Mbps
– HD: AVC 6 - 9 Mbps  HEVC 3 - 4.5 Mbps
– UHD 4K: AVC 16 - 30 Mbps (estimate)  HEVC 8 - 15 Mbps (for 2160p60)
› Fits in existing channel bandwidth of currently deployed HD!

› HEVC practical availability

– Driven by availability of receive devices
– First version of standard completed January 2013
– Software-based implementations available now; address a subset of applications
– First production silicon available during 2013, but does not support true 4K UHDTV
– Second production silicon available during 2014  first practical systemization 4Q14-1H15
– Implementation issues (impacts practical ability to deploy)
› Migration of deployed (legacy) vs. greenfield
› Intellectual property licensing needs to be settled  MPEG LA
› 4K UHDTV: Main 10 Profile settled but what higher frame rate?
› Better interlaced content support?
– Contribution market – Professional profiles (“RExt”) planned Jan. 2014  first systemization 1H15

© Ericsson AB 2013 | 2013-03-25 | Page 14

Industry drivers - Efficiency

Multi-screen DSNG UHDTV

Mobile TV • More HD • More HD • High bit-rate need
• Expensive • More screens • Expensive
bandwidth bandwidth
• Increasing
demand
xDSL reach
• More subscribers
Satellite
Distribution
• Spectrum efficiency

Terrestrial Broadcast
• Spectrum efficiency
© Ericsson AB 2013 | 2013-03-25 | Page 15
Increasing image formats

›“4KTV” = 3840 x 2160

© Ericsson AB 2013 | 2013-03-25 | Page 16

 4K Cinema

Workflow &
Distribution
Enabler
4K UHDTV

Capability
to Transmit
HEVC Applicability

HDTV
Picture Quality
Extra Chanels
Extra

SDTV
Channels

© Ericsson AB 2013 | 2013-03-25 | Page 17

LTE

Mobile
Bandwidth
HEVC version 1 contains
3 profiles
› Main Profile supports general consumer-grade video (4:2:0 8-bit)
› Main 10 Profile targets large screen consumer applications that require higher
resolution
– Main Profile + bit-depth up to 10 bits
– UHDTV and large screens
– Not for professional applications (contribution, mastering)
› Still Picture Profile targets specialty applications
– Strict sub-set of Main Profile
– One IDR picture
› Some practical concerns still remain …

© Ericsson AB 2013 | 2013-03-25 | Page 18

Interlaced tools disagreements
› De-interlace prior to encoding
– Shown to give great efficiency results
– Doesn’t fit all usage scenarios
› DTA, embedded transcoding
› Billions of pre-encoded MPEG-2 Video & AVC source content
› PAFF/MBAFF (AVC) vs. SAFF/Field Only Coding/Frame Only Coding (HEVC)
– What about PAFF without MBAFF?
› Chroma bleeding in 4:2:0 (no chroma motion vector offset when predicting top-
to-bottom / bottom-to-top field as AVC)
› WG11 has asked for more evidence before proceeding further

© Ericsson AB 2013 | 2013-03-25 | Page 19

Resolution revolution

© Ericsson AB 2013 | 2013-03-25 | Page 20

What is 4KTV? UHDTV?
› UHDTV as a standard
– ITU-R BT.1769 defines
› UHDTV Level 1 = 3840 x 2160
› UHDTV Level 2 = 7680 x 4320
›“4KTV” = 3840 x 2160 – ITU-R BT.2020 defines colorimetry

› Consumer Electronics Association

(CEA) October 2012
– “Ultra High Definition” or “Ultra HD”
› >8 million active pixels
› At least 3840 (H) x 2160 (V)
› At least 16:9 aspect ratio
› At least 1 digital input capable of
In context of broadcast television, carrying and displaying native
3840x2160 resolution without
“4KTV” is UHDTV Level 1 or relying on up-conversion
4K UHDTV
© Ericsson AB 2013 | 2013-03-25 | Page 21
UHDTVs are now available

© Ericsson AB 2013 | 2013-03-25 | Page 22

Visual perception - Resolution

0.01°

0.9x to 1.6x
diagonal
84” TV
6’ to 11' for
84” diagonal

© Ericsson AB 2013 | 2013-03-25 | Page 23

Driving UHDTV content & events

Movies

2014

TV dramas Live sports

© Ericsson AB 2013 | 2013-03-25 | Page 24
What format will industry
settle on for 4K UHDTV?
4K HEVC could require up to 80x more horsepower vs. HD AVC

SDTV HDTV 4KTV P30 8b 4:2:0 4KTV P60+ 10b 4:2:2

Film OCN

© Ericsson AB 2013 | 2013-03-25 | Page 25

Contribution: 4:2:2 vs. 4:2:0
4:2:0 production flow
Chroma samples

4:2: 4:2: 4:2:2 production flow

0 2

Chroma rate = ¼ Chroma rate = ½

Luma rate Luma rate

Saves 50% Saves 33%

bandwidth bandwidth
Keying with 4:2:0
Source: http://www.dv.com/dv/magazine/2006/November/DV0611.hdvout.fig4.jpg

© Ericsson AB 2013 | 2013-03-25 | Page 26

10-bit quantization
8-bit 10-bit
› Banding (posterization) with 8b, especially in
plain areas
– Sky, backgrounds, graphics, logo
– More noticeable with slow changes, such as fades

8 bit = 256 levels

10 bit = 1024 levels

10-bit operation does not cost bitrate …

© Ericsson AB 2013 | 2013-03-25 | Page 27

 Free bits!
› For gradually changing content
– 8b quantization of the prediction causes
residual errors
› See the error line in the graph
– 10b quantization makes a better predictor,
leaving less residual to be coded

› For complex content

– The quantization of the residuals results
means that any additional 10b data is likely
to be quantized out

› Net result
– Zero bit cost for 10b depth, just better
pictures!
© Ericsson AB 2013 | 2013-03-25 | Page 28
HEVC Range Extensions
› HEVC range extensions (RExt) will be an Annex to HEVC v1
– 4:2:2/4:4:4/RGB chroma formats
– Higher bit-depths, including 10/12/14
– Mixed Chroma formats (4:2:0 & 4:4:4) – mixed video (4:2:0) and computer graphics (4:4:4)
– Lossless Coding
› As of July 2013 meeting cycle
› Use square transform and square intra prediction like v1.0 (no rectangular transforms)
› Intra prediction angle adjustment for 4:2:2
› New tools for 4:4:4 / Screen Content Coding, such as intra motion compensation
› Potential Profiles, Main 12, Main 422 10, Main 422 12, Main 444 10, Main 444 12
› For Contribution, Exchange, & Distribution, likely 25-30% bitrate savings over AVC
Contribution encoders, 15% for HEVC-I
› Timeline
– July 2013 - Draft Amendment (DAM) completed
– January 2014 – Final Draft Amendment (FDAM) scheduled

© Ericsson AB 2013 | 2013-03-25 | Page 29

Possible profiles and
conflicting goals
Main-tier (lower bitrates)
Simple (few changes)
vs.
vs.
High-tier ([very] high [intra]
Complex (new tools)
bitrates)
Mixed Content
› Possible profiles
– Consumer – lower bitrates, few changes over v1
– Professional – high bitrates, new tools for performance gains, possible divergence from v1
› Mixed content – more applicable to consumer applications (e.g., gaming) but requires new tools
(divergence from v1)
– Is 4:4:4 now becoming relevant for consumer applications?

© Ericsson AB 2013 | 2013-03-25 | Page 30

 Until HEVC is ready … 4K UHD
Contribution using AVC
4K UHDTV display

4K source  4x 3G-SDI

or

fiber
4x AVC encoders 4x AVC decoders
synchronized phase-lock synchronization

© Ericsson AB 2013 | 2013-03-25 | Page 31

Scaleable HEVC (SHVC)

Add new reference pictures

for Scaleable and Multi-view Extensions
Annex F
Motion compensation on
up-sampled decoded base
Annex G layer pictures

New CU level coding tool for

Inter layer prediction
Annex H

© Ericsson AB 2013 | 2013-03-25 | Page 32