[PPT] - Image and Video Coding: Video Coding Standards s k [ x , y ] u k [ x PowerPoint Presentation

SLIDE 1

Image and Video Coding: Video Coding Standards

scaling &

inv. transform

transform & quantization intra mode decision intra-picture prediction entropy coding in-loop filtering decoded picture buffer motion-comp. prediction motion estimation coding mode decision buffer for current picture input pictures sk[x, y] uk[x, y] quantization indexes qk[x, y] u′

k[x, y]

s∗

k [x, y]

bitstream s∗

k [x, y]

s′

k[x, y]

intra prediction modes f iprd( s∗

k [x, y] )

ˆ sk[x, y] − motion data s′

ref[x + mx, y + my]

coding modes, intra prediction modes, motion data

SLIDE 2

Video Coding Standards / Scope of Standards

Video Coding Standards

scope of standards

pre- processing video encoder video decoder post- processing

input samples bitstream

utput

samples

Why Standards ? Bitstream generated by an encoder should be reliably decodable by decoders of other manufacturers Image and video coding standards define “interface” between encoder and decoder Freedom for manufacturers to adapt actual implementations Scope of Video Coding Standards

1 Bitstream syntax

(including constraints for transmitted and derived parameters)

2 Decoding process

(example decoding process for conforming bitstreams) Encoding, pre- and post-processing are out of the scope of standards No guarantee of image/video quality

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 2 / 35

SLIDE 3

Video Coding Standards / Scope of Standards

Bitstream Syntax

Coded Representation Samples are representated by coding parameters

Coding modes, motion parameters, intra prediction modes Quantization indexes ...

Bitstream Syntax Format for representating coding parameters in the bitstream Order of transmission Conditions for presence

Motion data are only included for inter-picture coded blocks Intra prediction modes are only included for intra-picture coded blocks

Entropy coding

Codeword table Binarization and probability models

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 3 / 35

SLIDE 4

Video Coding Standards / Scope of Standards

Example for Bitstream Syntax: Motion Data in H.265 | HEVC

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 4 / 35

SLIDE 5

Video Coding Standards / Conformance Points

Interoperability and Conformance

Conformance Points Standards are designed for broad range of potential application areas Some features may not be appropriate for all application scenarios (would unnecessarily increase complexity of decoder implementations) Define different conformance points using profiles & levels Interoperability between applications with similar requirements Profiles and Levels Profile: Subset of coding tools supported in standards Level: Maximum values for key parameters (picture size, frame rate, bit rate, buffer sizes) Conformance Conforming bitstream: Obeys all constraints on syntax elements and derived parameters Conforming encoder: Generates conforming bitstreams Conforming decoder: Produces same result as decoding process specified in standard

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 5 / 35

SLIDE 6

Syntax Features and Coding Tools / General Design

Syntax Features and Coding Tools

Development of Video Coding Standards Videos have by far the most data of all information types we use today Usage of videos as well as resolutions continuously increase Always desire higher compression capabilities Advances in computing devices enables new possibilities in video compression Research in video compression offers new directions Standard include well-tested features implementable for considered applications Standards included in Comparison H.262 | MPEG-2 Video (first version: 1995) H.264 | AVC: Advanced Video Coding (first version: 2003) H.265 | HEVC: High Efficiency Video Coding (first version: 2013) H.266 | VVC: Versatile Video Coding (first version: 2020)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 6 / 35

SLIDE 7

Syntax Features and Coding Tools / General Design

Basic Design: Block-Based Hybrid Video Coding

scaling &

inv. transform

transform & quantization intra mode decision intra-picture prediction entropy coding in-loop filtering decoded picture buffer motion-comp. prediction motion estimation coding mode decision buffer for current picture input pictures sk[x, y] uk[x, y] quantization indexes qk[x, y] u′

k[x, y]

s∗

k [x, y]

bitstream s∗

k [x, y]

s′

k[x, y]

intra prediction modes f iprd( s∗

k [x, y] )

ˆ sk[x, y] − motion data s′

ref[x + mx, y + my]

coding modes, intra prediction modes, motion data

Hybrid Video Coding All standards follow same basic design Differ in many details Analyze Features in Following Areas Supported temporal coding structures Partitioning into blocks Intra-picture prediction Transform and quantization Entropy coding Motion-compensated prediction Motion parameter coding In-loop filtering

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 7 / 35

SLIDE 8

Syntax Features and Coding Tools / General Design

Basic Design: Block-Based Hybrid Video Coding

scaling &

inv. transform

transform & quantization intra mode decision intra-picture prediction entropy coding in-loop filtering decoded picture buffer motion-comp. prediction motion estimation coding mode decision buffer for current picture input pictures sk[x, y] uk[x, y] quantization indexes qk[x, y] u′

k[x, y]

s∗

k [x, y]

bitstream s∗

k [x, y]

s′

k[x, y]

intra prediction modes f iprd( s∗

k [x, y] )

ˆ sk[x, y] − motion data s′

ref[x + mx, y + my]

coding modes, intra prediction modes, motion data

Hybrid Video Coding All standards follow same basic design Differ in many details Analyze Features in Following Areas Supported temporal coding structures Partitioning into blocks Intra-picture prediction Transform and quantization Entropy coding Motion-compensated prediction Motion parameter coding In-loop filtering

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 7 / 35

SLIDE 9

Syntax Features and Coding Tools / Pictures and Partitioning

Picture Types and Supported Temporal Coding Structures

H.262 | MPEG-2 Video Three pictures types: I, P, and B No support of multiple reference pictures Two types of possible coding structures:

IPPP: Only possibility for low delay (no bi-prediction) IBBP: B picture as non-reference picture (between two I/P)

H.264 | AVC, H.265 | HEVC, and H.266 | VVC I, P, and B slices (picture can consist of multiple slices) Decoupling of coding order and picture/slice types B pictures can be used as reference picture More advanced and more efficient coding structures

B picture and multiple reference pictures for low delay Hierarchical B pictures when high delay is possible

I P P P P P P P P I/P P P B B B B B B I/B B B B B B B B B I/B B B B B B B B B

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 8 / 35

SLIDE 10

Syntax Features and Coding Tools / Pictures and Partitioning

Basic Concepts for Picture Partitioning into Blocks

H.262 | MPEG-2 Video Fixed 16×16 macroblocks Coding mode (intra/inter) selected on macroblock basis H.264 | AVC 16×16 macroblocks with potential splitting Coding mode selected on macroblock basis H.265 | HEVC Quadtree (QT) partitioning of CTUs (up to 64×64) Coding mode selected on CU basis (result of QT partitioning) H.266 | VVC Multi-type tree (MTT) partitioning of CTUs (up to 128×128) Coding mode selected on CU basis (result of MTT partitioning)

quadtree multi-type tree

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 9 / 35

SLIDE 11

Syntax Features and Coding Tools / Pictures and Partitioning

Partitioning for Prediction and Transform Coding

Blocks for Motion-Compensated Prediction MPEG-2: 16×16 macroblocks H.264 | AVC: Sub-partitioning of macroblocks (4×4 to 16×16 blocks) H.265 | HEVC: Sub-partitioning of coding units (4×8/8×4 to 64×64 blocks) H.266 | VVC: Coding units (result of MTT) (4×8/8×4 to 128×128 blocks) Blocks for Intra Prediction and Transform Coding MPEG-2: 8×8 transform blocks H.264 | AVC: 4×4 or 8×8 (also 16×16 for intra) H.265 | HEVC: Additional QT partitioning of CUs H.266 | VVC: Coding units (result of MTT), intra sub-partitioning

N×N N×(N/2) (N/2)×N (N/2)×(N/2) N×(N/4) N×(3N/4) (N/4)×N (3N/4)×N CTU

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 10 / 35

SLIDE 12

Syntax Features and Coding Tools / Intra-Picture Prediction

Intra-Picture Prediction

H.262 | MPEG-2 Video DC prediction only (quantization index) H.264 | AVC 9 spatial intra prediction modes (8 angles) H.265 | HEVC 35 spatial intra prediction modes (33 angles) H.266 | VVC 67 conventional prediction modes (out of 93, wide angles) Position dependent prediction sample filtering (PDPC) Multi-reference line intra prediction (MRL) Matrix-based intra prediction (MIP) Cross-component linear model (CCLM) for chroma

45◦ −135◦

32

32
32

32

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 11 / 35

SLIDE 13

Syntax Features and Coding Tools / Motion-Compensated Prediction

Motion Vector Precision and Interpolation

Motion Vector Precision (in luma samples) MPEG-2 Video: Half-sample precision AVC, HEVC: Quarter-sample precision H.266 | VVC: Adaptive motion vector resolution (AMVR) block-adaptive (4, 1, 1/2, 1/4) Luma Interpolation MPEG-2 Video: Bi-linear interpolation H.264 | AVC: 6-tap filters (half) + linear (quarter) H.265 | HEVC: 8/7-tap filters H.266 | VVC: 8/7-tap + switched half-sample filters Chroma Interpolation MPEG-2, AVC: Bi-linear interpolation HEVC, VVC: 4-tap filters

filter coefficients h[k] phase

3
2
1

1 2 3 4 1/4 48 16 MPEG-2 1/2 32 32 3/4 16 48 1/4 1

5

52 20

5

1 AVC 1/2 2

10

40 40

10

2 3/4 1

5

20 52

5

1 1/4

1

4

10

58 17

5

1 HEVC/VVC 1/2

1

4

11

40 40

11

4

1

3/4 1

5

17 58

10

4

1

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 12 / 35

SLIDE 14

Syntax Features and Coding Tools / Motion-Compensated Prediction

Motion-Compensated Prediction

H.262 | MPEG-2 Video P pictures: Previous I/P picture B pictures: Preceding and succeeding I/P picture H.264 | AVC and H.265 | HEVC Multiple reference pictures (both P and B slices) Generalized bi-prediction, weighted prediction H.266 | VVC Multiple reference pictures, generalized bi-prediction Block-adaptive bi-prediction weights Geometric partitioning (merge mode only, 64 shapes) Prediction refinement based on optical flow Combined intra/inter prediction (weighted sum)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 13 / 35

SLIDE 15

Syntax Features and Coding Tools / Motion-Compensated Prediction

Motion Parameter Prediction and Coding

H.262 | MPEG-2 Video Motion vector prediction using left neighbor Temporal direct mode for motion inference (B pictures) H.264 | AVC Median prediction, special cases for 16×8 and 8×16 Temporal direct or spatial direct mode H.265 | HEVC Switched motion vector prediction Merge mode with spatial and temporal candidates H.266 | VVC Switched motion vector prediction (similar to HEVC) Extended merge mode with spatial candidates, temporal candidates, and history-based candidates

m(l) m A B C current block m1 m0 mcol

time

t0 tc t1

co-located block current block A0 A1 B0 B1 B2

current block

T0 T1 co-located area in reference picture

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 14 / 35

SLIDE 16

Syntax Features and Coding Tools / Motion-Compensated Prediction

Improvements for Motion Parameter Coding in H.266 | VVC

Improved Temporal Motion Vector Prediction Derive local motion field from reference picture Offsets are derived from local neighboring blocks Motion Vector Difference Coding Merge mode with restricted motion vector difference Symmetric motion vector difference (bi-prediction) Affine Motion Model Affine model with 4 or 6 parameters 4×4 motion compensation with 1/16-th sample precision Decoder Side Motion Vector Refinement (DMVR) For merge blocks with “true” bi-directional prediction Decoder search: Minimize SAD between reference blocks

m1
m0

∆m −∆m Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 15 / 35

SLIDE 17

Syntax Features and Coding Tools / Transform Coding

Transformation of Prediction Error Blocks

H.262 | MPEG-2 Video Separable 8×8 DCT-2 H.264 | AVC Integer approximation of DCT-2 Transform sizes 4×4, 8×8, cascaded 16×16 H.265 | HEVC Transform sizes 4×4, 8×8, 16×16, 32×32 (integer) DST-7 for intra-predicted 4×4 blocks H.266 | VVC Multiple transform types: DCT-2, DST-7, DCT-8 Low-frequency non-separable transform (LFNST) Subblock transform (SBT) for inter-coded blocks Joint coding of chroma residuals (JCCR)

primary transforms in VVC codeword horizontal vertical DCT-2 DCT-2 100 DST-7 DST-7 101 DCT-8 DST-7 110 DST-7 DCT-8 111 DCT-8 DCT-8

8×8 LFNST: 256 → 16 4×4 LFNST: 16 → 8

DCT-8 DST-7 DST-7 DST-7 DST-7 DCT-8 DST-7 DST-7 Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 16 / 35

SLIDE 18

Syntax Features and Coding Tools / Transform Coding

Quantization of Transform Coefficients

H.262 | MPEG-2 Video Uniform reconstruction quantizer (URQ) for intra Modified URQ with wider deadzone for inter H.264 | AVC and H.265 | HEVC Uniform reconstruction quantizer (URQ) HEVC includes optional sign data hiding (one sign is derived from parity of absolute sum) H.266 | VVC Uniform reconstruction quantizer (URQ) Optional sign data hiding Alternative: Trellis-coded quantizer (TCQ)

Very simple variant of vector quantization Decoder similar to URQ Encoder: Trellis search using Viterbi algorithm

∆ −∆ 2∆ −2∆ 3∆ −3∆ 4∆ −4∆ 1 −1 2 −2 3 −3 4 −4 ∆

−4 −3 −2 −1 1 2 3 4 s/∆

Q0 Q1

−2 2 −1 1 −2 2 −1 1

next state state Q even

dd

Q0 2 1 Q0 2 2 Q1 1 3 3 Q1 3 1 – → 0 → 1 → 2 → 3 → · · · · · · · · · · · · · · ·

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 17 / 35

SLIDE 19

Syntax Features and Coding Tools / Transform Coding

Entropy Coding and Transform Coefficient Coding

H.262 | MPEG-2 Video Variable-length coding (VLC) tables Coefficients: zig-zag scan + run-level coding H.264 | AVC Two entropy coding methods

1 Context-adaptive variable length coding (CAVLC) 2 Context-based adaptive binary arith. coding (CABAC)

Coefficients: CAVLC or CABAC H.265 | HEVC and H.266 | VVC Context-based adaptive binary arith. coding (CABAC) Coefficients: Advanced context modeling VVC includes improvements of general coding engine and the context modeling for quantization indexes

binary arithmetic coder bypass mode regular mode binarizer context modelling regular arithmetic coding engine bypass arithmetic coding engine

previous bin for update bins bin bin bin ctx

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 18 / 35

SLIDE 20

Syntax Features and Coding Tools / In-Loop Filters

In-Loop Filters

H.262 | MPEG-2 Video No in-loop filters included H.264 | AVC Deblocking filter H.265 | HEVC Deblocking filter Sample Adaptive Offset (SAO) filter H.266 | VVC Reshaper: Luma Mapping with Chroma Scaling Deblocking filter Sample Adaptive Offset (SAO) filter Adaptive Loop filter (Wiener filter)

block boundary block P block Q reconstructed samples before deblocking modified samples after deblocking p3 p2 p1 p0 q0 q1 q2 q3 p'

1

p' q' q'

1

a c b a c b a c b a c b a c b

category 1

a c b a c b

category 2

a c b a c b

category 3

a c b

category 4 Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 19 / 35

SLIDE 21

Comparison of Coding Efficiency / Comparison of Standards

Comparing Coding Efficiency of Standards

Measuring Coding Efficieny Average bit rate R = size of generated bitstream (in bits) nominal duration of video sequence (in seconds) Average PSNR with weighting of color components (for YCbCr 4:2:0) PSNRYCbCr = (6 · PSNRY + PSNRCb + PSNRCr) / 8 Encoder Configurations Enable all coding tools that contribute to coding efficiency Choose best known configuration (e.g., coding structure) for each standard Results show capabilities of video coding standards Encoder complexity largely varies over tested standards

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 20 / 35

SLIDE 22

Comparison of Coding Efficiency / Comparison of Standards

Unified Encoder Control

Encoder Control for Different Video Coding Standards Standards only specify bitstream syntax and decoding process During development of standards: Example encoding process (Test Model / Reference Software) Encoding technology has been improved over time Unified Lagrangian Encoder Control Use the same Lagrangian encoder control for all standards

Lagrangian mode decision (including transform, quantization, entropy coding) Lagrangian motion estimation (SAD for integer search, Hadamard SAD for sub-sample refinement) Rate-distortion optimized quantization

Improves encoders for older standards Fair comparision of achievable coding efficiency

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 21 / 35

SLIDE 23

Comparison of Coding Efficiency / Intra-Only Video Applications

Intra-Only Video Applications

I I I I I I I I I I I I

Considered Application Space Mainly still image coding, some professional editing applications Analysis of improvements for intra-picture coding Encoder Configuration All pictures are coded separately No inter-picture prediction enabled (all pictures are coded as I pictures) All intra-picture coding tools enabled that are supported in standard

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 22 / 35

SLIDE 24

Comparison of Coding Efficiency / Intra-Only Video Applications

Example: ”Kimono” (1920 x 1080, 24 Hz)

2 4 6 8 10 12 14 16 18 20 22 24 26 36 37 38 39 40 41 42 43

MPEG-2 AVC HEVC VVC

46% 27% 27% 71%

bit rate [Mbits/s] PSNRYCbCr [dB]

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 23 / 35

SLIDE 25

Comparison of Coding Efficiency / Intra-Only Video Applications

Subjective Comparison: “Kimono” in Intra-Only Configuration

H.266 | VVC @ 6 Mbits/s (50%) H.262 | MPEG-2 @ 12 Mbits/s (100%)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 24 / 35

SLIDE 26

Comparison of Coding Efficiency / Intra-Only Video Applications

Coding Efficiency Comparison: Intra-Only Video Applications

average over five HD sequences (1920×1080) bit-rate savings relative to ... MPEG-2 H.264/AVC H.265/HEVC H.264/AVC 32 % H.265/HEVC 45 % 21 % H.266/VVC 57 % 38 % 23 %

Average Bit Rate Savings Bit-rate savings based on PSNR as quality measure Averages over all sequences and reasonable quality range Large improvements from one standard generation to the next (about 20-30%) Roughly 50-60% bit-rate savings from MPEG-2 (1995) to VVC (2020)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 25 / 35

SLIDE 27

Comparison of Coding Efficiency / Interactive Video Applications

Interactive Video Applications

I P P P P P P P P I B B B B B B B B

Considered Application Space Video conferencing, video telephony, video chats Require very low encoding/decoding delay Encoder Configuration All pictures are coded in acquisition/display order, only one I picture at start of sequence Use B pictures, multiple reference pictures, and QP cascading when support in standard All coding tools enabled that yield coding efficiency gains

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 26 / 35

SLIDE 28

Comparison of Coding Efficiency / Interactive Video Applications

Example: ”Johnny” (1280 x 720, 60 Hz)

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 36 37 38 39 40 41 42 43 44 45

MPEG-2 AVC HEVC VVC

63% 54% 43% 90%

bit rate [Mbits/s] PSNRYCbCr [dB]

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 27 / 35

SLIDE 29

Comparison of Coding Efficiency / Interactive Video Applications

Subjective Comparison: “Johnny” in Low Delay Configuration

H.266 | VVC @ 100 kbits/s (20%) H.262 | MPEG-2 @ 500 kbits/s (100%)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 28 / 35

SLIDE 30

Comparison of Coding Efficiency / Interactive Video Applications

Coding Efficiency Comparison: Interactive Video Applications

average over six HD sequences (1280×720) bit-rate savings relative to ... MPEG-2 H.264/AVC H.265/HEVC H.264/AVC 59 % H.265/HEVC 75 % 39 % H.266/VVC 86 % 65 % 42 %

Average Bit Rate Savings Bit-rate savings based on PSNR as quality measure Averages over all sequences and reasonable quality range Large improvements from one standard generation to the next (about 40-60%) Roughly 85% bit-rate savings from MPEG-2 (1995) to VVC (2020)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 29 / 35

SLIDE 31

Comparison of Coding Efficiency / Entertainment Video Applications

Entertainment Video Applications

B0 I0 B0 B1 B2 B2 B1 B2 B2 – – – random access B0 B0 B1 B2 B2 B1 B2 B2 5 1 2 3 4 6 7 8

Considered Application Space Television broadcast, optical discs Video streaming (over the Internet or wireless) Encoder Configuration Random access points in regular intervals (about 1 second), open-GOP structures, high delay Use most efficient coding structures supported in standard (hierarchical B pictures when possible) All coding tools enabled that yield coding efficiency gains

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 30 / 35

SLIDE 32

Comparison of Coding Efficiency / Entertainment Video Applications

Example: ”BasketballDrive” (1920 x 1080, 50 Hz)

2 4 6 8 10 12 14 16 18 20 34 35 36 37 38 39 40 41

MPEG-2 AVC HEVC VVC

49% 40% 43% 83%

bit rate [Mbits/s] PSNRYCbCr [dB]

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 31 / 35

SLIDE 33

Comparison of Coding Efficiency / Entertainment Video Applications

Subjective Comparison: “BasketballDrive” in Random Access Configuration

H.266 | VVC @ 1 Mbits/s (20%) H.262 | MPEG-2 @ 5 Mbits/s (100%)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 32 / 35

SLIDE 34

Comparison of Coding Efficiency / Entertainment Video Applications

Objective vs Subjective Results: “MountainBay” (3840 x 2160, 10 Bit, 30 Hz)

1 2 3 4 5 6 33 34 35 36 37 38 39 HEVC VVC 26% average bit-rate savings bit rate [Mbits/s] PSNR 1 2 3 4 5 6 1 2 3 4 5 6 7 8 HEVC VVC ≈ 50% average bit-rate savings (subjective quality)

[ data from JVET-S0246 ]

bit rate [Mbits/s] MOS

Comparison of objective (PSNR) and subjective quality measure (MOS = Mean Opinion Score) Improvements in subjective quality are often higher than indicated by PSNR

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 33 / 35

SLIDE 35

Comparison of Coding Efficiency / Entertainment Video Applications

Coding Efficiency Comparison: Entertainment Video Applications

average over five HD sequences (1920×1080) bit-rate savings relative to ... MPEG-2 H.264/AVC H.265/HEVC H.264/AVC 47 % H.265/HEVC 65 % 34 % H.266/VVC 79 % 61 % 41 %

Average Bit Rate Savings Bit-rate savings based on PSNR as quality measure Averages over all sequences and reasonable quality range Large improvements from one standard generation to the next (about 35-40%) Roughly 80% bit-rate savings from MPEG-2 (1995) to VVC (2020)

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 34 / 35

SLIDE 36

Summary

Summary of Lecture

intra only low delay random access H.262 | MPEG-2 H.264 | AVC H.265 | HEVC H.266 | VVC

Video Coding Standards All today’s standards follow the basic approach of hybrid video coding Standards specify bitstream syntax and decoding process Main improvement: More coding options for representing block of samples Coding Efficiency Comparisons Unified and highly efficient Lagrangian encoder control Substantial improvements in coding efficiency from one standard generation to the next

Heiko Schwarz (Freie Universität Berlin) — Image and Video Coding: Video Coding Standards 35 / 35