Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Transform and Entropy Coding in AV2

Alican Nalci, Hilmi E. Egilmez, Madhu P. Krishnan, Keng-Shih Lu, Joe Young, Debargha Mukherjee, Lin Zheng, Jingning Han, Joel Sole, Xin Zhao, Tianqi Liu, Liang Zhao, Todd Nguyen, Urvang Joshi, Kruthika Koratti Sivakumar, Luhang Xu, Zhijun Lei, Yue Yu, Aki Kuusela, Minhua Zhou, Andrey Norkin, Adrian Grange

TL;DR

AV2 strengthens AV1 by enhancing transform and entropy coding through redesigned kernels, data-driven transforms, expanded partitioning, and mode/coefficient–dependent signaling. Key innovations include IST, CCTX, MDTX, DCTX, and ATC for improved energy compaction, along with TCQ, PARA, FSC, PH, and TR-based HR coding for more efficient coefficient signaling. The framework preserves high throughput and moderate memory growth, achieving notable BD-rate gains under all AV2 Common Test Conditions across all profiles. Together, these advances offer meaningful bitrate reductions while maintaining practical encoder/decoder complexity for natural and screen content, including lossless paths.

Abstract

AV2 is the successor to the AV1 royalty-free video coding standard developed by the Alliance for Open Media (AOMedia). Its primary objective is to deliver substantial compression gains and subjective quality improvements while maintaining low-complexity encoder and decoder operations. This paper describes the transform, quantization and entropy coding design in AV2, including redesigned transform kernels and data-driven transforms, expanded transform partitioning, and a mode & coefficient dependent transform signaling. AV2 introduces several new coding tools including Intra/Inter Secondary Transforms (IST), Trellis Coded Quantization (TCQ), Adaptive Transform Coding (ATC), Probability Adaptation Rate Adjustment (PARA), Forward Skip Coding (FSC), Cross Chroma Component Transforms (CCTX), Parity Hiding (PH) tools and improved lossless coding. These advances enable AV2 to deliver the highest quality video experience for video applications at a significantly reduced bitrate.

Transform and Entropy Coding in AV2

TL;DR

AV2 strengthens AV1 by enhancing transform and entropy coding through redesigned kernels, data-driven transforms, expanded partitioning, and mode/coefficient–dependent signaling. Key innovations include IST, CCTX, MDTX, DCTX, and ATC for improved energy compaction, along with TCQ, PARA, FSC, PH, and TR-based HR coding for more efficient coefficient signaling. The framework preserves high throughput and moderate memory growth, achieving notable BD-rate gains under all AV2 Common Test Conditions across all profiles. Together, these advances offer meaningful bitrate reductions while maintaining practical encoder/decoder complexity for natural and screen content, including lossless paths.

Abstract

AV2 is the successor to the AV1 royalty-free video coding standard developed by the Alliance for Open Media (AOMedia). Its primary objective is to deliver substantial compression gains and subjective quality improvements while maintaining low-complexity encoder and decoder operations. This paper describes the transform, quantization and entropy coding design in AV2, including redesigned transform kernels and data-driven transforms, expanded transform partitioning, and a mode & coefficient dependent transform signaling. AV2 introduces several new coding tools including Intra/Inter Secondary Transforms (IST), Trellis Coded Quantization (TCQ), Adaptive Transform Coding (ATC), Probability Adaptation Rate Adjustment (PARA), Forward Skip Coding (FSC), Cross Chroma Component Transforms (CCTX), Parity Hiding (PH) tools and improved lossless coding. These advances enable AV2 to deliver the highest quality video experience for video applications at a significantly reduced bitrate.
Paper Structure (44 sections, 29 equations, 13 figures, 14 tables)

This paper contains 44 sections, 29 equations, 13 figures, 14 tables.

Table of Contents

  1. Introduction
  2. Background on AV1 Transform Coding
  3. Transform Block Partitioning in AV1
  4. Primary Transforms in AV1
  5. Transform Type Signaling in AV1
  6. Quantization Design in AV1
  7. Entropy Coding Engine
  8. Coefficient Level Map and Sign Coding
  9. AV2 Transform and Entropy Coding Design
  10. Transform Partitioning Improvements
  11. Primary Transforms and AV2 Improvements
  12. Intra Mode Dependent Transform Set Derivation (MDTX)
  13. Intra/Inter Secondary Transforms
  14. Cross-chroma Component Transforms (CCTX)
  15. DC-based Transform Restriction (DCTX)
  16. ...and 29 more sections

Figures (13)

  • Figure 1: Mapping between $q_{\text{index}}$ and normalized QStep for AV1 (AC/DC) and AV2 extended quantization. AV1 QStep values are normalized by 8, while AV2 QStep values are normalized by 64 to account for the quantization precision.
  • Figure 2: Legacy AV1 transform partition types (NONE, SPLIT) and AV2 extensions (HORZ, VERT), (HORZ4, VERT4) and (HORZ5, VERT5).
  • Figure 3: An example list of transform identifiers depending on the set index, which shows the mapping from signaling index ($\textit{tx\_type}$) to transform candidates for different $m$, where $M$=39 in the AV2 design. The transform candidates $0$ and $3$ denote DCT_DCT and ADST_ADST, respectively.
  • Figure 4: Block diagram illustrating the operation of the intra/inter secondary transform (IST) and examples of predefined support regions.
  • Figure 5: TCQ with dual quantizers Q0 and Q1
  • ...and 8 more figures