<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Anna Gladkova on Medium]]></title>
        <description><![CDATA[Stories by Anna Gladkova on Medium]]></description>
        <link>https://medium.com/@anna.gladkova?source=rss-7677d9ce97f4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*eb03BxL27Y7eQg4G.jpg</url>
            <title>Stories by Anna Gladkova on Medium</title>
            <link>https://medium.com/@anna.gladkova?source=rss-7677d9ce97f4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 16:37:10 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@anna.gladkova/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[­­ Partitioning in Video Codecs]]></title>
            <link>https://medium.com/@anna.gladkova/partitioning-in-video-codecs-64b6941095ce?source=rss-7677d9ce97f4------2</link>
            <guid isPermaLink="false">https://medium.com/p/64b6941095ce</guid>
            <category><![CDATA[video-codec]]></category>
            <category><![CDATA[vvc]]></category>
            <category><![CDATA[partitioning]]></category>
            <category><![CDATA[hevc]]></category>
            <dc:creator><![CDATA[Anna Gladkova]]></dc:creator>
            <pubDate>Tue, 01 Sep 2020 14:33:24 GMT</pubDate>
            <atom:updated>2020-09-01T14:33:24.254Z</atom:updated>
            <content:encoded><![CDATA[<p>Fast broadband and an ever-increasing bandwidth are attributes that we often take for granted in the contemporary world. In doing so, we perhaps fail to appreciate the transformative change that is taking place around us in the information technology space. We are communicating with each other using the electronic mode at an ever-increasing frequency and pace. A lot of that communication is based on video files capturing snippets of our day-to-day life, be it some funny moments involving the pet dog or a momentous speech made by a national leader. Being trapped in this gigantic maze often referred to as life, we often lose sight of the specifics. How do we actually record, compress, and transfer video content across the World Wide Web and multiple devices? In this article, we embark on a journey to uncover some of the technical secrets associated with video files. In particular, we describe some of the key advancements in the video codecs literature associated with partitioning. In the first section, we compare Versatile Video Coding (VVC) to its predecessor High Efficiency Video Coding (HEVC). We discuss their relative advantages concerning partitioning. In the second section, we compare VP9, an open access video coding format developed by Google, with AOMedia Video 1 (AV1), an emerging open access internet video transmission tool developed by the Alliance for Open Media (AOMedia).</p><p><strong>High Efficiency Video Coding (HEVC) vs Versatile Video Coding (VVC)</strong></p><p>VVC’s picture structure adopted a concept from its predecessor, HEVC, wherein a picture is divided into a sequence of coding tree units (CTUs). One CTU consists of an N×N block of luma samples together with two corresponding blocks of chroma samples. The maximum permitted VVC block size is 128×128. This is in contrast to HEVC which supports a block size of 64×64. However, both VVC and HEVC support resolutions up to 8192×4320.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Qn6x9KjZjUxz25HpNkbSBQ.png" /><figcaption><strong>Figure 1 — Example of a picture partitioned into CTUs</strong></figcaption></figure><p>A picture is divided into one or more tile rows and one or more tile columns.</p><p>· A tile is a sequence of CTUs that covers the rectangular region of a picture.</p><p>· A slice consists of a whole number of complete tiles, or of a whole number of consecutive and complete CTU rows within a tile of a picture.</p><p>There are 2 modes of slices</p><p>· raster-scan slice: a slice that contains a sequence of complete tiles in a tile raster scan of a picture. <strong>Figure 2</strong> shows an example of raster-scan slice partitioning of a picture, where the picture is divided into 40 tiles (8 tile columns × 5 tile rows) and 3 raster-scan slices.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/429/1*sMD9-BnE632x8poHTFYGfQ.png" /></figure><p><strong>Figure 2 — Example of a picture partitioned into tiles and raster-scan slices</strong></p><p>● rectangular slice:</p><p>o a slice that contains either a number of complete tiles that collectively form a rectangular region of the picture</p><p>…or</p><p>o a number of consecutive complete CTU rows of one tile that collectively form a rectangular region of the picture.</p><p>Tiles within a rectangular slice are scanned in tile raster scan order within the rectangular region corresponding to that slice.</p><p><strong>Figure 3</strong> shows an example of rectangular slice partitioning of a picture, where the picture is divided into 40 tiles (8 tile columns × 5 tile rows) and 6 rectangular slices (size in units of tiles 3x3)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e2alWlH0npNP4ZOsL72Nhg.png" /></figure><p><strong>Figure 3 — Example of a picture partitioned into tiles and rectangular slices</strong></p><p><strong>Figure 4</strong> shows an example of rectangular slice partitioning of a picture with several slices in one tile, where the picture is divided into 12 tiles (3 tile columns × 4 tile rows) and 16 rectangular slices.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/499/1*9XXmrHsQsg0lNRrdDOT6ew.png" /></figure><p><strong>Figure 4 — Example of several slices in one tile</strong></p><p>The main difference between HEVC and VVC in picture partitioning is that the latter is endowed with a new structure called subpicture. Subpicture is developed under the influence of VR, which is currently in vogue. Since the subpictures are completely independent, users could decode only a part of the picture instead of the entire image. Another advantage of subpicture’s independence is that they could be encoded/decoded in parallel. Note that the same applies to slices and tiles as well as they are also independent. They could be encoded/decoded in parallel. A key difference however is that the partitioning of a decoded frame (or a reference frame) into tiles and slices become redundant whereas partitioning of a reference frame into subpictures could bolster the inter prediction process by treating the subpicture boundaries (optionally) as picture boundaries.</p><p>Note that subpicture is a rectangular region of one or more slices within a picture. It is crucial to appreciate how the boundaries of tiles, slices and subpictures are related. Each subpicture boundary is also a slice boundary, and every vertical subpicture boundary is always a vertical boundary of some tile.</p><p>As we consider dependence of subpictures and tiles, at least one of the following conditions perpetually hold:</p><p>· All subpicture’s CTUs belong to the same tile;</p><p>· All tile’s CTUs belong to the same subpicture.</p><p><strong>Figure 5</strong> shows an example of rectangular slice partitioning of a picture with several slices in one tile, where the picture is divided into 6 tiles (2 tile columns × 2 tile rows), 6 rectangular slices and 4 subpictures</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/674/1*v5mGupvMBrTq6O3xqON0-A.png" /><figcaption><strong>Figure 5 — Example of tiles, slices and subpictures combination</strong></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/174/1*gQLO2GHUgbxp9pkkecyAKw.png" /></figure><p>Partitioning of the CTUs uses a multi-type tree structure. A quadtree, with nested multi-type tree using binary and ternary splits segmentation structure, replaces the old HEVC concepts of multiple partition unit types<em>. </em>That’s to say, it removes the separation of the CU, PU and TU concepts (except as needed for CUs) that are too large for the maximum transform length, and it supports more flexibility for CU partition shapes.</p><p>In the coding tree structure, a CU can have either a square or a rectangular shape. A coding tree unit (CTU) is first partitioned by a quaternary tree (a.k.a. quadtree) structure.</p><p>Then, the quaternary tree leaf nodes can be further partitioned by a multi-type tree structure. There are four splitting types in the multi-type tree structure,</p><p>· vertical binary splitting (SPLIT_BT_VER)</p><p>· horizontal binary splitting (SPLIT_BT_HOR)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/530/1*EBbpa8PG1a4I13XauXEYpQ.png" /></figure><p>· vertical ternary splitting (SPLIT_TT_VER)</p><p>· horizontal ternary splitting (SPLIT_TT_HOR)</p><p>The multi-type tree leaf nodes are called coding units (CUs). This segmentation is used for prediction and transform processing without any further partitioning. This implies that, in most cases, the CU, PU and TU have the same block size in the quadtree with nested multi-type tree coding block structure. The exception is when the maximum supported transform length is smaller than the width or height of the colour component of the CU.</p><p><strong>Figure 6</strong> shows a CTU divided into multiple CUs, with a quadtree and nested multi-type tree coding block structure. The quadtree with nested multi-type tree partition provides a content-adaptive coding tree structure comprised of CUs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/484/1*OndOYw3PE_RZxOq1dsT5sw.png" /></figure><p><strong>Figure 6– Example of a quadtree with nested multi-type tree coding block structure</strong></p><p>A CU may be as large as the CTU or as small as 4×4. In the case of the 4:2:0 chroma format, the maximum chroma CB size is 64×64 and the minimum size chroma CB consist of 16 chroma samples.</p><p>In VVC, the maximum supported luma transform size is 64×64 and the maximum supported chroma transform size is 32×32. When the width or height of the CB is larger than the maximum transform width or height, the CB is automatically split in the horizontal and/or vertical direction to meet the transform size restriction in that direction.</p><p>Luma and chroma may have a separate block tree structure for I slices, but not for P and B slices. Block tree structure for the P and B slices will remain the same.</p><p><strong>VP9 vs AV1</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/222/1*YJPPuhugwEoV-a9sV6e-4g.png" /></figure><p>VP9, as a predecessor of AV1, is limited to 4 partition types. The maximum block size is 64x64 and the minimum is 4x4. Each square block can be split recursively down to 4x4 level (the lowest level).</p><p>AV1 expands the maximum permitted block size to 128×128 (superblock), and also expands the partition-tree (significantly) into a 10-way structure. AV1 introduced ¼ rectangular partitions that did not exist in VP9. Rectangular partitions cannot be further recursively subdivided, in contrast to squares. In addition, AV1 adds more flexibility to the use of any partitions below the 8x8 level, making 2x2 chroma inter prediction possible. Figures 7–10 presents examples of partitioning in VP9 and AV1.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/394/1*zckeInXetzcXumXaDV0nQA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/218/1*wMIc7ec79--4TDpaDi-MUQ.png" /><figcaption>Fig. 7 — Partition Types in AV1, as from VP9</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/399/1*v6o9UcqU00RUVa7dpLeC1Q.png" /><figcaption><strong>Fig. 8 — New Partition Types (Horizontal and Vertical A and B)</strong></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/204/1*KEm1nfm_RDhIWXq9qCxZKg.png" /><figcaption><strong>Fig. 9 — New Partition Types (Horizontal and Vertical 4)</strong></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/324/1*TYeVF-ZC7hnWSDX0LGoLJQ.png" /><figcaption><strong>Fig 10 — an example of AV1 block partitioning</strong></figcaption></figure><p>Partitioning is fundamental in the work of the main products <a href="https://vicuesoft.com/products/analyzer">VQ Analyzer </a>and <a href="https://vicuesoft.com/products/probe">VQ Probe</a> by <a href="https://vicuesoft.com/index">ViCueSoft</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=64b6941095ce" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>