This book aims to bridge the increasing gap that has emerged between the classical principles of compression and the rapidly growing world of AI-driven and neural video coding. The acceleration ongoing in the area urgently requires an integrated reference that links theoretical underpinning with practical realities of modern media systems. This book intends to bring the different threads of royalty-bearing and royalty-free codec ecosystems together, set the compression research in perspective against real-life applications that span from large-scale streaming services to immersive VR/AR environments.
The need for such a book has never been greater. Video now constitutes more than 80% of global internet traffic, fuelled by an explosion of 4K/8K, HDR, VR, and AR content. Streaming platforms deliver billions of viewing hours each day, placing unprecedented pressure on network infrastructure. Without continual advances in compression, the global internet would struggle to sustain the accelerating demand. At the same time, the technology landscape has become fragmented and fast-moving: H.264 evolved into H.265, H.266/VVC, and onward to AV1, AV2, and emerging neural codecs, while industry continues to navigate conflicting ecosystems of patent-encumbered standards and royalty-free alternatives. These codecs also serve increasingly diverse use cases-streaming, broadcasting, video conferencing, immersive media, autonomous systems-yet no single resource explains how these technologies relate, compare, and complement one another.
The main challenges that come with this are: engineering teams struggle to pick codecs that are suitable, developers employ implementations without necessarily understanding the underlying theory behind it, and decision-makers make infrastructure decisions at scale without a comprehensive technical framework. Researchers often work in silos: traditional compression communities and AI-based communities rarely interact with each other, despite clear opportunities for synergy. These educational gaps are accentuated further: university curricula often stop at H.264 or introductory concepts, existing textbooks are either outdated or narrow, and professional training resources are often proprietary, expensive, or incomplete. In turn, this means that graduates and new engineers enter the field unprepared for the modern video technology stack.
This book addresses critical needs on both the technical and broader landscape. Standards professionals must have a coherent historical and technical context; open-source developers seek to understand codec internals in depth; the task of streaming engineers is to optimize end-to-end systems. Beyond the technical community, efficient compression has direct societal impact, including providing access to video in bandwidth-constrained regions, supporting sustainability through reducing energy consumption in data centers preserving digital heritage, and driving innovation across sectors. Existing literature suffers from several shortcomings: highly theoretical, too narrow, outdated, siloed by application, or entirely separate between traditional and AI-based approaches. Video Compression Theory and Practices cover everything from foundational theory to modern AI models, practical insights into implementation, and comparative analysis of recent standards introduced between 2020 and 2025. The book combines academic rigor with industrial pragmatism and shows how fifty years of compression research have informed today's hybrid and neural codecs.
Reka Sandaruwan Gallena Watthage
Video compression Video codecs Network-based codecs H.264 / AVC HEVC / H.265 VVC / H.266 AV1 Volumetric video compression End-to-end learned video compression Rate-distortion optimization Transform coding (DCT/DST) Motion estimation and compensation Entropy coding (CABAC) Deep learning for video