When I published When Metrics Mislead: Evaluating AI-Based Video Codecs Beyond VMAF earlier this year, the takeaway was blunt: traditional video quality metrics like PSNR, MS-SSIM, and VMAF simply can’t be trusted for AI-based codecs. My tests with Deep Render, and supporting research from JPEG AI and Microsoft’s MLCVQA project, showed that those metrics consistently underrate the perceived quality of …

InterDigital’s acquisition of Deep Render marks one of the first instances where a traditional codec IP powerhouse has invested in AI-native compression. Deep Render was built from a clean slate, with an architecture that doesn’t trace its roots to the block-based model that has defined every standard codec for the last 30 years. At some point, that old architecture will …

Three recent Alliance for Open Media presentations on YouTube shed new light on AV2’s performance and utility. Andrew Norkin, Director of Codec Development at Netflix, presented the current status and architecture of AV2. He outlined the codec’s design goals, early performance results, and hardware-focused development approach, noting that the low-level toolset is now finalized. The YouTube video is here. Ryan …

I recently visited Florida Atlantic University’s Multimedia Lab to record the first real-time demonstration of Feature Coding for Machines (FCM), a new approach to video compression designed for AI and edge-based applications. The system, developed by OP Solutions in collaboration with FAU, demonstrates low-power, efficient operation. You can watch a video of the demonstration and its implications here, and it’s …

I recently interviewed Dominic Sunnebo, Commercial Director at Worldpanel by Numerator, for Streaming Media. We discussed new data from the Q2 2025 Entertainment on Demand report. Topics covered include: Sports driving nearly 1 in 4 new streaming subscriptions Why sports are a rented audience—loyal to teams, not platforms The rapid rise of women’s sports and its impact on rights deals …

Every streaming service runs on a business model which shapes everything from content acquisition to monetization. This PDF walks through the core models every professional should know: SVOD, AVOD, TVOD, FAST, vMVPD, and Hybrid. Each slide explains how the model works and highlights a key insight into its strengths and challenges. For a deeper look at how these models combine …

As you may have seen, I’ve been spending a lot of time analyzing multiview solutions and studying the technical alternatives, client-side vs. server-side (see here for an extensive analysis of the two). Briefly, client-side multiview uses existing encoded streams assembled into the multiview by the client, while server-side essentially creates a new multiview channel on the server, necessitating a new …

Multiview, or the ability to view multiple live feeds simultaneously, is quickly becoming a must-have feature for sports and live-event streaming. The core technical question for providers is how to implement it.  There are two main architectural options: client-side and server-side multiview. Both can display multiple games or camera angles on a screen, but they work in very different ways, …

I recently sat down with Marc Todd, CEO of Skreens, to discuss how multiview has evolved from a niche feature into a must-have capability for sports and live-event streaming. Skreens has been at the center of this transformation, powering multiview deployments for millions of subscribers. What follows is a slide-by-slide look at Todd’s presentation, told mostly in his own words. …

Traditional and streaming content and service providers, MVPDs, and other rights holders that dismiss multiview as an unnecessary expense are leaving revenue on the table. Wherever it has been deployed, multiview has delighted customers, driven acquisition, and reduced churn. Just as important, it adds new ad inventory and formats, creating monetization opportunities that weren’t possible in a single-screen world. The …