While the Alliance for Open Media (AOMedia) aims to finalize the AV2 specification by the end of 2025—with the standard still in draft form—VideoLAN developers have been quietly forging ahead. This weekend, they published dav2d, an open-source AV2 decoder. This move mirrors their earlier success with dav1d for AV1, signaling a strong commitment to next-generation video codec support. Below, we answer key questions about this development.
What Is Dav2d and Why Was It Created?
Dav2d is an open-source decoder for the upcoming AV2 video compression standard, developed by VideoLAN—the team behind VLC media player. Its primary goal is to provide a fast, efficient, and royalty-free software implementation that can decode AV2 streams in real time. VideoLAN created dav2d to ensure that open-source projects and users have access to high-performance AV2 decoding early in the standard's lifecycle. By publishing the code while AV2 is still in draft status, they give developers a head start on integration, testing, and optimization. This proactive approach helped dav1d become the gold standard for AV1 playback, and the team aims to replicate that success with dav2d.
Why Is Dav2d Published Before the AV2 Spec Is Final?
Although AOMedia originally targeted a finalized AV2 specification by late 2025, the standard remains a draft. Publishing dav2d now offers several advantages: it allows developers to start experimenting with AV2 decoding, identify edge cases, and provide feedback to the specification authors. VideoLAN’s early work on dav2d also helps ensure that once the spec is finalized, a mature, optimized decoder will be available immediately. This strategy reduces the usual lag between standardization and real-world adoption, as seen with dav1d, which was widely deployed soon after AV1’s release. The open-source community benefits from early access, and VideoLAN gains valuable field testing before the codec becomes mainstream.
Who Is Behind the Dav2d Project?
Dav2d is developed by VideoLAN, the nonprofit organization renowned for creating the VLC media player and the earlier dav1d AV1 decoder. The project is led by VideoLAN’s core developers, many of whom also contributed to dav1d and other multimedia libraries like libavcodec (FFmpeg). Key contributors include Jean-Baptiste Kempf (president of VideoLAN) and a team of volunteers and sponsored engineers. While the initial code was published by VideoLAN, they welcome contributions from the broader open-source community. The project is licensed under the permissive BSD 2-Clause License, encouraging commercial and academic use without restrictive copyright obligations.
How Does Dav2d Compare to Dav1d?
Dav2d is conceptually a successor to dav1d, reusing its architecture and optimization techniques but adapted for AV2’s more complex codec tools. Both prioritize high performance and low latency for real-time playback, with heavy use of SIMD instructions (e.g., AVX-512, Neon) and multi-threading. However, AV2 introduces new coding tools like NeRSC (neural network–based intra prediction) and enhanced transform blocks. Dav2d must implement these efficiently while maintaining the same lean, portable codebase philosophy. Early benchmarks suggest dav2d achieves decoding speeds comparable to dav1d’s initial releases, with room for optimization as the spec stabilizes. VideoLAN plans to maintain both projects in parallel until AV2 fully replaces AV1 in consumer applications.
What Is the Current Status of the AV2 Specification?
The AV2 specification is still in draft status as of early 2025, despite AOMedia’s earlier goal of finalization by the end of 2025. The codec is being designed as a royalty-free successor to AV1, targeting a 30–50% bitrate reduction for equivalent perceptual quality. New features include improved motion compensation, adaptive filtering, and neural-network–based modules (as mentioned). AOMedia continues to refine the draft based on submissions from member companies like Google, Amazon, Netflix, and Intel. VideoLAN’s early decoder work indirectly influences this process by revealing implementation challenges. While no official final date has been set, industry experts expect a completed specification by 2026 or later, depending on testing and convergence of features.
What Does Dav2d Mean for Open-Source Video?
Dav2d strengthens the open-source video ecosystem by ensuring that a high-quality decoder is available before the AV2 codec reaches mass adoption. This preemptive approach prevents proprietary decoders from dominating the market, as happened with earlier codecs. It also integrates seamlessly with existing open-source multimedia frameworks like FFmpeg, GStreamer, and VLC. The publication of dav2d under a permissive license encourages innovation in video applications, from streaming services to archival tools. Moreover, VideoLAN’s commitment to transparency means that anyone can inspect, modify, and redistribute the code, fostering trust and security. In the long run, dav2d will help make AV2 a truly open standard, reducing barriers for developers and end-users worldwide.