--- title: "The Future of Video Codecs: AV1, AV2, VVC, and Why Modern Codecs Are Collections of Tools" author: "Cutsio Team" date: "2026-05-14" lastmod: "2026-05-14" category: "Video Technology" excerpt: "Modern codecs like AV1 and VVC are not single algorithms — they are collections of dozens of coding tools that adapt to the content being compressed. This increasing complexity is reshaping how video is encoded, decoded, and delivered, and it matters for every editor and content creator." tags: ["AV1", "VVC", "Codecs", "Video Compression", "FFmpeg", "dav1d"] --- ## Why are modern codecs described as collections of tools instead of single algorithms? Modern codecs like AV1 and VVC are described as collections of tools because they contain dozens of specialized coding techniques that can be selectively applied depending on the type of content being compressed — screen sharing, talking heads, animation, live sports, and cinematic footage all benefit from different combinations of tools. Earlier codecs like MPEG-2 and even H.264 had a relatively fixed set of coding tools. The encoder made a limited number of decisions: block size, motion vector, quantization level. The tools were general enough to handle most content reasonably well, but they were not optimized for any specific type. Modern codecs are different. AV1 includes tools specifically designed for screen content (like palette mode for flat-color regions), tools for film grain synthesis (which preserves the look of film without encoding actual grain), tools for different block partitioning schemes, multiple transform types, and advanced loop filters. Each tool provides a few percent improvement in compression efficiency for specific types of content. "The thing to understand about modern codecs like AV1 and VVC is that they are actually not codecs," the developers explain. "They are a collection of tools. There are multiple tools, multiple codecs in the same codec, depending on the image, to get more compression." This means that an encoder needs to decide in real time which tools to apply to each region of each frame. A screen share containing mostly text and flat colors uses different tools than a cinematic scene with complex lighting and motion. The encoder must evaluate the content and choose the optimal tool set — a computationally expensive optimization problem. ## How does dav1d prove that software decoding still matters? dav1d proves that software decoding still matters because it demonstrates that a clean-slate, hand-optimized software decoder can decode AV1 efficiently enough for real-time playback on commodity hardware — contradicting the industry assumption that AV1's complexity required hardware decoding. When the Alliance for Open Media designed AV1, the conventional wisdom was that the codec was too complex for software decoding. The standard was so feature-rich that decoding it in software on general-purpose CPUs would be too slow. Hardware decoders would be necessary. The VideoLAN project responded by building dav1d from scratch. The project was designed with a single goal: create the fastest possible software AV1 decoder using every optimization technique available. The result proved the conventional wisdom wrong. dav1d uses aggressive low-level optimizations that go beyond what most software projects attempt. It uses custom calling conventions — the functions do not follow the standard ABI because the standard calling convention adds overhead. It reorders instructions to maximize pipeline utilization. It uses platform-specific SIMD for every critical function. It takes processor detection so seriously that it identifies the specific microarchitecture and adjusts its code paths accordingly. "I have never heard of any other project than dav1d doing that," Kieran notes about the custom calling conventions. The approach worked. dav1d became the reference AV1 decoder and is now used by most major platforms, including VLC, FFmpeg, and many streaming services. ## What is the compression improvement from generation to generation? Each new codec generation improves compression by approximately 30% over the previous generation — H.265 is 30% better than H.264, AV1 is 30% better than H.265, and AV2 is targeting 30% better than AV1 — but each generation requires dramatically more computational power to encode. The 30% improvement means that for the same perceived quality, the bitrate can be reduced by 30%. A video that required 10 Mbps with H.264 needs about 7 Mbps with H.265 and about 5 Mbps with AV1. For streaming services like Netflix and YouTube, this translates directly into reduced bandwidth costs and better quality for users with slower connections. The catch is that the encoding complexity grows much faster than the compression improvement. AV1 encoding can be 100 to 1,000 times more computationally expensive than H.264 encoding. This is not a problem for large-scale cloud encoding where the cost is absorbed by the service provider, but it limits the adoption of new codecs for real-time encoding and consumer applications. "The encoding is two orders of magnitude more complex," the developers explain. "So you need to be able to leverage the CPU and sometimes GPU, like you mentioned." | Codec | Year | Compression vs Previous | Encode Complexity | |---|---|---|---| | H.264 / AVC | 2003 | Baseline reference | 1x | | H.265 / HEVC | 2013 | ~30% better than H.264 | ~10x H.264 | | AV1 | 2018 | ~30% better than H.265 | ~100x H.264 | | AV2 | 2024-2025 | ~30% better than AV1 | ~500x H.264 | | VVC / H.266 | 2020 | ~30% better than H.265 | ~200x H.264 | ## What is the difference between AV1, AV2, and VVC (H.266)? AV1 and AV2 are developed by the Alliance for Open Media, an industry consortium including Google, Netflix, Amazon, and Mozilla, using a royalty-free licensing model. VVC (Versatile Video Coding, also called H.266) is developed by the MPEG standards body and uses a patent-pool licensing model. The licensing difference is the most significant practical distinction. AV1 and AV2 are designed to be royalty-free — any company can implement and use them without paying license fees. VVC uses a patent pool, similar to the model used for H.264 and H.265, where implementers pay royalties to the patent holders. For streaming services, the royalty-free model of AV1 is attractive because it eliminates a significant ongoing cost. For broadcast and professional video, VVC may offer better tooling and ecosystem support from established vendors. "The main codecs for transmission are H.264, H.265, H.266 on one side, and AV1, AV2 on the other," JB explains. "There is a difference in the standardization process — in AV1 and AV2, patents are not even discussed, whereas VVC uses a traditional patent pool." Both ecosystems will coexist for the foreseeable future. AV1 and AV2 are dominant in internet streaming. VVC is more common in broadcast and professional applications. ## How should editors prepare for the next generation of codecs? Editors should prepare for the next generation of codecs by developing a workflow that separates mastering from delivery — master in a high-quality intermediate codec and deliver in whatever format the platform requires, which allows you to adopt new delivery codecs without changing your editing pipeline. The most important principle is to never edit in your delivery codec. Edit in ProRes, DNxHR, or another editing-friendly format. Create your master file in that format. Then transcode to H.264, AV1, or whatever the target platform requires. This decouples your editing workflow from the delivery format, so when AV1 becomes the standard for YouTube, you do not need to change how you edit. The second principle is to understand the hardware you are targeting. If your audience primarily watches on recent smartphones, AV1 hardware decoding is widely supported and provides excellent quality at low bitrates. If your audience includes users with older devices, H.264 remains the safest fallback. The third principle is to stay informed about codec developments without overreacting to announcements. AV2 will be incrementally better than AV1, just as AV1 was incrementally better than VP9. The fundamentals of good video production — good lighting, good audio, good storytelling — matter far more than the specific codec used for delivery.

Future-proof your workflow. Master in quality, deliver for any platform.

Cutsio supports this philosophy by handling format-agnostic pre-processing. Upload your footage in any format, remove silences with AI, generate transcripts, and export clean XML to your NLE. Your master stays high quality. Your delivery adapts to the platform.

Try Cutsio Free

No credit card required. 60 minutes of free processing.

## FAQ **Is AV1 better than H.264?** AV1 provides approximately 50% better compression than H.264 at the same quality level. However, AV1 encoding is significantly more computationally expensive, and older devices may not have hardware decoding support. **Do I need to use AV1 for my YouTube videos?** YouTube accepts and re-encodes whatever you upload. If you upload H.264, YouTube will transcode to its preferred formats. Uploading AV1 may reduce your upload time at the same quality, but the final playback quality is determined by YouTube's encoding pipeline. **What is dav1d and why is it important?** dav1d is a hand-optimized software AV1 decoder developed by the VideoLAN project. It proved that AV1 could be decoded efficiently in software, contradicting the assumption that hardware decoding was necessary. **Will AV2 replace AV1?** AV2 will eventually replace AV1, but the transition will take years. Hardware support for AV1 is still being deployed, and AV2 hardware decoders will not be widely available until the late 2020s. **What codec should I use for archiving my projects?** For archiving, use a visually lossless codec like ProRes 4444, DNxHR HQ, or FFV1. These preserve maximum quality for future re-encoding and are independent of the delivery format landscape.