Aomedia Announces Year-End Launch of Next-Gen Video Codec Av2
Posted4 months agoActive3 months ago
aomedia.orgTechstory
calmmixed
Debate
60/100
Video CodecsAv1Av2Video Compression
Key topics
Video Codecs
Av1
Av2
Video Compression
The Alliance for Open Media (AOMedia) announced the year-end launch of AV2, a next-generation video codec, sparking discussion on its potential adoption, competition with other codecs, and hardware support.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
51m
Peak period
62
Day 1
Avg / period
16.5
Comment distribution66 data points
Loading chart...
Based on 66 loaded comments
Key moments
- 01Story posted
Sep 15, 2025 at 1:44 PM EDT
4 months ago
Step 01 - 02First comment
Sep 15, 2025 at 2:35 PM EDT
51m after posting
Step 02 - 03Peak activity
62 comments in Day 1
Hottest window of the conversation
Step 03 - 04Latest activity
Sep 24, 2025 at 3:04 PM EDT
3 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
ID: 45252701Type: storyLast synced: 11/20/2025, 2:52:47 PM
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
~~And iPhones and Macs since the A15 / M3 chips~~
* https://en.wikipedia.org/wiki/AV1#Hardware_encoding_and_deco...
But to be honest, video codec have gotten so good these days, I just hope that we get a reasonably fast open-source encoder. AV1 and HVEC can already turn a 50Go bluray into an almost visually lossless 5~3Go file.
And H.267 ECM is miles ahead of AV2. There are not even in the same category as AOM has stated AV2 intends for 1080P and efficiency. While ECM at its current form is 50x to 100x slower than VVC encode, and 8 - 10x slower to decode. I am not even sure if it will ever be useable.
Yes:
> VVC is not alone in the video coding race. AV1, backed by AOMedia, has already gained traction, although its performance does not make it a direct competitor to VVC in high-end applications. The upcoming AV2, as well as AI-driven encoding techniques, could pose challenges to VVC’s success. Nevertheless, VVC’s strong technical foundation, industry support, and clear intellectual property structure position it as a promising long-term solution for video coding.
* https://www.nokia.com/blog/the-future-of-video-compression-i...
> For businesses focused on reducing operational costs, this is a key point in the h.266 vs av1 debate. While the H.266/VVC codec offers powerful compression improvements over h.265, AV1—and eventually AV2—may be more attractive thanks to simpler licensing and long-term affordability.
* https://www.dacast.com/blog/h266-vvc-versatile-video-coding/
Not qualified to answer.
https://engineering.fb.com/2023/02/21/video-engineering/av1-...
While the space savings and quality improvements are good, the encoding speed is an order of magnitude slower than using h264/vp9. In the end the user experience of causing people to wait significantly longer for an AV1 encode wasn't worth the tradeoff. To fix the user experience problem, I still had to encode a h264 version anyway, which kinda defeats the point when it comes to space savings. You still get data transfer improvements, but the break even point for when the encoding costs offset the data transfer costs were around 1000 views per min of video encoded, and as an average I'm far below that.
IMO there's a reason why YouTube only encodes AV1 for certain videos - I suspect it's based off of a view count. Past that point they trigger a AV1 encode, but it isn't worth it to do all videos, at least right now.
Worth keeping in mind I was looking at this ~2 years ago, so things may have evolved since then.
But how can they do that without storing the original uploaded video until it hits that view count?
Do they actually store the original uploaded video somewhere, but reencode for the edge servers to save data/storage?
YouTube has always stored the original video indefinitely. When they added 60FPS support, videos going back years were suddenly available with 60FPS without having to re-upload them. Not many people bothered to upload in 60FPS before YouTube supported it, but those that did noticed. (I know from Rooster Teeth/Achievement Hunter, which did 60FPS before YouTube supported it possibly because they also had their own platform in parallel.)
But for higher quality encoding, I personally found that h265/HVEC almost always beat it, with similar encoding time.
As for AV2, I just hope that we get a good open-source encoder.
See my blog: https://catskull.net/libaom-vs-svtav1-vs-rav1e-2025.html
For most “normie” use cases, I’d recommend cloudflares image transforms which are available on free tier. I actually wrote a small Jekyll plugin for my site to auto prefix images with their transform. Idk why but shipping optimized images is just one of those things that tickles me!
https://developers.cloudflare.com/images/transform-images/
Also, video encoding pretty much always comes with the tradeoff of more efficient = uses more processing power
But before it is widely used and accepted, here's AV2 for you to have compatibility issues with in the wild
With the ubiquity of h.264 and the patents expiring, will anyone but streamers care?
By the time h.265 encoding was trying to gain traction, h.264 encoding speeds were very fast. The image improvement was negligible with the main benefit being smaller file sizes. For the average user, the increased encoding times did not justify that. The switch from MPEG-2 to h.264 had very noticeable quality improvements so it did make it worth while for the slower encodes until h.264 was locked and key code included in CPUs. It was similar to the adoption rates of DVD from VHS compared to Blu-ray from DVD.
That's a contradiction because quality improvement and file size improvement are just two sides of the same coin. You can't have a large quality improvement at the same bit rate without having a large file size reduction at the same quality.
The average user is a consumer of media, not doing encoding themselves. A one time cost for higher encoding to save bandwidth / storage space many times over is almost always going to make some amount of sense.
The real issue here is just a standard chicken-and-egg problem. To use a new codec, you need it to be supported in end user devices. To get it to be supported, you need to show demand... for a thing that nobody can use yet.
Another reason to get a new video codec standard finalized sooner is that hardware implementation and deployment in mass market consumer SOCs is glacially slow. On the software side, encoder and decoder performance tends to improve meaningfully in the first few years as optimization occurs. And those running large media distribution platforms prefer at least 12-18 months to evaluate and implement a new codec.
Perhaps:
* https://en.wikipedia.org/wiki/Lindy_effect
If people know that AV2 is coming and competitive they may avoid adopting h266 and wait for the open alternative to ship.
h267 is still in development and due to be released in 2028. That's the actual competitor with AV2.
So I guess neither of these line up 1:1. I tend to see h265 and AV1 competing pretty hard right now so I tend to think of those as one generation and presumably h266 and AV2 will compete as the next generation.
AV1 and VP9/VP8 before that have, in contrast, been pretty much static after they were released. AV1 has had a single errata after it's release.
So I could see why you'd see H265 as the competitor. I mostly don't simply because I believe they explicitly stated that they were trying to be competitive with 266.
I personally prefer the way AOMedia is running things and I suspect hardware manufacturers do as well. No licenses and AOM is creating open source reference encoders/decoders. They are working very hard to make it easy for manufacturers to be able to pick up the spec and run with it. Keeping the stream standard static for a long period also means manufacturers don't have to worry that they won't get a new extension next year. Content encoders are also reasonably guaranteed that their encoding with today's software still works with yesteryear hardware.
MPEG, on the other hand, is paywalling the crap out of everything.
That is some major revisionist history.
AV1 was created to suck all the oxygen out of the room for h265, and hopefully be finished soon enough that most chipsets would implement AV1 instead of h265, because no one would want to pay the royalty fees. This would then be a major boon to Google, Facebook, Microsoft, Netflix and Amazon. Only Apple had vested interest in h265.
Instead, h265 was finished very fast and Apple almost immediately implemented it on their iPhone chipsets, with their end goal being much smaller video recording sizes at high resolution and FPS. Then Intel added h265 decode to Quick Sync. This was the nail in the coffin for AV1. AV1 is only now seeing some limited uptake.
Hopefully AV2 can have its bitstream / spec frozen in time and garner hardware support before h266 does, otherwise the AV codecs will be forever niche because the MPEG group will always have the lead in hardware decode.
But why hurry ? AV3 will come soon and it will be better. /s
You will always have to provide H.264 as baseline due to compatibility. And that has a cost of storage. Bandwidth cost have been declining fast, and with AI Capex it doesn't seems to be slowing down either. Meanwhile Storage cost hasn't drop, if anything recent trend suggest it may have plateau and went up for HDD.
H.264 1080P 2Mbps is good enough for a lot of things. Just like how MPEG-2 is still getting encoder improvement ~30 years later so is H.264 encoder.
There are other codec like LCEVC which you can apply on top of H.264 can provide up to 60% bitrate reduction for 4K content. This saves on storage cost and provide enough benefits.
It is only in streaming services like Netflix where the catalog of video are low enough they could afford to re-encode it every 5- 8 months and storage cost is minimal.
Again a new codec introduction is easily a 10 years task. Higher Speed PON is already being tested, while others are working on NGS-PON2 roll out. 5G Home Broadband with Massive MIMO. The true free and open Video Codec may not be AV1 or AV2, but H.264.
(Safari has a low market share but I have an above-average number of Mac / Safari users using my site)
Apple devices have had accelerated VP9 devices for literally years, which didn't stop them from simply not supporting VP9 during that time period, despite no reason (i.e. battery life) and despite that it has been patent unencumbered for that entire period.
And so much for battery life... If you are on a Wikimedia website on an iPad, it will use open codecs no matter what, and if you are on an Apple device, you might just get a software decoder in WebAssembly instead, if you're not on a new enough OS.
Apparently Ogg Opus and Ogg Vorbis has been supported since 18.4 (Mar 2025)[1]
I think it would be great to see more adoption of Opus in podcasting personally.
[1]: https://developer.apple.com/documentation/safari-release-not...
Apple is hopefully noticing that while major publishers can be swayed by this sort of behavior, lack of support for open codecs and open formats is going to make the iPhone and iPad web experience much worse for people who veer outside of the mainstream for a bit.
[0] https://en.wikipedia.org/wiki/Batterygate
That is 1.6B iPhone + iPad. I wouldn't say that is low market share. It still have 25-30% of devices.
Another great thing JXL has is lossless recompression of .jpg files, which is a smaller improvement than a whole new format, but much easier to deploy. Saving 22% beats saving 0%. Harder, of course, to see how that one would connect to any of AOMedia's other priorities.
A lot of good streaming hardware does jpeg-xs, ISO/IEC 21122, for real-time encoding basically from the camera to the mix station. It's still extremely high bit-rate mostly, but very low latencies. https://en.wikipedia.org/wiki/JPEG_XS