Live data from Hacker News

Dav2d

jbkempf.com

211–212 of 212 posts

Re: Dav2d

#211

This seems like an interesting case to test AI agents on. Like we had weird examples like C compilers and Bun. This is a much more interesting example because its highly nontrivial. AV1 exists, Dav1d exists. Lets see AI take the AV2 spec and Dav1d code and try to make a working high performance AV2 decoder.

> Lets see AI take the AV2 spec and Dav1d code and try to make a working high performance AV2 decoder. That sounds like one of these high-risk, high-reward things that are great for people / projects / companies who have nothing to lose, but is not a great baseline strategy for an established market player. AV2 is here with support from aomedia and its members. AV2 will be used, and we need a production-grade decoder…

What's the high risk? That you end up with useless code?

I didn't mean that the Dav1d people should yolo vibe code Dav2d. My point was this this is a very interesting possible experiment since there is no existing Dav2d contamination in the training data.

Re: Dav2d

#212
>The reported gains vary depending on the test conditions, but improvements around 25% compared to AV1 are commonly seen

>AV2 decoding is roughly five times more complex than AV1 decoding.

For the past 8 to 10 years VideoLan are the only guys on the AOM camp who doesn't care about optics and willing to spell out the truth.

>because we did not believe hardware decoding would become available quickly enough, or on enough devices.

The we was basically VideoLAN only. All other members ( at least on the AOM ) thought hardware decoder would be ready by 2019. And we all know how that works out.

Post reply on HN