66

2 years too early

One of the differences is the size of the video. H.266 is 50% smaller size with the same quality. For example one fullhd movie will be 700mb instead of 1.4gb without any loss of quality. I love compression. One day big video files will be very small size with amazing quality. Better quality at lower sizes, baby!

Only 1.8x decoder time increase, it's not that much

@@morfx9911 1.8 that is nearly 2 time longer, nearly double the decode time, nearly double the power draw. if 1x equal to 30 seconds then 1.8 equal to 54 seconds. Cpu, soc advertises as mostly at 20% faster than older gen rarely at 50% faster never 80% fater and at most 50% less power draw. Here 1.8 time increase decode time is 80% slower, %80 more power draw. So all the cpu, soc advancement become null. You may not need to read what is below. Just a example: 2chips of 2 generations, 1st gen chip draw 1W for 1 second. the 2nd gen chip is 20% faster, draw 50% less power than its older gen chip. 1 clip h.265 and that same clip in h.266, 1st gen would need 1 second to decode h.265 clip draw 1W and 1.8 second draw 1.8W to decode h.266 clip 2nd gen chip would need 1.6 second, 0.8W to decode h.266 clip, 2.25 time battery time, 0.8 seconds, 0.4W for h.265 clip, 4.5 time longer battery time. So with example of 20% processing increase and 50% less power draw. If both machine has same battery and 1st gen machine can stay 20hours if decode h.265, then 2nd gen machine can stay 90 hours (3day 18H), but if 2nd gen machine decode h.266 it would only last 45H (1day 21H). 3day 18H sure is more useful than 1day 18H.

Вопрос актуален

Sofware decoding is kinda ok

@@kimisaacbuelagala1314 if you're willing to wait few hours to play 1 sec of a AV1 video that is

@@0wynlol my 3 year old phone can play 720p on mx player

Not so much about the consumer. On the Scale of Netflix, 40% bandwidth saving on 4K could be tens of million of saving. And that is already with their appliance sitting locally with ISP.

@@ksec6631 But at same time waste billions on consumer side for bad access time. It does not matter for consumer, one consumer does not pay for other million consumer, one consumer pays only for his traffic, which is fraction of cent in hour. But pays for waste of time 0,2-50 dollars depending on his wage and amount of rewind.

Brilliant work!!! Thanks for sharing.

@@lastmanustanding What is brilliant work? This codec is has very heavy complexity and therefore very bad access time.

On a much larger scale? Yes. Billions of devices connected and streaming video content and everyone is reducing 40% on data bandwith? You dont need math to understand that the cost will be tremendously cut

They only show that when the thing it's standardized

If you mean the orthonormal transforms I read somewhere they were dropped because they were brought in too late and hardware would have to redesign a lot risking potential delays. Maybe they will make it for AV2.

​@@Ashtree81 I don't know if this is an orthonormal transform, but it's a special kind of lifting scheme with a little post-adjustments to make it resemble a DCT, & it only used adds subs & small shifts to achieve that - & it wasn't brought in late, it was there from rather early on (it's already described in Monthy's first daala demo page) but it got dropped late in development. Also I see no way how this could be difficult to implement in hardware because it's fundamentally simpler than a DCT on so many levels: people.xiph.org/~xiphmont/demo/daala/demo1.shtml

@@FrankHarwald Ok, yeah its not the same as the ortonormal ones so dont know why it was dropped.

@@MileHighVideo Great!