VP9
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
Why isn't this used more often?
because it takes forever to encode.
that also means you cant do live streaming with it
Because nothing has dedicated VP9 decoding chips yet aside from a few SoC's. Will waiting for at least 5 years or more for VP9 and also H.265 to become more mainstream. VP9 4K video would absolutely rape a 6700K.
>>54293421
Because CZ is better.
>>54293421
Tried converting that shit? Takes a million years compared to vp8.
>>54293421
Ask Chinese moot.
>>54293731
Yeah, that's why I'm asking. I converted a 2 minute h264 video to VP9 (~200kbps) and it looked acceptable at least. But when I tried to upload it to 4chan, it said that only VP8 is accepted. So I split the video into two parts, because quality-wise, VP8 is nowhere near VP9.
>>54293639
But I'm a watching VP9 4K YouTube video on my 6700K as we speak and it's doing just fine.
>>54293731
Why don't you support VP9, Hiroshimoot? WHY?
And also, audio. I need audio. Opus would be good.
>>54293715
No, VP9 is actually faster, though you have to use the proper encoding parameters, sacrificing a lot of potential quality, but it will still beat VP8 with equal encoding times.
>>54293950
VP9 has better superior for the same bitrate.
>open video page in browser
>long movie, want to skip to the part on my mind
>click around seek bar
>video is encoded in vp8
>seeks nearly instantly
>happy
>video is encoded in vp9
>seeking requires buffering the entire video up until that point, followed by long waits for the video to decode
>no happy
>>54294081
That's because it's not well-supported.
>>54294161
>Why isn't this used more often?
>because it doesn't work well
>Why doesn't it work well?
>because it's not well supported
>>54294198
That's like saying that people should not use airplanes because there are not enough people who know how to fly them.
>>54294223
>what is a catch 22
>>54294252
A book, a paradox named after the book, and a movie adaptation of the book.
Fuck off with your meme encoding. Every time YouTube is taking too much CPU is because I'm watching some video that doesn't use based H.264 and uses instead the retarded new Google codecs. Wonderful for battery life.
>>54293918
>youtube video
Reminder VP10 comes out this year.
Google is going to continuously release new iterations of these codecs and hardware manufacturers and consumers are never going to be able to keep up with them.
>>54294292
VP8/9 aren't bad but they don't/didn't have hardware support. Hardware manufacturers were afraid of implementing VP8 decoders because MPEG-LA would have sued the shit out of them. VP8 was like a ripoff of H.264/AVC's baseline profile or something and everyone knew it so no one wanted to risk it. Google caved and just paid the licensing fees for VP8/9. It should be okay now so eventually hardware decoders should start showing up but Google wants a new version of VPx to be released every 1 and a half years so good luck expecting hardware manufacturers or consumers to keep up.
>>54293421
Because the jap moot won't enable it on 4chan.
>>54294292
>Wonderful for battery life.
Yep, and that's why Edge won't have it enabled by default unless your chip supports VP9 hardware acceleration.
https://blogs.windows.com/msedgedev/2016/04/18/webm-vp9-and-opus-support-in-microsoft-edge/
>>54294611
All mts are faggots, it seems.
>>54294621
Is there any PC hardware that supports VP9 currently?
>>54294635
moots*
>>54294638
Probably only new graphics cards. Some of the newer phones might have decoders integrated by now.
>>54293421
>vp9
why. you can buy a glock for less and its better
I wonder if Apple is ever going to support VPx, Safari is now the only browser that doesn't support VP8 or VP9.
>>54294638
Skylake has partial support, IIRC.
GTX 950 & GTX 960 has VP9 hardware decoder
Pascal will have HEVC Main10 & VP9 hardware decoder top to bottom of the family of GPUs
>>54294582
fuck MPEG-LA I hope they burn in hell
>>54294479
Yup, this has been clear for years.
They abandoned the codecs they had out in some TV hardware already too meaning no updates to them, it was clear from then that VP* is just a meme to waste cash on stupid shit.
Meanwhile Apple and Microsoft have capitalised on h264+h265 HLS streaming with HW decoding easy & widely supported and chrome (and FF) don't support it leaving them unable to watch encrypted streams (like some sites I use rely on flash with chrome unless I use Edge or Safari).
>>54294691
They will eventually.
>>54294479
>>54294582
VP10 is currently being reincarnated as the baseline for this new "Alliance for Open Media" codec initiative that Google announced last year. By now, almost every major browser vendor and hardware manufacturer has joined this thing.
Everyone's tired of this MPEG-LA licensing bullshit.
>>54294689
If they had a physical product they were protecting I would understand what they were doing but audio/video codecs are mathematical algorithms. They never really "invented" these things so I don't see how they can claim ownership of them. Fucking patents on math.
>>54294691
>H264/HEVC
>widely supported
Yeah, keep telling yourself that.
>>54294771
Meant for >>54294479
>>54294799
H.264 is widely supported, H.265 not really
http://caniuse.com/#feat=mpeg4
>>54294853
>>54294799
Sorry, meant H265, not H264
>>54294662
I have Skylake, is there any way to make use of it? Is there even an API available for it?
>>54294691
I think everyone should just agree on VP9 and ignore future memes, I mean it's what Youtube uses for the best quality videos, that never happened with VP8. I just want something that's better than H264 while simultaneously being free.
>>54295336
It's probably better that everyone just agree to adopt every even or odd numbered release instead. That will give hardware manufacturers and consumers approximately 3 years between each new version. VP9 wont be great forever. While it's possible to skip over VP10 it might not be as easy to skip over VP11 and it will only get more and more difficult to ignore new codecs with each new generation.
How do we convince Hiroshima to add VP9 and Opus to 4chan?
Reasons depend, but fansubs for example won'T touch it until they fix the terrible worse-than-nonexistent rate control libpvx has.
libvpx(vp9): https://0x0.st/ZWu.png
x264: https://0x0.st/ZWS.png
x265: https://0x0.st/ZWQ.png
>>54295977
nani the fuck, it surely can't be that bad?
>>54293421
It's slow as shit and not necessarily better than x264
>>54295977
x264 seems to be the least bad there.
>>54293639
I'm watching 4K VP9 just fine on a 4770. It's really efficient to decode, close to H.264. It's just encoding that's slow.
Almost all youtube videos are delivered in VP9 where possible.
>>54294479
>Google is going to continuously release new iterations of these codecs and hardware manufacturers and consumers are never going to be able to keep up with them.
Sounds like all video technology ever
>>54296041
>It's slow as shit
If you mean it's slow to encode, that's true, but only because hardware manufacturers are not making hardware that fully supports VP9.
>Not necessarily better than x264
Oh, it is! Just try it yourself. Encode a video with x264, then with VP9 at the same bitrate. You'll see the difference.
>>54295977
>>54296058
No surprise there. x264 is the most mature of those by far.
>>54296089
>Encode a video with x264, then with VP9 at the same bitrate. You'll see the difference.
>>54295977
>>54296030
>>54296093
>>54296112
Basically those pics illustrate it isn't well tuned in allocating the bits (its actual compression techniques are close to VP9 or rather copied over from it, so it shouldn't be that bad).
In that case it is a scenario with very little bitrate which brings out the inefficiency more than real use would, but it does show its rate control is shit which sadly applies broadly. Sadly they are building VP10 and AOM AV1 on the same codebase (which has been accumulating half-functioning cruft for 20 years, they reportedly keep hacking it from one codec generation to next in On2).
>>54296253
>(its actual compression techniques are close to VP9 or rather copied over from it
Close to HEVC I meant.
>>54293421
Because VP10 and AOM AV1 are coming soon and improving quite a bit on VP9
>>54293639
A lot of stuff has SIP decoders for VP9 nowadays, specially SoC's
>VP9 4K video would absolutely rape a 6700K
My i5-2520M can decode 4K@30p at an average of 80 fps
VP9 is much faster than H265, you're thinking of those weird low bitrate H265 clips that rape everything
>>54296253
That's great when you're coding keyframes and still images, feel free to use VP9 there if you want to.
But for movies, I'll stick with x264 thanks.
>>54294656
enjoy your fragmented hands, /k/omrade
>>54295301
Read the Skylake instruction set document.
Write software that takes advantage of the instruction sets in assembly.
>>54296369
>Because VP10 and AOM AV1 are coming soon and improving quite a bit on VP9
They are at least 1 year away from freezing the bitstream (IMHO, delay is likely). But after that, you will still have to wait for properly implemented encoder. VP9 encoder has to this day grave flaws (rate control, limited threading), so I don't have high hopes for quick AV1 spread. 4 years from now to usability is my guess. The actual compression benefit compared to x265 I don't dare to predict, but I'm sceptical of it being close to revolutionary.
