Announcement

Collapse
No announcement yet.

Netflix Talks Up SVT-AV1 Video Encode/Decode Effort

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • andreano
    replied
    Originally posted by OneTimeShot
    Hmm... the patent trolls are out to get VP9 and AV1.

    let’s hope Google can hit back hard with their video patent stack and Android support.
    Originally posted by Toggleton View Post
    Hope it will be Fun to watch once the Alliance for open media does fight against these claims
    What now? If a large percentage of those patents are invalid, it's still a thousand. This is quite devastating for royalty-free video. Is it back to VP8 now?

    Also, they forgot to have a software exemption, which is akin to declaring war against Mozilla. I hope at least this can be negotiated.

    Here is what I think:


    To protect against bad surprises in the future, AV2 should have bitstream switchable features, like XVC or EVC.

    Leave a comment:


  • pal666
    replied
    Originally posted by nuetzel View Post
    If avx2 or avx512 is a must, it sucks.
    Sure the 'other' implementation works with ssse3.
    Intel crippled our systems right enough...
    not every intel cpu supports avx2. on the other hand my amd cpu does support it

    Leave a comment:


  • nuetzel
    replied
    If avx2 or avx512 is a must, it sucks.
    Sure the 'other' implementation works with ssse3.
    Intel crippled our systems right enough...

    Leave a comment:


  • CommunityMember
    replied
    Originally posted by profoundWHALE View Post
    hardware decoding needs to be there before we'll see adoption
    AV1 had, as one of it's goals, efficient software decoding (because, as you rightfully point out, hardware decoding is going to take quite time to get out to the masses). And while no where near as battery efficient as hardware assisted decoding, AV1 decoding performance and power utilization is already adequate for many purposes on mobile devices today (using dav1d with neon on arm devices for example), as long as you don't insist on 4K decoding at 120fps for hours upon hours.

    Leave a comment:


  • profoundWHALE
    replied
    hardware decoding needs to be there before we'll see adoption

    Leave a comment:


  • Toggleton
    replied
    Originally posted by shmerl View Post
    How does it compare to rav1e?
    Depends on your use case SVT-AV1 is build for server/PC with lot of RAM(last time i tested 6GB for 1080p, but that was a while ago could be better now) and needs avx2 or avx512 https://www.reddit.com/r/AV1/comment...ifferent_simd/

    Rav1e has more broad CPU support cause they use the assembly from dav1d(av1 decoder) avx2 ssse3, arm and uses ~2GB for 1080p.

    I think both projects still need work and they are in heavy development so quality and speed wise can things change quite a bit.

    My personal opinion is so far that aom (build from git) is quality wise, with https://github.com/master-of-zen/Av1an even speed wise quite competitive. Cause av1an will split the video and encode it in parallel.(for me a sweet spot is cpu-used 3 --end-usage=q --cq-level=30 ## 0–63. Lower values mean better quality and greater file size.)

    Quality wise(at a reasonable speed) is it hard to compare and there is no up to date test. This one is 1 month old https://www.reddit.com/r/AV1/comment...vp9_vs_svtvp9/
    Last edited by Toggleton; 03-15-2020, 09:56 AM.

    Leave a comment:


  • shmerl
    replied
    How does it compare to rav1e?

    Leave a comment:


  • Toggleton
    replied
    Originally posted by CommunityMember View Post

    While AV1 techniques have undergone review of possible patented technique infringement, A/V codecs have always been a patent minefield (some of the patents have covered broad fundamental functionality that it is hard to avoid getting close to). Since the current list of potentially applicable patents is not public, there is no way to know what the claims being asserted may be, nor what, if anything, will need to change for AV1 to be either non-infringing or freely implementable if the validity and applicability of those patents is sustained.
    Actually they have made the list public https://sisvel.com/licensing-program...atform/patents
    https://sisvel.com/images/documents/...1_10032020.pdf AV1 list (so far)
    You can search for the 'Pat.Nr' on google or duckduckgo and will find most of the time a readable version on patent.google
    This should be the first in the list https://patents.google.com/patent/CA...n?oq=CA2406459

    TBH as a non lawyer, is it quite interesting to randomly look thru the list. Hope it will be Fun to watch once the Alliance for open media does fight against these claims
    Last edited by Toggleton; 03-14-2020, 06:44 PM.

    Leave a comment:


  • CommunityMember
    replied
    Originally posted by OneTimeShot View Post
    Hmm... the patent trolls are out to get VP9 and AV1.
    While AV1 techniques have undergone review of possible patented technique infringement, A/V codecs have always been a patent minefield (some of the patents have covered broad fundamental functionality that it is hard to avoid getting close to). Since the current list of potentially applicable patents is not public, there is no way to know what the claims being asserted may be, nor what, if anything, will need to change for AV1 to be either non-infringing or freely implementable if the validity and applicability of those patents is sustained.

    Leave a comment:


  • OneTimeShot
    replied
    Hmm... the patent trolls are out to get VP9 and AV1.

    let’s hope Google can hit back hard with their video patent stack and Android support.

    Leave a comment:

Working...
X