Announcement

Collapse
No announcement yet.

Open-Source Software Encode/Decode For H.266/VVC Progressing

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

  • #11
    Originally posted by MorrisS. View Post
    What's the necessity of VVC, having been developed AV1?
    As others have stated, VVC is better than AV1 in some use cases. You may not care about those use cases, but some at least think they do (or will).

    Comment


    • #12
      Originally posted by DanL View Post
      I'm not really looking forward to new codecs. AV1 should be good enough for most cases and it already took way too long for AMD and Nvidia to get hardware support, especially in their midrange and lower end GPU's.
      If you aren't why are you leaving a comment in this topic?
      If you believe "AV1 is good enough" why are you leaving a comment in this topic?
      If you know very little about codecs in general and it's well outside your interests why are you leaving a comment in this topic?

      AV1 has been good enough for no one so far. It's extremely computationally expensive, so expensive in fact only Google can afford to encode some of top watched videos using it. Yes, Intel Arc, RDNA3 and Ada Lovelace all contain HW encoders for it but they are so bad they are only OK for streaming. I can imagine x264 --preset veryslow being better than hardware encoded AV1 despite being more than a decade older.
      Last edited by avis; 11 February 2023, 04:38 PM.

      Comment


      • #13
        AV1 is slightly better than H265 and VVC is a bit better than AV1, AV2 might compare to VVC. Is that an accurate conclusion of todays insights?

        Comment


        • #14
          Originally posted by Velocity View Post
          AV1 is slightly better than H265 and VVC is a bit better than AV1, AV2 might compare to VVC. Is that an accurate conclusion of todays insights?
          There's no AV2 to speak of, the rest of what you're saying is true. Only VVC is quite a bit better than AV1, not just a bit better.

          Comment


          • #15
            Originally posted by archkde View Post
            It's the same story as with other MPEG codecs.
            As you say, the code is considered to be source available, but you can't legally use it without a license in the jurisdictions that have patented the codecs. And while MPEG-LA's patent portfolio license allows free software projects to distribute a number of copies of their software without paying any royalty, one needs to enter into the license, and to be able to track and show those numbers, and start paying when the number of copies is exceeded, which makes it's inclusion a non-starter in many environments. A workaround for some is to create and distribute their software from locations that have not granted the patents, as long as the parent organization(s) have no presence in the locations where the patent is valid. A more viable workaround for a number of Linux distros has been to have a company provide the library and pay the yearly maximum royalty (Cisco pays for a license to H.264 and creates the binaries for many Linux distributions), but the yearly max for H.266 is quite a bit higher, and unless VVC takes off in the consumer electronics world no company may decide to take on that ongoing burden.

            Comment


            • #16
              Originally posted by DanL View Post
              I'm not really looking forward to new codecs. AV1 should be good enough for most cases and it already took way too long for AMD and Nvidia to get hardware support, especially in their midrange and lower end GPU's.
              Few hardware vendors finalizes their VPU (Video Processing Unit) codec blocks until the standard is complete and released, and inclusion into future products can then begin (preliminary work will be ongoing, but until the ink is dry, you don't want to commit). Two years after the standard is finalized is about the earliest one is likely to see silicon from the major vendors, and it can easily take another year or two for full capabilities to be available in released products. Those delays frustrate those who want it all yesterday, but unless you are willing to accept a possibly defective implementation (should the standard be revised before final release) it is unlikely to get a lot better.

              Comment


              • #17
                Originally posted by Velocity View Post
                What about AV2? AV1 codec bitstream is locked, but it is less 'good' than VVC i heard. So AV2 might compare to VVC?

                If one wouldn't care about licensing, what would be the ideal codec to use at home for transcoding all your videos to, AV1, upcoming AV2 or VVC?
                Yeah, as avis said the best thing to do is leave them be unless the files are enormous or are something ancient like MPEG2.

                If you are trying to optimize copies for remote viewing with Plex or something, av1an's VMAF target mode with aom av1 (or x265 for better client support) is state-of-the-art afaik, and you can prefilter them with a denoiser or an AI filter I guess.


                Its too early for VVC to be anything but a toy, and AV2 is way past the horizon (but may be quite interesting if they end up using some machine learning filters).
                Last edited by brucethemoose; 11 February 2023, 04:38 PM.

                Comment


                • #18
                  Originally posted by nazar-pc View Post
                  It is a dead on arrival codec unless situation around patents is dramatically better than with HEVC, which I don't think is realistically going to happen.
                  Whether VVC is a success will depend on whether the major consumer and business product vendors find the advantages valuable to their customers use cases (which has been true for all the previous video codecs too), and nothing to do with its licensing costs (which will just be added to the cost of the various products, just as all previous video codecs licensing has been (for what it is worth, the VVC license will have a discount if you also license HEVC)). TTBOMK important software (in the video industry) such as Adobe Premiere Pro does not yet support VVC but I would expect if Adobe's customers want it it will happen soon enough.

                  Comment


                  • #19
                    Originally posted by CommunityMember View Post

                    Whether VVC is a success will depend on whether the major consumer and business product vendors find the advantages valuable to their customers use cases (which has been true for all the previous video codecs too), and nothing to do with its licensing costs (which will just be added to the cost of the various products, just as all previous video codecs licensing has been (for what it is worth, the VVC license will have a discount if you also license HEVC)). TTBOMK important software (in the video industry) such as Adobe Premiere Pro does not yet support VVC but I would expect if Adobe's customers want it it will happen soon enough.
                    Some of the biggest users will probably be streaming services, where even tiny bandwidth reductions translate to millions of dollars saved. And streaming box/smart tv SOCs appear to be the first hardware decode adopters: https://en.m.wikipedia.org/wiki/Versatile_Video_Coding


                    IDK about the big editing suites. They still default to x264 (with not very optimal settings) or ProRes from what I've seen, so they seemingly arent in a rush to optimize the quality of exports.

                    Comment


                    • #20
                      Originally posted by brucethemoose View Post
                      Its too early for VVC to be anything but a toy, and AV2 is way past the horizon (but may be quite interesting if they end up using some machine learning filters).
                      I do wonder how much silicon area will be needed to implement some of the various proposed AI algorithms. Maybe we will go back to having dedicated encoding/decoding chips (like the Broadcom Crystal HD chip).

                      Comment

                      Working...
                      X