No announcement yet.

SVT-AV1 0.5 Released As Intel's Speedy AV1 Video Encoder

  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Originally posted by sophisticles View Post
    I nearly always use CRF 15, preset very slow and tune PSNR.
    At CRF 15, the bitrate will be so ridiculously high, the tuning will highly likely not matter. This, btw, is another point of bad encoder tests - too high a bitrate.

    Originally posted by sophisticles View Post
    And my "issue" with Jason/Fiona, is that he was full of crap, he lied about so many things, GPU acceleration, psychovisual optimizations, the list goes on.
    GPU encoders went nowhere, what we have nowadays are dedicated encoder ASICs, so she was right about GPU acceleration. And it's only you who claims psy enhancements to be a lie, the quality of x264 clearly demonstrates otherwise. If the list goes on, please do continue it.

    Originally posted by sophisticles View Post
    He took credit for shit he didn't invent or conceive of
    Where's the proof of that? You're making bold claims here, but unless you can back them up with links to where she said these things, all you have is baseless attacks against a person, because of some personal beef.

    Originally posted by sophisticles View Post
    He has built up a cult-like following, almost becoming a folk hero by claiming to have invented and created stuff that predated his software by nearly a decade.
    Unless you show proof, what you're saying here is entirely false. What she did is do is not invent these things, but managed to do a really good implementation of them, which resulted in a kick-ass encoder. There's no "cult of personality" here, but a technically extremely well implemented encoder. The only thing she claims was her idea is MB-tree: <- but even here she clearly states the inspiration was observing what another encoder did and trying something similar on a different scale (not limited to I-frames like in that other encoder).

    Basically, back up your claims, otherwise all you have is personal attacks.


    • #32
      Originally posted by Gusar View Post
      This, btw, is another point of bad encoder tests - too high a bitrate.
      That entirely depends on your use-case. When my use case is to archive raw video footage from my camera at reasonable cost, then encoding at bitrates high enough to not suffer from any visible quality loss is exactly what I am interested in, and comparing encoders for this scenario makes perfect sense for me.

      x264 for example is very good for that purpose, while x265 or any AV1 encoder are still way too slow for me to wait until my recording media has been backed up and is ready to re-use for the next shots. Plus, the claims that HEVC allows for half the bit-rate compared to H.264 is clearly wrong for high quality targets - I have tested that repeatedly myself, and see lots of compression artifacts in HEVC encodings that are half the size of H.264 encodings that look pristine.

      As for Netflix, I understand that their use case is very different - encoding costs are almost irrelevant for them, as encoding is done just once or a few times, while every byte saved on delivery raises their profit, as does lowering quality.


      • #33
        Originally posted by sophisticles
        I really think Jason decided to become Fiona so he could go fuck herself.
        Wow. Just... wow...
        Last edited by Gusar; 22 May 2019, 04:02 PM.