Netflix Main@L4 Profile

Discussion in 'AnyStream' started by LightEssence, Dec 8, 2020.

Tags:
  1. LightEssence

    LightEssence Active Member

    Is it possible to select main@L4 not high@L4 profile? At Netflix, the main@L4 profile is preferred, since the quality is higher
     
    Last edited: Dec 8, 2020
    getol and itofeddy like this.
  2. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

    Out of curiosity, why do you say "quality is higher"?
     
  3. DeepSpace

    DeepSpace Active Member

    I think this is only possible if NF includes this information inside the playlist, because AS does not convert the files.
     
  4. LightEssence

    LightEssence Active Member

    I had experience in 2018,2019 using various tools... 300MB against 700 on average
     
  5. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

    Huh?
     
  6. LightEssence

    LightEssence Active Member

    tomorrow I'll throw off the examples, now it's just night outside, but as far as I remember main and high are available for L3 devices in this case Google chrome, vp9 and hevc only for L1 widevine
     
  7. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

    That's not what I asked; you said Main@ videos were of higher quality cf. High@ ones, I asked why you think that is so...
     
  8. LightEssence

    LightEssence Active Member

    because I've already worked with both profiles... now I sit with the phone MediaInfo can't throw off... only as I will be at the PC
     
  9. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

  10. LightEssence

    LightEssence Active Member

  11. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member


    You're kidding right? The High-coded video has much more complexity and more than twice the size of the DPB constraint. Bitrate isn't everything, you need to understand video coding and how that works! For example you can stick a bunch of pictures into your DPB and keep referring to them, or keep having to re-supply mostly similar pictures thus wasting the bitrate.
     
  12. LightEssence

    LightEssence Active Member

    I've been doing video for a long time and am familiar with video encoding and frame comparison firsthand.if you're a fan of the 1500kbps bitrate at 1080p, I sympathize with you... active scenes will have broken frames which is unavoidable with this compression ratio
     
  13. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

    Well, if you are familiar with video encoding, you'd have spotted that both in that example are VBR. The Main-coded one has peak rate of 6330 kbps, lower DPB constraint, and a VBV-bufsize cap of 8440 kb; whereas the High-coded one has a peak rate of 20 Mbps, axiomatically higher DPB constraint, and a VBV-bufsize cap of 25 Mb. (You know what happens in the encoder when it hits the VBV-bufsize cap, what happens to the quantizer?) Now tell me, which of the two is way more likely to preserve more entropy in "active scenes," as you call them?
     
  14. LightEssence

    LightEssence Active Member

    these 20mbits of yours will pull out one scene but along the way they will ruin a couple of dozen others, since they simply do not have enough bitrate for this, and small details will be lost...
     
  15. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

    not how vbr works
     
  16. LightEssence

    LightEssence Active Member

    are you a person from the early 2000s? when was it possible to ruin quality for the sake of size? I know how vbr works, but you will only get a good result for 1080p with an average bitrate of at least 5000kbps
     
  17. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member


    It's very easy to have larger file with worse quality than a smaller file with better quality; for a "home encoder" size-quality is a trade-off, for NF, it's not (which is why they can computationally afford motion estimation range of 120!) There are plenty of analyses that produce better size but take quadratic, if not worse, time, for NF it's not a problem: they have reserved computational capacity at Amazon. The point is: NF wouldn't supply higher-complexity lower bitrate if it were worse than lower-complexity higher bitrate (incidentally, iTunes videos that I've watched and that are bigger than NF ones actually look visually worse because of Apple's coding decisions). You simply can't say that the quality is worse if there are that many variables and you have no master to compare against (which NF do, btw). Your hard and fast rule of 5Mbps is naive.
     
  18. libris

    libris Member

    you know nothing about the quality.
    The only reason to you defend the high profile is that anystream cannot download the main profile.
     
    towboat likes this.
  19. 0x0x0x0x0

    0x0x0x0x0 Well-Known Member

    Way get some objective argument in there! Like I said to OP, SHOW ME THE EVIDENCE! The argument that comes from both of you is higher b/w demand => higher quality, well, let's just stick with MPEG-2 then! In the absence of evidence, anything you say is but supposition posing as fact.

    It's actually very simple: get the Big Buck Bunny, and encode that with the parameters that the two profile videos from NF expressly say they use, then feel free to judge visually or use VMAF.
     
  20. LightEssence

    LightEssence Active Member

    Example based on pokemon season 22 episode 54
    Code:
    https://screenshotcomparison.com/comparison/8865
    High@L4
    Code:
    Видео
    Идентификатор                            : 1
    Формат                                   : AVC
    Формат/Информация                        : Advanced Video Codec
    Профиль формата                          : High@L4
    Настройки формата                        : CABAC / 4 Ref Frames
    Параметр CABAC формата                   : Да
    Параметр RefFrames формата               : 4 кадра
    Идентификатор кодека                     : avc1
    Идентификатор кодека/Информация          : Advanced Video Coding
    Продолжительность                        : 25 м. 26 с.
    Вид битрейта                             : Переменный
    Битрейт                                  : 1 198 Кбит/сек
    Максимальный битрейт                     : 20,0 Мбит/сек
    Ширина                                   : 1 920 пикселей
    Высота                                   : 1 080 пикселей
    Соотношение сторон                       : 16:9
    Режим частоты кадров                     : Постоянный
    Частота кадров                           : 23,976 (24000/1001) кадра/сек
    Цветовое пространство                    : YUV
    Субдискретизация насыщенности            : 4:2:0
    Битовая глубина                          : 8 бит
    Тип развёртки                            : Прогрессивная
    Бит/(Пиксели*Кадры)                      : 0.024
    Размер потока                            : 218 Мбайт (90%)
    Библиотека кодирования                   : x264 core 148 r2717 0d2410d
    Настройки программы                      : cabac=1 / ref=4 / deblock=1:0:0 / analyse=0x3:0x111 / me=umh / subme=10 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=120 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=8 / lookahead_threads=2 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / stitchable=1 / constrained_intra=0 / bframes=16 / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=240 / keyint_min=121 / scenecut=0 / intra_refresh=0 / rc_lookahead=240 / rc=crf / mbtree=1 / crf=20.0 / qcomp=0.50 / qpmin=6 / qpmax=51 / qpstep=4 / vbv_maxrate=20000 / vbv_bufsize=25000 / crf_max=0.0 / nal_hrd=vbr / filler=0 / ip_ratio=1.40 / aq=1:0.80
    Дата кодирования                         : UTC 2020-12-09 07:04:46
    Дата пометки                             : UTC 2020-12-09 07:04:46
    Codec configuration box                  : avcC
    Main@L4
    Code:
    Видео
    Идентификатор                            : 1
    Формат                                   : AVC
    Формат/Информация                        : Advanced Video Codec
    Профиль формата                          : Main@L4
    Настройки формата                        : CABAC / 3 Ref Frames
    Параметр CABAC формата                   : Да
    Параметр RefFrames формата               : 3 кадра
    Идентификатор кодека                     : V_MPEG4/ISO/AVC
    Продолжительность                        : 25 м. 28 с.
    Вид битрейта                             : Переменный
    Битрейт                                  : 3 160 Кбит/сек
    Максимальный битрейт                     : 4 740 Кбит/сек
    Ширина                                   : 1 920 пикселей
    Высота                                   : 1 080 пикселей
    Соотношение сторон                       : 16:9
    Режим частоты кадров                     : Постоянный
    Частота кадров                           : 23,976 (24000/1001) кадра/сек
    Цветовое пространство                    : YUV
    Субдискретизация насыщенности            : 4:2:0
    Битовая глубина                          : 8 бит
    Тип развёртки                            : Прогрессивная
    Бит/(Пиксели*Кадры)                      : 0.064
    Размер потока                            : 561 Мбайт (55%)
    Библиотека кодирования                   : x264 core 148 r2717 0d2410d
    Настройки программы                      : cabac=1 / ref=3 / deblock=1:0:0 / analyse=0x1:0x111 / me=umh / subme=10 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=0 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=8 / lookahead_threads=1 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / stitchable=1 / constrained_intra=0 / bframes=2 / b_pyramid=0 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=0 / intra_refresh=0 / rc_lookahead=48 / rc=2pass / mbtree=1 / bitrate=3160 / ratetol=1.0 / qcomp=0.50 / qpmin=6 / qpmax=51 / qpstep=4 / cplxblur=20.0 / qblur=0.5 / vbv_maxrate=4740 / vbv_bufsize=6320 / nal_hrd=vbr / filler=0 / ip_ratio=1.40 / aq=1:1.00
    Default                                  : Да
    Forced                                   : Нет
     
    Last edited: Dec 10, 2020