Personal View site logo
Official Low GOP topic
  • 1003 Replies sorted by
  • thanks @stip and @stray. no visible damage so far.
  • @Braamokiev
    settings=@stonebat ;-)
  • a simple tweak on @driftwood's settings ;)
  • @ stonebat

    "The h setting is 110Mbps, GOP3, AQ2, X1.5 on FBs and FL. The j setting is 110Mbps, GOP3, AQ3, X1.39 on FBs and FL. Both worked. The attached diagram is from the j setting. My gut feeling is that... the B-frame seems little too big. Or is that a typical behavior from AQ3 setting?"

    I think it is typical of AQ3 and 4... and to my mind it's a good thing, those big b-frames, especially with non moving shots. Going to test your latest right now.
  • @stonebat
    This is a graph from settings 110Mbits, GOP3, AQ3 and 1.37 on FBs anf Fl. This was simple shot of my girlfriend.
    110Mbits GOP3 AQ3 Fb Fl X1.39.JPG
    1288 x 670 - 149K
  • Thanks to:

    @Mihuel
    @Stonebat
    @Drifwood
    @Stip
    @Stray

    And special thanks for Vitaliy!

    I wait the best Setting 100mbs
    ;-)
  • @Mihuel that doent look right, close streamparser reopen it and reopen that same file.
  • @driftwood
    ok, I will try again , thanks for advise :)
    P.S. I'm loading patch with 1.3 FBs and Fl right now .... so after this ....
  • @driftwood
    Unfortunately stream parser several times showed the same thing. Maybe this card? Maybe this is my model of camera?
    When I changed FBs and Fl to 1.3 ....
    Hmmm, I think I'm doing something wrong. Anyone can help me ?
    110Mbits GOP3 AQ3 Fb Fl X1.30.JPG
    1288 x 669 - 148K
  • @Mihuel download Stonebat's110 aq3 (setj) and load it and confirm a test matches his with streamparser
  • @drifwood
    Ok, good idea ( better then mine ) ;)
  • When you lower frame limit it will end up raising the effective QP value to accommodate the lower bandwidth allocated to I frames and do other strange things. It's sort of like clipping - not really a good thing. Better to leave frame limit a little higher and lower AQ until it works. According to my calculations (based in part on your feedback), the ideal AQ value for GOP3 at 100Mb would be between 1 and 2. It seems that 2 works pretty well, 3 is clearly too high according to my latest math. With GOP1, you probably want to use AQ1.

    Chris
  • Mihuel, is your girlfriend a human? :) The B-frames are abnormally high.
  • Orly? Ok. Back to AQ2~!~!
  • @stonebat
    I think so, but you know how a woman is pregnant. Maybe it is a small elephant? ;)
  • @cbrandin
    My jaw dropped to the floor and I already have :-)
  • @Mihuel never mind :)

    @cbrandin
    How do we verify the clipping?

    What about GOP6 for 66Mbps and 88Mbps? AQ3 is the practical limit?
  • For GOP 6 I would estimate that you could go to AQ3 at any bitrate. With GOP1 I think it's AQ 1-2 no matter the bitrate. This is because AQ is automatically scaled according to bitrate.

    Chris
  • @cbrandin Have you worked out a stable-ish multiplier for AQ3 at 132mbps?

    @stonebat elecard buffer analysis verifys the clipping and where it will fail.

    edit: still testing my buffer theory.
  • Interesting. Thanks Chris.

    So we have a matrix?

    GOP12 AQ4 for 44, 66, 88Mbps
    GOP6 AQ3 for 66, 88, 110Mbps
    GOP3 AQ2 for 88, 110, 132Mbps
    GOP1 AQ1 for 110, 132, 154Mbps

    The lower GOP, the higher bitrate range, the lower AQ.

    If AQ and multiplier are set for each mode, it's basically 2-d matrix of Mbps and GOP.
  • @stonebat
    I tried GOP6, AQ3 and FL and FBs x1, 5 Charter does not withstood and the Stream Parser showed a few seconds I = P = B
  • @cbrandin would know a multiplier for the case.
  • ok, let's wait for the opinion of the wiser of us on this issue :-)
  • @cbrandin
    Do you have any suggestions for 6GOP and 88 or 110Mbits? Unfortunately, my knowledge is, at least, enough so that I can handle myself.
This topic is closed.
← All Discussions