markfilipak's query on video presentation

[MasterInQuestion]

[[
[ MasterInQuestion @ CE 2024-06-16 02:46:16 UTC:
https://github.com/EricBerendsen/dvbinspector/issues/71#issuecomment-2171020092
    "The video is DTS-ordered (i.e., physical order is by DTSes) and 100% open GOP."
<^>    After more analysis on actual relevant implementations:
    I find perhaps your statement outright doesn't make sense...

    Current Open GOP implementations mostly involve B-frames:
    Which invariably contradicts the (frame presentation be) DTS-ordered assertion:
    https://trac.ffmpeg.org/ticket/11055#comment:16 ]

----

[ markfilipak-windows @ CE 2024-06-17 17:52:37 UTC:
https://github.com/EricBerendsen/dvbinspector/issues/71#issuecomment-2173994543
    How so?
    I've browsed packets from a couple dozen M2TS. They were all DTS-ordered.
    This movie is from Criterion. It is DTS-ordered.
    And all modern M2TS appear to be Open GOP.
    Has my thinking gone into the weeds?

    FFmpeg apparently assigns DTS to B-frames equal to the B-frames' PTS.
    So:
    PTS: I B P B I
    DTS: I P B I B
    Isn't it? ]
]]
    Your example is over-complicated: as frame referencing may be arbitrary.
    (thus there may be other outcomes)

    You seem to lack adequate understanding on the video presentation things:
    Notably the concept of B-frame.
    .
    Mostly all explained since:
    https://trac.ffmpeg.org/ticket/11055#comment:16
    (search also later comments: keyword "B-frame")

    Allow using OGOP doesn't mean it must: used only when beneficial.
    (relevant concepts also explained in aforementioned: deep trace links)

[MasterInQuestion]

[ markfilipak @ CE 2024-06-17 17:25:29 UTC:
https://trac.ffmpeg.org/ticket/11055#comment:118
    While I have your eye, MasterQuestionable, may I ask a question?

    What I did was rip all 4 episodes: then trim them, concatenate them, transcode them to HEVC-MP4.
    I did it in that order because I thought I'd get the best results ... all streams with a single timebase, you know.
    The end result: HEVC-MP4 did not play well at the joins, and crashed PowerDVD on loading.

    I'm now redoing everything in this order:
    Rip-convert to HEVC-MP4, then: trim, concatenate.
    .
    The HEVC-MP4 will have differing stream timebases that may complicate the trimming.
    I'm crossing my fingers.
    What do you think? ]

----

    Encoding can be lossless: which you can use as intermediate.
    (also archive source; but typically inapplicable for general playback: due to size and speed)
    .
    Compression may also trade size for speed (useful for intermediate case) or vice versa (archive).
    [ No compression may actually result in worse speed due to storage bandwidth cap. ]

    However strict lossless may not be meaningful whatsoever.
    See: https://trac.ffmpeg.org/wiki/denoise
    .
    And lossy lossless may be essentially the same thing.
    (lossy ≈ pre-filtered lossless)

    Time base mostly wouldn't matter for modern implementations. (mentioned in comment:86)
    Also: https://github.com/MasterInQuestion/talk/discussions/3#discussion-5219034