Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
25p official topic, aka Don't cry for me Argentina
  • 236 Replies sorted by
  • _gl_gl
    I'll repeat (loosely) what I sent to Ian_t:

    1) audio resampling changes pitch - unacceptable in many cases (people sound older, or younger & the feel of music changes)

    2) audio stretching _without affecting pitch_ is not flawless - cheap algorithms are terrible, better ones like DIRAC3 (eg. in Wavelab 7) are passable but still have artifacts, and are very slow to process. It won't matter to everyone, it matters to me.

    3) even if audio could be perfectly stretched, you are still changing the timing/pacing of the audio and video. For example, actors are now talking faster, music has a faster beat.

    All of this distorts reality, and my artistic choices. It won't matter too much for some shots, it's a disaster for others - if you care.

    Lots of video people care more about visuals than audio (that's why you still often see bad lip syncing in music videos). I care about both. So just because it doesn't seem important to you, don't assume that's true for everyone. I could also say 'why bother with higher bitrate, isn't the GH2's 24mbps already good enough'? It is for some - but if you really care about quality, you want more. Case closed.

  • PtchawPtchaw
    @stolpis100 That's correct, it would affect any camera with rolling shutter regardless of frame rate, though it it is something to be aware of.

    Can't tell you without doing some tests, but from what I remember it has to be pretty high!
  • PtchawPtchaw
    @_gl I know the pace is changing, but I think you really need to try it before you say it has an effect on the footage; I've never been able to tell the difference. You're only being shown 1 more frame per second. Colour grading and polarising filters 'distort reality'.

    Again with audio, I really doubt you'd ever tell the difference, let alone would it be a 'disaster'. I don't know what algorithm editors use, but from experience it has not just been an issue of 'good enough', but indistinguishable from the true speed.
  • andresandres
    Here in Argentina, 99% of the work I do is 25P when I do a documentary for theatrical release it is 25P in festivals I always send 25P, only in fiction we use 24p and not always. Even in short films we work in 25P. The 50i can be good for some things but I alway end up converting to progressive (fast motion gets artifacts).

    Here is a 60Hz 50Hz Map From Kameramann Jahrbuch
    60Hz NTSC
    50Hz PAL
    _1070872.jpg
    1962 x 1344 - 1M
  • _gl_gl
    @Ptchaw, see this is why I'm unsually irritable on the subject. Don't assume I don't know what I'm talking about. Just yesterday I tested it again by resampling a CD track - the resulting speedup (equivalent 24->25p) or slowdown (the other way) are totally unacceptable to me. It's the same pacing difference as I noticed between the US and UK versions of Frasier.

    Let me turn the tables, why don't you try it. Take a music track (with a beat) you like and know, and speed it up to 104.2%, with whatever algorithm you have available.

    As I said, not everyone will care and maybe you don't - but I do.
  • _gl_gl
    BTW, the idea that 'the industry does it, so it must be perfect' is absurd. First we have a large range of quality on our TV screens, not all of it good. 2nd, the PAL TV industry _had_ to convert 24p to 25p because it had no choice, not because the results are flawless.

    People in PAL countries are used to it, especially as they don't get to compare it to the original (like when I finally saw a US rate Frasier episode) - that's when I noticed how far off it actually is. In the Frasier case, the actors sounded much older on the US version, and the dialog and joke rate seemed to drag as everything was happening slower. I also frequently notice it with music that I know well when it turns up on a PAL-conformed movie, it sounds more hyperactive and like it was sung by a younger singer and aimed at a younger audience.

    For another example, many non-English speaking countries dub their English imported TV shows. For people growing up in those countries, it seems perfectly normal, they don't notice the weird lip-sync errors, or that the same voice-artist is doing voices for different actors, and all kinds of weird things like that - because they're used to it, it's what their local industry does. It's only when they start watching the originals with subtitles that they notice how weird it actually is.
  • PtchawPtchaw
    My industry point was that people were saying they required 25p for use in broadcast, when conforming 24p is widespread in broadcast.

    Just tried resampling Beck - Beercan using Soundtouch, which I understand uses the SOLA algorithm. I honestly didn't notice apart from with a direct comparison.

    For non-music work, the difference would be even less noticeable.
  • _gl_gl
    > I honestly couldn't tell the difference apart from via direct comparison.

    Fair enough, but I can.
  • PtchawPtchaw
    @_gl Just curious, do you find it noticeable for dialogue?
  • _gl_gl
    Yes (I mentioned the Frasier example) - but in fairness only if I know what the actor should sound like, or if I can compare with the original version.

    But I don't like not seeing/hearing something the way it was intended, even if I don't know it. In the Frasier example, I watched it for years so I really got used to the actors' pitch and pacing. So it's weird to see them interviewed and think 'wait, that's not what you sound like!'. Feels like it's another person.
  • PtchawPtchaw
    @_gl Interesting. I can see why it would be a lot more noticeable for Frasier, being slowed down by 17%!

    Did some more tests, I think how noticeable it is depends a lot on the song - Beck didn't show it much, but some electronic stuff did seemed a bit quicker to be fair.
  • fehafeha
    Conclusion: 25p is a must :-)
    wah, wah, wah ...
    If PAL camera why not PAL Cinema 25 p ?
    wah, wah, wah ... uff Pana...
  • _gl_gl
    > I can see why it would be a lot more noticeable for Frasier, being slowed down by 17%!

    Apparently it was shot on 24p film, so the difference is only 4%, like a normal 25p conform - it tallies with the audio resampling I did yesterday. It doesn't sound like much on paper, but it really is quite different to my ears.

    @feha, I'd be happier if 25p would just go away and everybody would use 24p. SD TV's were mostly stuck with their local fps, but luckily HD TVs all support 24p now for Blu-ray movies. We just have to kill 25p in broadcasting : ). But for now we're stuck with it ...
  • Vitaliy_KiselevVitaliy_Kiselev
    Guys, just to note.
    GH2 don't have 25p mode amongst sensor setup modes.
    24fps, 50fps, 60fps, 120fps, but no 25fps.
  • zigizigizigizigi
    End of story. Thanks for clarifying.
  • Vitaliy_KiselevVitaliy_Kiselev
    @zigizigi
    Interesting stuff is that according to our testing of GH1.
    MJPEG uses 60fps setup. At least this is how I understand it.
  • skiandbikeskiandbike
    > 24fps, 50fps, 60fps, 120fps, but no 25fps

    120fps is just for fast focus, correct?
  • fehafeha
    Thanks, no hope ...
    Other question:
    There is a variable movie mode ...
    Can you hack that with percent ?
    104 % would be a 25p :-)
    The max is 300%, if it could be hacked to 1000 or 2000 % perfect for timelapse ...
  • _gl_gl
    > GH2 don't have 25p mode amongst sensor setup modes.
    24fps, 50fps, 60fps, 120fps, but no 25fps.

    50p is a nice second-best : ).
  • sparedogsparedog
    120fps? is this possible at 720 or 1080?
  • mpgxsvcdmpgxsvcd
    50 FPS divided by 2 equals 25 FPS. That would work just as well if not better.
  • sparedogsparedog
    @_gl frazier is shot 24p on 35mm right? but we dont know what it is cut to. i assume it would be cut at 29.97 because of part durations needing to be exact.
    the uk versions are supplied on tape, right? so they are would be supplied ntsc and then transcoded to pal. in sd days, i dont remember ntsc>pal conversions having a duration change, so there shouldnt be a pitch shift, or speed shift between american and english versions, only between the live studio viewings and the final cuts. was there a duration change in ntsc>pal conversions of sd?
    tricky one, gonna to try to find out more on this
  • Vitaliy_KiselevVitaliy_Kiselev
    @feha
    Variable modes are separate modes, they are not defined by percent.
    But it could be interesting to look how they skip frames.
  • MimirsanMimirsan
    High bitrate 720 50p will be fine for me as long as I dont have to use the 24p cine record first doobery.
    There wasnt 25p on GH1 originally but you managed to get native 25p out of the 50i...so if possible this way?
  • Vitaliy_KiselevVitaliy_Kiselev
    In GH1 we changed encoder.
    In GH2 it don't work this way.
    In upcoming PTool you'll see quite stable 25p, but. It only has first field in top part, below is garbage.
Add a Comment