HELP: First attempt at DV capture - corrupt signal

Discussion in 'Video Cameras' started by H Bergeron, Aug 25, 2004.

  1. H Bergeron

    Matej Artac Guest

    In
    I am a bit too sleepy to get myself through the entire conversation that
    goes on further on in this thread, so it might have been pointed out
    already. But just to be a bit of a nitpick here - in principle one could
    employ DirectShow routines to do the demuxing and other operations by
    programming in VB, so that DirectShow does it in real-time like it would for
    a video playback, and the VB routines do less work. At least I think it
    could be done as I haven't done it myself.

    * Matej goes to bed now.
     
    Matej Artac, Sep 1, 2004
    #81
    1. Advertisements

  2. H Bergeron

    :::Jerry:::: Guest

    You could be right if it was having to do it RT, but I suspect that
    processing video files on a computer in this manor doesn't have to be RT,
    just as rendering is not always RT - it can take many hours to process.

    VB scripting is certainly mentioned on various Vegas forums.....
     
    :::Jerry::::, Sep 2, 2004
    #82
    1. Advertisements

  3. H Bergeron

    SjT Guest

    There's nothing i like more than finding a love letter from
    You told me this 'Thread drift' was acceptable Jerry, and that a
    cretin like me should be able to understand usenet.

    Pot, Kettle, Black.
     
    SjT, Sep 2, 2004
    #83
  4. H Bergeron

    SjT Guest

    There's nothing i like more than finding a love letter from Tony
    I was thinking of the old 60s film :D
     
    SjT, Sep 2, 2004
    #84
  5. H Bergeron

    :::Jerry:::: Guest

    Sorry but I missed the smiley off the post, it was meant as a bit of irony
    after all the time Mr Morgan has complained about the discussion of pro and
    semi pro matters on the group, implying that they were OT here because this
    is a 'rec' group....
     
    :::Jerry::::, Sep 2, 2004
    #85
  6. H Bergeron

    Matej Artac Guest

    In
    It certainly would have been good to be able to export my projects from
    Pinnacle Studio 9 demo into EditStudio, should I decide to buy it (a bit too
    tight on budget right now to go for Vegas just yet...)

    Matej
     
    Matej Artac, Sep 2, 2004
    #86
  7. H Bergeron

    Dave R Guest

    Either through the time code on the tape having a large (configurable)
    gap, or by examining the contents of the picture itself. The latter is
    the cleverer bit, and I assume it works by examining the colours at each
    co-ordinate, from frame to frame these don't tend to change *that* much,
    so form one clip. If they suddenly change a lot - over a certain
    threshold, then it's a new scene.
     
    Dave R, Sep 3, 2004
    #87
  8. H Bergeron

    :::Jerry:::: Guest

    One hopes....
     
    :::Jerry::::, Sep 3, 2004
    #88
  9. H Bergeron

    Matej Artac Guest

    In
    There are plenty of ways do describe an image and compare it to other images
    like sampling the colour histograms, resampling the image and doing some
    sort of a correlation... anything that gives you some measure of the image
    similarity. If a dissimilarity exceeds a threshold, you have a scene change.

    However, just last night I wanted to have my software detect scene changes
    in a DV converted from a VHS source, and it produced none - I guess the
    silly program was trying to find the gaps in the time code...
     
    Matej Artac, Sep 4, 2004
    #89
  10. H Bergeron

    :::Jerry:::: Guest

    So a taped event, were the camera is recording the whole time can not be
    scene detected, well as I keep saying, scene detection is bloody
    useless.....

    Log, Capture decision list, Capture, Edit - it's the only way !
     
    :::Jerry::::, Sep 4, 2004
    #90
  11. H Bergeron

    Matej Artac Guest

    In
    As I said in the next paragraph of my post:
    So my explanation of the scene detection method applies to when the software
    actually looks at the _content_ of the video. However, as you've pointed
    out, some programs simply don't bother as they rely on the metadata stored
    with the video stream.

    By the way, I did a quick test of my theory by extracting a short clip from
    my recording referred to eariler and converted it into a DivX. Then I
    imported it into my program (NeroVision Express 2) and did a scene
    detection. And, guess what, not having an easier way out, it produced quite
    a few scenes! I guess this makes another case of a software trying to be
    smarter than the user ;)

    Matej
     
    Matej Artac, Sep 4, 2004
    #91
  12. I have been busy so have only just caught up with this thread which seems to
    have deteriorated in the usual way.

    I'm not going to get further involved after this post but the case is that I do
    happened to know better.

    Try looking at the AVIFile Functions, try looking at the visual basic samples in
    the DirectX SDK, try looking at http://www.shrinkwrapvb.com/avihelp/avihelp.htm,
    try doing a search for Microsoft RIFF, try doing a search for Microsoft RIFF
    avi2.

    The ability to read and write avi files, with the correct codec being
    automatically used via the RIFF fccHandler element, are all there in the AVIFile
    Functions and easy to access from any language.

    As each video frame is returned as an uncompressed DIB, you could, for example,
    detect a scene change by a very simple algorithm that compares the 41+K bits of
    the previous frame with the next frame.

    Of course, the Blue Book contains basic information but equally microsoft
    publish the actual structure format of the riff based avi files which means that
    you can actually view and follow the structure stream using any binary file
    reader/viewer.

    For goodness sake, just for once admit that a google collection of snippets
    doesn't mean that you have any real understanding of what you can do with SDK
    objects.

    Stuart

    www.mckears.com
     
    Stuart McKears, Sep 5, 2004
    #92
  13. H Bergeron

    Dave R Guest

    You are so wrong.
    No. *Your* scene detection software may be limited to detecting timecode
    changes, but not all software works that same way.
    Wrong. Some software works just like I described a few posts back.
    I have, thanks. :)
     
    Dave R, Sep 13, 2004
    #93
  14. H Bergeron

    :::Jerry:::: Guest

    Tony Morgan and his little Blue book, funny, he does sometimes come over
    like a little dictator !.. :~)
     
    :::Jerry::::, Sep 13, 2004
    #94
  15. H Bergeron

    Dave R Guest

    Yep, plenty of times.
    Live on in ignorance then, it matters little to me.

    Once again you are wrong. Once again you resort to belittling. We've
    almost come full circle. If you now put me in your killfile again, the
    circle will be complete, and that suits me just fine.
     
    Dave R, Sep 13, 2004
    #95
  16. H Bergeron

    :::Jerry:::: Guest

    NO I have not you total plank, as anyone with half a brain cell can see,
    stop lying Mr Morgan, stop trying to disguise the fact that most people on
    thins group have you well and truly marked out as nothing but a class one
    cretin.

    Mr Morgan, you are nothing but a LYING TROLL.

    But you wouldn't know that if I'm meant to be in you kill file, you are
    nothing but a lying troll.
    Yes you are, A LYING ONE AT THAT.
     
    :::Jerry::::, Sep 13, 2004
    #96
  17. H Bergeron

    Dave R Guest

    It's not inability, I just couldn't be bothered.

    You see, you persist in this notion that scene detection can only work by
    looking for changes in the timecode. That is an incorrect statement. One
    of my previous posts explained how software might look at RGB values and
    detect shifts in those to know a scene has changed. I don't know the exact
    criteria used by such programs, and other suggested alternative means, I
    simply know it works.

    That you don't believe me doesn't change the fact that I have done this
    several times with success. Maybe you'll never require this feature, in
    which case your ignorance won't be detrimental to you, this time.

    You're so focused on your newly discovered book of standards, and your
    desire to show-off some understanding of the interleaved data, you've
    missed the point that there many ways to skin a cat.
    Oh I realise it just fine, but it all depends on whether the software that
    they use only supports timecode based scene detection, or whether it's
    better than that.

    But how many of them really care? It's just another useless arguement on
    this group involving you.
     
    Dave R, Sep 13, 2004
    #97
  18. H Bergeron

    Dave R Guest

    What are you talking about?

    7 weeks or so ago, Jerry posted as such...

    Date: Mon, 26 Jul 2004 20:29:06 +0100
    From: ":::Jerry::::" <>

    The post that you replied to, was this...

    Date: Mon, 13 Sep 2004 14:40:19 +0100
    From: ":::Jerry::::" <>

    Where is the change in his name or e-mail addy? What is wrong with you?
     
    Dave R, Sep 13, 2004
    #98
  19. H Bergeron

    Dave R Guest

    Actually I offered 2 explanations of how it worked. One by timecode
    gaps and other by scene differences.

    If you care to correct yourself, you can look it up. The Message Id was
    Well obviously that would depend on the mode in which you captured. Do
    keep up at the back.
    Keep digging.
    If I were using that, then I wouldn't get *any* scene detection. You're
    so blind it's almost amusing.
     
    Dave R, Sep 13, 2004
    #99
  20. H Bergeron

    SjT Guest

    There's nothing i like more than finding a love letter from
    I come over a little dictator once.

    Can't remember his name though, short guy with a funny moustache.
     
    SjT, Sep 14, 2004
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.