Analog to AVI or Analog to Mpeg?

Discussion in 'Amateur Video Production' started by Brian, Jan 12, 2004.

  1. Brian

    Brian Guest

    I read that some analog to digital cards create Mpeg on the fly (in
    real time). If the analog is changed to an Mpeg file then can I remove
    bits of the video that I don't want then create an DVD video.
    Will the software creating the DVD video want to re-compress the mpeg
    file again?
    I like the idea of the card creating an Mpeg file in real time which
    would save me a lot of time that is wasted when rendering the file
    from AVI format.

    Regards Brian.
     
    Brian, Jan 12, 2004
    #1
    1. Advertisements

  2. Brian

    Rudy Benner Guest

    If you are planning on editing, then capture to AVI, do your editing, then
    render to MPEG.

    http://www.lordsmurf.com has a nice explanation that even I can follow.
     
    Rudy Benner, Jan 12, 2004
    #2
    1. Advertisements

  3. Brian

    Nomen Nescio Guest

    Some DVD-authoring software will, some DVD-authoring software won't.
    Check out below to see how one person does it:

    <http://www.lordsmurf.com/>
     
    Nomen Nescio, Jan 12, 2004
    #3
  4. Brian

    Toneburst Guest

    Hi,

    it needn't be a 'card'. I use the Pinnacle Moviebox DV. It renders my HI8
    analogue to DV (720x576) in realtime. The software that comes with it is ok
    after updating to 8.12. Burns DVD from within program. There is an USB2.0
    version that looks ok, but if you buy the Firewire-version you can use any
    other editing program (like premiere 6.5) for capturing. The Moviebox is an
    external device, connected to you PC through Firewire, capable of converting
    analogue to DV and VV in realtime. It can even be used as a standalone
    device for converting. Check it out on there website:
    http://www.pinnaclesys.com.
    There is a lot of negative comment about this product but this is mostly
    based on software incompatabilities. If you have a nice streamlined
    Intell-based system it's flawless.

    Jos.
     
    Toneburst, Jan 12, 2004
    #4
  5. Brian

    geno_cyber Guest

    Either use DV or MPEG I-Frame only formats for editing then you can re-encode the result to MPEG IPB
    and create DVD-Video compliant streams.
    Just avoid using MPEG 1/2/4 IPB as your source for editing.
    Also, it would be good to deinterlace everything and not work with interlaced video at all. Although
    MPEG-2 and higher support interlacing it's much better to avoid it, the optic flow axis gets heavily
    corrupted and MPEG compression works better with non-interlaced video.
    So using a good deinterlacing filter to make all your final videos progressive only would result in
    higher quality streams, if properly set that is.

    The Canopus DVStorm2 Pro seems to be the best DV/MPEG Hardware capture card to me and might be my
    next purchase as soon as I can afford it. In the meantime I might upgrade from my now ancient DC10+
    MJPEG Hardware PCI codec card to a Canopus ADVC-100 DV Hardware Firewire simply because DV to MPEG-2
    transcoding usually takes less time than MJPEG to MPEG and because Pinnacle drivers support for
    standard APIs has been quite lame so far and their drivers are quite crippled to force users to use
    their applications only, a very bad marketing policy that has no reason to exist other than to
    frustrate customers and limit hardware use.
    Canopus products have no such restrictions AFAIK and their drivers,where needed, are fully compliant
    with standards.
     
    geno_cyber, Jan 12, 2004
    #5
  6. Brian

    Gavin Hill Guest

    Ok, so mpeg editing is a no no. If you already have a card that
    captures to mpeg (harware encoding) what about transcoding to AVI,
    editing, then back to mpeg? Will there be much lost format? I realize
    it's a goofy way of doing it but it's what I have at the moment.
     
    Gavin Hill, Jan 12, 2004
    #6
  7. Brian

    Nomen Nescio Guest

    MPEG editing is problematic, even with expensive software like Mediaware's
    M2-Edit Pro.

    Simple cuts can be done nicely with M2-Edit Pro or Womble's MPEG2VCR, but
    complex editing is best (and maybe only able to be) done in another format
    such as AVI.
    That's a poor choice and normally only an option of "last resort". If you
    plan on doing a lot of that either (1) realize your quality will suffer or
    (2) get a device that captures in AVI such as a Canopus ADVC-100.

    You could also see if you could capture in I-frame only MPEG, that way cuts
    and edits could be done without much loss (about the same as AVI).
    I think the quality loss would be unpleasant, but people have different
    expectations, What I find unacceptable, you might find tolerable.

    I would recommend you capture to MPEG with your hardware device and save
    that as a file. Then convert that to AVI, and re-render to MPEG. Then
    compare the two results. Then you will discover for yourself if the
    quality loss is acceptable.
     
    Nomen Nescio, Jan 12, 2004
    #7
  8. Brian

    Nomen Nescio Guest

    That depends on your source destination. If your source is interlaced and
    it's going to be viewed on a destination that is interlaced (most
    televisions), then keep it interlaced.

    See <http://www.lukesvideo.com/interlacing.html> and
    <http://www.lordsmurf.com/> under the section entitled "Interlacec vs.
    De-interlaced".

    If your source is interlaced and it's only going to be viewed on a computer
    monitor (progressive), then you can either keep it interlaced and use a
    software player that de-interlaced on the fly (e.g., Cyberlink's PowerDVD
    XP) or de-interlace it prior to encoding.
    I agree here. In a perfect world, all DVD-Video sources would be
    progressive. It's best to avoid interlaced material, but many times, you
    cannot.
    Once again, quality will suffer if you take an interlaced source and
    deinterlace it if you intend to view it on an interlaced destination
    primarily.
    Smart move there.
     
    Nomen Nescio, Jan 12, 2004
    #8
  9. Brian

    Will Dormann Guest

    No need to create an intermediate file. Use VirtualDub-MPEG2 and
    frameserve if your editing software isn't MPEG2 aware.


    -WD
     
    Will Dormann, Jan 12, 2004
    #9
  10. Brian

    geno_cyber Guest

    That's wrong. Digital encoding and compression schemes like MPEG are not suited for legacy
    interlaced video formats.
    Interlacing is a thing of the past and should be avoided.
    It's unfortunate that the industry still forces interlaced video to the market, especially digital
    interlaced video.
    While it's true that standard Tv formats are still interlaced and that HDTV allows progressive
    video, digital encoding could be progressive only and interlaced sources could be deinterlaced
    properly and interlaced hybrid MPEG formats (and others still interlaced based) could be completely
    avoided.
    Interlaced video can be generated on the fly on any digital device, DVD players for sure, it's part
    of the design that progressive video is automatically encoded to interlaced by the video output
    enconding chipsets, so there's no real reason to use MPEG Interlaced coding.
    Interlaced MPEG was in practice forced to be put into the standard but there would have been no real
    reason for it, it doesn't improve quality nor it saves computing time, instead it's even more
    complex to compute the vectors properly at every algorithm levels...

    I tested deinterlaced MPEG-2 Vs interlaced MPEG-2 on DVD at the same bitrate with same codec and
    same settings and the deinterlaced one looked much better every single time, although it was still
    interlaced when watching on my Sony 29" Tv set and playing it back with a surely not top-notch Sony
    DAV-S550. Even ancient DVD players like Samsung 709 did a better job with progressive MPEG streams
    than with interlaced ones.
    Obviously if the deinterlacer is no good then the result would simply suck to the eye but due to the
    nature of interlaced video, the fact that it was invented just to cut broadcasting costs and needed
    bandwidth when Tv standards were set and then the fact that digital video encoding gives better
    results with progressive sources simply means that any interlaced video should be deinterlaced and
    converted to the best progressive that can be obtained before the encoding stages take place.

    Well, deinterlacing algorithms do exist, it's a real mistery why many DVD production studios force
    interlaced video formats.... even considering the fact that the film sources are usually 24fps
    progressive and so forcing legacy 3:2 pulldown techniques surely corrupts the optic flow axis even
    more which simply contributes to degrade the signal overall. Any good deinterlacer with proper
    motion compensation could ensure much better results with true progressive video even if a producer
    would like to convert from 24fps progressive to 25fps PAL/SECAM or 29.97 NTSC progressive frame
    rates. However I don't see that many DVDs encoded in 24fps progressive, the majority simply forces
    the 3:2 pulldown setting of MPEG.
    Not necessarily if deinterlacing is done properly. Proper motion compensation algorithms can ensure
    almost no noticeable corruption of the optic flow axis.
    Video encoding chipsets for video output can regenerate fields properly on the fly from progressive
    sources and if the interlaced->progressive conversion was done with a good deinterlacing algorithm
    implementation and the resulting optic flow axis is mantained in a good or almost optimal state then
    even and enhancement over the original interlaced source could be obtained only thanks to motion
    compensation techniques that could fix the optic flow axis issues introduced by interlacing.
     
    geno_cyber, Jan 13, 2004
    #10
  11. Brian

    Will Dormann Guest


    When viewing on a regular television, a 60 field per second source
    (interlaced) will look smoother than a 30 frame per second source
    (progressive).


    -WD
     
    Will Dormann, Jan 13, 2004
    #11
  12. Brian

    Bariloche Guest

    Mpeg 1 is not meant for interlaced video. So it was developed in order
    that it would be able to cope with it -and Mpeg 2 was born.
     
    Bariloche, Jan 13, 2004
    #12
  13. Brian

    Bariloche Guest


    Who knows, depends on the software. Tmpgenc, for instance, does not
    re-compress. It just doesn't provide frame-precise cuts. There are
    Mpeg2 editing programs (Mpegvcr, M2edit) than can do frame-precise
    cuts, and only recompress a few of the frames. Capturing as mpeg, then
    trimming your material with those editors, would be the best way in
    terms of speed. OTOH, Quality-wise, one would capture to AVI, edit,
    then encode to mpeg.
     
    Bariloche, Jan 13, 2004
    #13
  14. Brian

    geno_cyber Guest

    I didn't say MPEG-1 there....
    MPEG-2 wasn't developed to just include support for legacy interlaced video.
     
    geno_cyber, Jan 13, 2004
    #14
  15. Brian

    geno_cyber Guest

    Interlaced looking smoother than progressive ? I doubt so. The optical flow axis gets heavy
    corrupted due to interlacing and so every single object on the screen simply doesn't move that
    smoothly.

    30fps progressive is converted to interlaced by the video out encoding chipset and if there are no
    major bugs with it then the output source would still be field-based but the MPEG source would be
    progressive which even avoids a 24fps film progressive -> 3:2 pulldown to 25fps or 29.97fps
    interlaced video conversion.
     
    geno_cyber, Jan 13, 2004
    #15
  16. Brian

    Nomen Nescio Guest

    You posted a lot of good info that I snipped, but I still disagree from my
    own personal tests.

    Interlaced sources look best when captured interlaced and encoded
    interlaced if they are going to be displayed on an interlaced destination
    (e.g., NTSC TV).

    Taking an interlaced source (1), deinterlacing it (2), encoding it to
    progressive MPEG-2 (3), converting it back to interlaced for viewing on a
    TV (4), and (5) displaying it on a TV involves at least 5 steps where there
    is quality loss.

    Taking an interlaced source (1), encoding it for interlaced MPEG-2 (2) and
    displaying it on a TV (3) only introduces 3 steps.

    My, and many others, tests lead to the conclusion that the small amount of
    improvement and increased bitrate you get from being able to encode
    progressive MPEG-2 is MORE THAN OFFSET by the quality hit one takes in the
    process of interlacing -> deinterlacing and deinterlacing -> interlacing.

    I understand you think differently. I don't. My results reflect those
    given at <http://www.lordsmurf.com> under the "Interlaced vs.
    De-interlaced" section of the Capture Guide there.
     
    Nomen Nescio, Jan 13, 2004
    #16
  17. Brian

    Ken Maltby Guest

    This is one case where we both agree, the "optic flow axis"
    effect in video translations is swamped by several others.

    This poster's mother must have been scared by John
    Watkinson or he is trying to apply research into robotic
    vision concepts to this issue.

    Lol,
    Ken
     
    Ken Maltby, Jan 13, 2004
    #17
  18. Brian

    geno_cyber Guest

    Do you think to be funny by insulting my mother, uh ?
    Well, that's not funny at all. Would you like if I did the same talking about your parents ,
    relatives or acting like you did above ?

    You should read some good books and study a lot to understand what I was talking about the optic
    flow axis there...

    Insulting people to look funny when you don't know anything about something it's just pretty
    disgusting and plain stupid.

    I surely didn't say anything personal to you nor tried to insult you but still you wanted to look
    "funny" by trying to start a flamewar here.. well, that's just pathetic and not funny at all.
     
    geno_cyber, Jan 13, 2004
    #18
  19. Brian

    geno_cyber Guest

    The number of steps required to achieve a result doesn't necessarily equate to a loss of quality.
    Although there are limits over which you'd surely get quality loss the number of steps or how much
    complex an algorithm is simply don't equate to bad quality.
    If algorithms and filters are designed efficiently and a good implementation overall then there
    should be no quality loss.

    If deinterlacing is done properly there's no way to get lower quality than the original interlaced
    source.
    I repeat once again that what matters is the optic flow axis that your eyes follow for every single
    object on the screen and properly deinterlacing means that a good motion compensation technique
    should be used to find the best possible vectors in order to fix the optic flow axis vectors that
    are always corrupted due to interlacing.
    The wrong technology is interlacing and not progressive. Interlacing was invented to save costs when
    drafts for Tv standards had to be set over 50 years ago. If at the time the original proposed specs
    of 48fps progressive could have been achieved and it hadn't been so expensive to broadcast radio
    signals suitable for analog video coding then we would have got HDTV like quality from the
    beginning. Instead they needed to save on transmission costs and so interlacing was invented but
    it's just a workaround and it should be avoided completely. Conversion from interlaced to
    progressive could even fix video and give it a very good approximation of the original progressive
    source or the best progressive that could have been instead.

    Personally I'll keep converting every interlaced source to progressive using the best available
    algorithms I can find online, like some available with AviSynth (although they're quite slow
    implementations).
     
    geno_cyber, Jan 13, 2004
    #19
  20. Brian

    Nomen Nescio Guest

    True. But the steps we are talking about DO incur a loss of quality.
    Not necessarily true -- it depends on the algorithm/filter.

    Some algorithms, just as gzip do not incur a quality loss, because they are
    lossless. Taking an interlaced source and deinterlacing it is similar to
    performing a JPEG compression on an image. Depending on the amount of
    compression, the quality loss will be severe or be almost unnoticable.
    BUT, there will ALWAYS be some quality loss on JPEG conversions.

    Likewise, with deinterlacing, because you are taking two fields taken at a
    slightly difference slice in time and trying to estimate how they would
    really look if they were taken at one point in time there will always been
    some quality loss because all it is is an estimate or guess.
    Yes there is. There ALWAYS is some reduction in quality because of what
    deinterlacing does. It takes two interlaced frames, that were taken at
    slightly different points in times and tries to estimate what these two
    fields would look like if they were actually taken at the same point in
    time. Depending on what is going on and the deinterlacing algorithm, the
    results are pretty close to accurate, other times, even with the best
    deinterlacing algorithms, the results are poor.
    Personally, I hate interlacing and wish all sources and destinations were
    progressive. However, my comments and opinions on deinterlacing interaced
    sources for viewing on an interlaced destination remain unchanged.

    P.S. Could I get you to deinterlace your word-wrap to ~72 characters so I
    don't have to? :)
     
    Nomen Nescio, Jan 13, 2004
    #20
    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.