Need Advice in Capturing VHS Video to MPEG

Discussion in 'Amateur Video Production' started by Jay Chan, May 10, 2004.

  1. Jay Chan

    Jay Chan Guest

    I have some VHS videos that I recorded using expanded recording (EP),
    and I have already transferred them into miniDV tapes. I have tried
    capturing them into AVI files in hard disk using Ulead Video Studio 7
    SE. But when I convert them into MPEG-1 or MPEG-2 files, I find that
    their file size is way too big (20 minutes video takes 1.2 GB in MPEG
    file (the original AVI file of the 20 minutes video is around 4 GB)).
    I think the settings for capturing and rendering are probably too
    high. I am wondering what settings to use in order to get a reasonably
    good video and use relatively little storage space.

    I was using 720x480 resolution and more than 8000 bitrate to capture
    the video and render the MPEG file. According to messages posted here,
    I am supposed to use 352x480 resolution and something like 4000
    bitrate to get good video while keeping the file size small. I intend
    to do this in this evening; but I have two questions:

    - Should I use the low setting for both capturing video from miniDV
    to AVI, and rendering from AVI to MPEG? Or should I use the high
    setting (720x480 at above 8000) for capturing into AVI, and use the
    low setting (352x480 at 4000) in rendering into MPEG? By the way, I
    don't intend to keep the AVI file, I only use it as a way to get the
    MPEG file.

    - Some messages indicated that I will get a smaller file size if I
    choose MPEG-1 instead of MPEG-2. But the result in my test (using high
    setting in both capturing and rendering) only show a 2% reduction in
    file size. Is this all I can expect from using MPEG-1 file? What's the
    trade off between using MPEG-1 and MPEG-2 anyway?

    Thanks in advance for any suggestion.

    Jay Chan
     
    Jay Chan, May 10, 2004
    #1
    1. Advertisements

  2. Jay Chan

    Will Dormann Guest

    Assuming you're using firewire, it's not really a "capture" from the
    camera, but more of a transfer. What you end up with on your PC
    should be identical to what you've got on tape. There shouldn't be any
    "high" or "low" setting with respect to the process of moving the video
    from your DV tape onto the PC. DV is fixed bitrate.

    Go ahead and grab the DV video from the camera (into AVI format), and
    then render it into MPEG2 of your liking. (Half D1 at 4000 sounds
    reasonable to me)


    MPEG1 does not have the ability to represent interlaced video. If
    you're using the full vertical resolution (480 lines with NTSC), then by
    all means use MPEG2.


    -WD
     
    Will Dormann, May 10, 2004
    #2
    1. Advertisements

  3. Jay Chan

    Jay Chan Guest

    Assuming you're using firewire, it's not really a "capture" from the
    Thanks for correcting me that I should not need to play around with
    high/low settings when I capture the video from miniDV source. This
    saves me one variable to test with. I will try two 480 lines
    variations (720x480, 352x480) with various bitrates to see which
    setting of rendering will give me the best compromise in term of
    quality and file size.

    By the way, which bitrates would you suggest for rendering at 720x480
    and 352x480 resolutions?
    Good to know that. I will stick with MPEG-2. The file size saving is
    so small that this doesn't worth my time and effort to try MPEG-1.
    This saves me one variable to test with.

    Jay Chan
     
    Jay Chan, May 11, 2004
    #3
  4. Jay Chan

    Will Dormann Guest

    The software shouldn't even give you the option. DV is 3.6MB/sec,
    720x480 (NTSC). If it lets you select any other setting than that for
    capture, I don't know what it's doing. (capturing and re-rendering at
    the same time perhaps?)
    It depends on how much you need to fit on a disc. When I make DVDs, I
    usually use Half D1 @ 3000. That lets me put just over 3 hours on a
    single disc. Depending on the quality of your source, encoder, and
    what filters you're using, you may want to increase the bitrate to get
    acceptable results. See:
    http://dormcam.mine.nu:8080/restest

    All of those samples are filtered with virtualdub. Ideally, I should
    have used the highest bitrate possible for my source file, but you
    should still get the idea. I think it was a 5000kbit MPEG2 from my
    PVR-250. See http://mythhelper.sourceforge.net if you want to get the
    filters and presets that I use. (Just grab the source code ZIP for the
    presets)


    -WD
     
    Will Dormann, May 11, 2004
    #4
  5. Jay Chan

    Bariloche Guest

    Encoding quality can be measured in bits/pixel. 352x480 (half D1) has
    about half the pixels of 720x480 (full D1), and thus only requires
    half the bits for the same encoding quality. IOW, 352x480 at 4000
    kbps is aprox. equivalent to 720x480 at 8000 kbps -only the image is
    somewhat less sharp, because of the reduced horizontal resolution.

    Now, a usual way of encoding D1 is CQ 65%, which usually revolves
    around 8500 kbps. What I would do is preparing a small half D1 clip,
    and encoding it at increasing CQ from 65% up: 65%, 70%, 75%, 80%, 85%,
    90%. Then notice the results, and decide by yourself. IMO, half D1 at
    CQ 85% may be undistinguishable from D1 at CQ 65% for many people -but
    the bitrate is substantially lower.
     
    Bariloche, May 12, 2004
    #5
  6. Jay Chan

    Jay Chan Guest

    The software shouldn't even give you the option. DV is 3.6MB/sec,
    Actually, I am following a tutorial that is available in internet on
    the way to capture and rendering using Ulead Video Studio. That
    tutorial established the high-setting in the project properties before
    going on to talk about capturing. That gave me an impression that the
    high-setting is for capturing. Now, I think about this, I have a
    feeling that I might have understand this incorrectly. The high seting
    in the project properties may not have to do with capturing; they are
    being done one after the other may just be a coincidence.
    Good to know that. I will use 3000 bitrate at the starting point, and
    adjust the bitrate upward until I cannot see much improvement or until
    the file size is too big. I am going to try 3000, 4000, 5000 bitrates.

    3 hours video in one DVD should be more than enough for my purpose. I
    am happy to just to fit 1.5 hour of video in one DVD. This means I
    have some room to increase the bitrate. Thanks.
    Great! I will receive my order of PVR-250 TV tuner card very soon. I
    hope I can use the hardware encoder in it to speed up the rendering
    process. This sounds very promising. May I ask how long did you take
    to render the video into MPEG using your PVR-250?

    Jay Chan
     
    Jay Chan, May 12, 2004
    #6
  7. Jay Chan

    Jay Chan Guest

    Encoding quality can be measured in bits/pixel. 352x480 (half D1) has
    Thanks for sharing this info. This means I really don't need to try
    bitrate near as high as 8000 if I set the resolution as 352x480. This
    saves me one test case to try.
    May I ask what is "CQ", and what is the percentage 65%...90% means? I
    probably don't need this info right now for rendering those relatively
    low quality VHS videos that I am doing now. But I will likely need
    this info when I start copying my wedding video and my kid videos from
    Hi8 to DVD -- I will need to get very high quality rendering, not just
    "good enough".

    Thanks in advance for any further info on this issue.

    Jay Chan
     
    Jay Chan, May 12, 2004
    #7
  8. Jay Chan

    Will Dormann Guest

    That's probably the case.

    Sounds like a plan. I use AC3 for my audio, too. LPCM takes up way
    too much space, and MPEG2 audio isn't a requirement for the NTCS spec.
    (Though the compatibility is there most of the time)

    The PVR-250 doesn't accelerate the rendering in any way. It can capture
    directly into MPEG2 format. I have one in my MythTV machine, and I
    take the recordings and re-render them (minus commercials) on my Windows
    machine. Aside from removing commercials, the other reason I re-render
    them is so that I can use VirtualDub filters for further noise reduction
    so that I can use a lower bitrate and still have acceptable results.
    (along with resampling into Half D1 resolution)

    In your case, it may make more sense to use the PVR-250 to capture
    directly into your final format. (and skipping the intermediate DV step)
    Just pick the bitrate you like and you'll be able to use the hardware
    encoder in real time. With the filters and settings I use, it takes
    about 3.5 hours to encode 23 minutes of video on my Athlon XP 1800. I
    haven't used the PVR-250 under windows, but I would assume that you will
    have control over bitrate, resolution, and other capture features.


    -WD
     
    Will Dormann, May 12, 2004
    #8
  9. Jay Chan

    Jay Chan Guest

    I have rendered an one-minute medium-to-low quality VHS video clip
    into MPEG-2 files in various rendering settings. The good thing is
    that the file size is quite small if I choose a low bitrate. The
    strange thing is that the file sizes for 352x480 and 720x480
    resolutions are very close; actually in most test cases, the file size
    for 352x480 is actually greater than that's for 720x480 -- I expect
    the file size of 352x480 should be almost half of the file size for
    720x480. See below:

    KB for Each
    Minute of
    Resolution BitRate Video Clip
    ---------- ------- -----------

    352x480 3000 21,325
    4000 27,809
    5000 33,753
    8264 56,881

    720x480 3000 20,739
    4000 27,719
    5000 34,531
    8264 54,733

    (The original video clip is 210 MB; the rendering process reduces it
    to as little as 1/10 of the original size).

    I have a couple questions on this if you don't mind:

    - Do you know why using 720x480 doesn't increase the file size?

    - Does this have to do with the fact that the video clip is only one
    minute long and may not be long enough to be a good test case? I would
    have tried a longer video clip for testing; but I don't know how long
    is long enough.

    - Does this have to do with the fact that the original video clip is
    relatively low quality (VHS video recorded at EP mode not SP mode) and
    set the horizontal resolution to high doesn't add anything new to the
    rendered video and simply got compressed away?

    Thanks in advance for any further info on this strange problem.

    By the way, I also have another post called "Put Different
    Resolution/Bitrate Video Clips in One DVD?" that is sort-of related to
    this post. Is this possible for you to also take a look on that post?
    Thanks again.

    Jay Chan
     
    Jay Chan, May 13, 2004
    #9
  10. Jay Chan

    Will Dormann Guest

    File size is a factor of bitrate and running time *only*. If two equal
    length videos use the same bitrate, they will take up the same amount of
    space regardless of resolution. Using a lower resolution means that
    more bits will be used to represent each pixel, since there are less of
    them.


    -WD
     
    Will Dormann, May 13, 2004
    #10
  11. Jay Chan

    Bariloche Guest

    As Will Dorman tells you, final file size is = Bitrate x Time. That's
    what bitrate means, after all. It's a size per time measure; bits
    (kilobits, bytes, Mb... whatever) per second (minute, hour...
    whatever).

    That's why you can know beforehand the final size of your mpeg when
    encoding in CBR or VBR mode. In the case of VBR, what applies is the
    average bitrate.

    But it's a different case when you encode in CQ mode. Then, the lower
    the resolution, the less bits you need to encode it at the same given
    quality. Only, you do not know beforehand how many bits does the
    encoding require. It may be easier to understand if you think it in
    terms of a Jpeg image. These are compressed in CQ mode. You cannot
    know beforehand the size of the compressed image from a BMP one. You
    only know, the bigger the resolution the bigger the file size you get.
    And movies are, after all, but a series of images. Only, movies use to
    be encoded in CBR to ensure the player can cope with them, and to
    ensure they fit on a disc. When those concerns do not apply, one can
    encode in CQ mode. In the case of 352x480, CQ makes more sense,
    because you are not going to violate the max. bitrate allowed by DVD
    standard. But if you need an exact file size, you would stick to VBR.
    On the other hand, at 720x480 if you want the highest quality/bitrate,
    you can do that with CBR, which needs only 1 encoding pass, because
    you cannot play with higher bitrates without surpassing the DVD
    limits. While at a lower quality (say, 6000 kbps) one would use VBR
    (which needs more than 1 pass) for the best results, at the expense of
    encoding time.
     
    Bariloche, May 13, 2004
    #11
  12. Jay Chan

    Jay Chan Guest

    As Will Dorman tells you, final file size is = Bitrate x Time. That's
    Thanks both of you to explain the reason why the file sizes are the
    same at a given bitrate. This puts me at ease.
    I don't really need to know the file size in advance. Seem like I may
    want to try CQ mode just to see if it can save me even more space. But
    there is one question: What's "CQ mode"? Does this mean "Constant
    Quality"? I am trying to find this mode in my Ulead Video Studio 7 SE.

    Thanks.

    Jay Chan
     
    Jay Chan, May 14, 2004
    #12
  13. Jay Chan

    Bariloche Guest

    In the case of Tmpgenc, it's "Constant Quality". For other encoders it
    might be "Constant Quantizer", which is about the same -only it goes
    in opposite direction. The higher the quantizer, the lower the
    quality. Quality is measured in %, while the quantizer measures how
    much information is discarded. The DivX codec configuration shows both
    quality and quantizer.
     
    Bariloche, May 14, 2004
    #13
  14. Jay Chan

    Madra Rua Guest

    I remember suggestions some time back that buying a video camera would be a
    cost-effective way to put videos onto a HD on the way to DVDs.

    Any suggestions for video cameras now? My WinTV card seems to have got some
    bug - SARS maybe. Have just reinstalled the drivers and software 5 times in
    a row and still nothing

    TIA

    Sd
     
    Madra Rua, May 16, 2004
    #14
    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.