How long to render?

Discussion in 'Video Cameras' started by Jim McLean, Sep 6, 2006.

  1. Jim McLean

    Jim McLean Guest

    Dear all, hope you can help.

    I have a DV camcorder - new, not brilliant stuff, a modest canon mvx460. I
    have quite a powerful pc with plenty of hard disk space.

    My problem is that the length of time it takes to render a video. After I
    copy my 60 minutes of home video onto the pc and edit this using say, Nero,
    or Videoworks or the package that came with the camera, I find that once I
    get this down to a 20 minute product it takes absolutely ages to complete.

    I know that the rendering process is the longest part of video making. I was
    dismayed to read in various websites and tutorials that people let this
    happen overnight.

    Is this something I need to learn to live with? The faff of trying to make a
    decent little home video is such that I am tempted just to leave it,
    transfer the lot to my hard disk DVD recorder and at least it will all be
    done in 60 minutes and after I can just ff the rubbish bits.

    What are your experiences and more importantly, what should my expectations
    be? I have a 2.8Ghz pc with 1gb RAM and 250gb hard disk space.

    Many thanks for any advice.
    Jim
     
    Jim McLean, Sep 6, 2006
    #1
    1. Advertisements

  2. Not sure if this is any help, but upgrading from Ulead's VideoStudio 6SE
    through 8 to 9 seems to have reduced the rendering time. (My PC now has a
    spec similar to yours - previously it had half the RAM, and 1/3 the
    clockrate. It was s l o w .)
     
    Malcolm Stewart, Sep 6, 2006
    #2
    1. Advertisements

  3. The trick is to render in the same format as the input file. If every
    frame has to be converted to a new resolution, colour depth,
    frame-rate etc. it's going to take a long time. Choose an output
    format that doesn't require this.

    If you DO need to change format, what's wrong with leaving it
    overnight anyway? Having edited you don't have to render Right Now.
    Save the project and set it running at bedtime, if you want to use the
    computer for something else in the meantime.
     
    Laurence Payne, Sep 6, 2006
    #3
  4. Jim McLean

    Tony Morgan Guest

    Relatively recently I got myself a notebook with the same CPU
    clock-speed as my desktop, but with a dual-core processor, 2G RAM, and a
    video card with 512M memory. Rendering seems (so far) remarkably fast
    compared with my desktop.

    But then again, when I do rendering I usually either go down the pub or
    (if the pub's shut) go to bed.

    To Jim, I'd suggest that you don't stand watching the wall for the paint
    to dry. I'm afraid it's something you have to adjust to.
     
    Tony Morgan, Sep 7, 2006
    #4
  5. Jim McLean

    franklin.sr Guest

    I remember years ago reading a computer magazine where the guy was
    extolling the virtues of this $8000 computer he'd bought. He was raving
    about how quick it was with photoshop and he said that he could apply
    an effect and walk away and have a cup of coffee and in 15 minutes it
    would done - just like that.

    Now we complain if completely re-rendering 90,000 frames (1 hours
    footage) is less than real time.

    There are of course real time rendering hardware cards available from
    companies like Matrox that complete the task in real time or better
    than real time.

    The next range of intel processors should bring this time down
    considerably...
     
    franklin.sr, Sep 7, 2006
    #5
  6. Jim McLean

    Jerry Guest

    Nah, we will be back to square one, what with people now wanting to
    shoot HD...
     
    Jerry, Sep 7, 2006
    #6
  7. Jim McLean

    Tony Morgan Guest

    In message <>,
    writes
    Snipped....
    The current crop of dual-core processors seems to have noticeably
    improved matters.
     
    Tony Morgan, Sep 7, 2006
    #7
  8. Jim McLean

    G Hardy Guest

    How long is "absolutely ages"?

    It depends very much on a lot of factors - CPU speed, memory and HDD space
    are just three.

    If you can, render the output file to a different hard drive. I'm not
    talking different drive letter, I mean a physically different piece of
    hardware (some PCs come with a single large hard drive configured as three
    different drive letters, C, D, E - that's not what we want). Ideally, you
    want the destination drive to be on a different IDE cable than the source
    drive. I don't know enough about SATA to know if the same holds true for
    that.

    Make sure your drives are being accessed as quickly as possible - with IDE
    drives you should be in UDMA5 mode.

    Another big factor is the efficiency of the encoding software. I used to use
    Sonic DVDit, which produced appalling quality renders and took ages to do
    it. It's possible, though, that speed and quality were affected by the
    issues I've mentioned above, because I didn't know about them back then.

    I never use Nero, but I just tried dropping a 38 minute AVI onto its
    "timeline" and rendering out. I'm using two-pass video encoding and AC3
    output. It's showing a 79 minutes as its estimate to completion, so
    extrapolating that to your situation (factoring CPU difference and assuming
    the use of the same hard drive for input and output), I'd suggest 20 minutes
    of footage should take 45 mins or so. That's rendering to a VIDEO_TS folder
    on the hard drive for burning later. You should always do that so you don't
    have to encode again if your blank turns out to be a dud.
     
    G Hardy, Sep 7, 2006
    #8
  9. Not sure I'd agree disk access speed is all that much of a factor.
     
    Laurence Payne, Sep 7, 2006
    #9
  10. Jim McLean

    G Hardy Guest

    It's not - I was talking about throughput speeds. Access speed is only
    relevant* for fragmented drives.

    It's easy enough to check - copy a "full" video_ts folder to another
    directory on the same drive and time it. Copy it to a different drive on the
    same IDE controller and time that. Copy it to a different drive on the
    second IDE controller and time that.

    The last one will be much lower than the other two. A chunk of this saving
    will be lost when encoding, because the OS will perform the writes in the
    background irrespective of the destination, but the saving is still there.

    (Assuming you mean throughput) On a well configured system, you're right -
    hard drive speed shouldn't be a problem - less than 10% of the time taken
    should be due to hard drive delay. But it doesn't sound like Jim (the OP)
    has a well configured system, and the whole point we're in this thread is to
    establish what's wrong. If anything - it could be that he has unrealistic
    expectations about MPEG encoding, and that his 20 minutes of edit are indeed
    taking 45 minutes, but in his opinion that's "absolutely ages". Who knows?
    We need a definition of "absolutely ages"... :eek:)


    * In my opinion. Statements like that are usually the catalyst for a URVD
    slagging match. Tony said some time ago that sequential writes to defragged
    discs are still affected by access times because the disc head moves back
    and forth to the FAT - or something along those lines - but if that were the
    case there would be no benefit to defragging your hard drive.
     
    G Hardy, Sep 7, 2006
    #10
  11. Jim McLean

    Tony Morgan Guest

    Please don't misquote me. I have never suggested that on fragmented
    disks is there a need to revisit the FAT.

    The trailer of each piece of data on a fragmented disk give the *disk*
    address of the next (subsequent) piece of data for that file on that
    disk. I have never said that there is a need to go back to the FAT -
    which gives the address of the *start* of a file (amongst other things).

    However, if the file fragments are scattered all over the disk (often on
    separate platters) then there obviously will be an overhead associated
    with the head seeking different sectors here, there and everywhere over
    the disk. Conversely on a defragged disk, there are only seeks when
    moving from one track to the next - the head movement will be small and
    occur much less often.

    It's perhaps worth mentioning that the defrag program PerfectDisk, while
    defragging, puts the "often-used" files on the outer tracks, and the
    "rarely-used" files nearer the centre of the disk. This is because the
    length of each track on the outer tracks is much longer (data-wise) and
    therefore requires less seeks to the next (and subsequent) tracks in
    each file.
     
    Tony Morgan, Sep 7, 2006
    #11
  12. Jim McLean

    Jerry Guest

    It is when you ask the system to read *and* write to the same drive
    simultaneously, which is what G Hardy was eluding to I think.
     
    Jerry, Sep 7, 2006
    #12
  13. We're not streaming it. We're loading finite chunks into memory for
    heavy processing. I'd need to see proof that in/out times was
    noticeably different with separate drives
     
    Laurence Payne, Sep 7, 2006
    #13
  14. Jim McLean

    G Hardy Guest

    I didn't - you've misread what I typed. I said "defragged" not "fragmented".

    You did - 11 months ago on Oct 7 2005.
     
    G Hardy, Sep 7, 2006
    #14
  15. Jim McLean

    Tony Morgan Guest

    That is very easy to do. Download and install SiSoft Sandra (I use
    Sandra Engineer).

    Select Benchmarks/Filesystems and
    Select Benchmarks/Physical Disks

    For each of your hard drives. You can save your own results as a
    benchmark to make comparison easier.
     
    Tony Morgan, Sep 7, 2006
    #15
  16. Jim McLean

    Andy Champ Guest

    Tony,

    a few points.

    (1) I know stuff-all about video, but a reasonable amount about
    computers. I think you may be the other way around.

    (2) If you're still using FAT based file systems you need to rebuild
    your discs using NTFS. This might require a new OS, if you're still on
    Win98. NTFS is far more resilient to things like power fails. You can
    use rebuild them without any re-installs.

    (3) The FAT doesn't (didn't!) contain the start of a file. The start is
    in the directory record; the FAT chains subsequent parts of the file
    onto the first bit.

    (4) Seeks - you're nearly right here. However, switching platters on a
    disc is very fast - it merely requires the electronics to switch to a
    different head, with no physical movement. This is why reading discs
    reads the whole of a cylinder (all the tracks of the same number on all
    the platters) before performing a seek to move the heads to a different
    cylinder. That said, you're absolutely correct that a fragmented disc
    with the files scattered all over the place is slow because of all the
    seeks.

    Going back to your earlier post: All the CRC calculation is carried out
    by the disc hardware, not by the main processor. And hard disc sector
    sizes are almost invariably 512 bytes. I don't think I've ever seen one
    as big as 4096. (A cluster is usually 8 sectors, adding up to 4096
    bytes; space allocation is done in clusters)

    That aside... see my other post.

    Andy
     
    Andy Champ, Sep 7, 2006
    #16
  17. Jim McLean

    Andy Champ Guest

    All this talk of slow edits makes me think ne of you guys may know the
    answer to this.

    I have a terabyte or so of MPEG2 video, and I want to convert it all to
    MPEG4. Rumour reaches my ears that ATI have a driver/application set
    called AVIVO that will use the power of the graphics card to do this
    several times faster than the CPU can.

    Do any of you know (a) if it's true (b) which graphics cards and (c) can
    you connect these filters into a DirectX graph (as in GRPHEDIT) so you
    can get proper control over what it's doing? I'm due a new PC...

    Thanks

    Andy
     
    Andy Champ, Sep 7, 2006
    #17
  18. I think you're ignoring the context.
     
    Laurence Payne, Sep 8, 2006
    #18
  19. Jim McLean

    Tony Morgan Guest

    I don't think so. You *did* say "I'd need to see proof that in/out times
    was
     
    Tony Morgan, Sep 8, 2006
    #19
  20. Jim McLean

    G Hardy Guest

    No, I was actually "alluding" to it.

    You're thinking of Tony.

    ;o)
     
    G Hardy, Sep 8, 2006
    #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.