Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

Improving MPEG Quality

  • 19-03-2003 8:30pm
    #1
    Moderators, Computer Games Moderators Posts: 2,972 Mod ✭✭✭✭


    As the heading says.. I've got a load of D&D cartoons in MPEG format and was wondering if it's possible to improve the picture after they've been encoded. Thanks


Comments

  • Registered Users, Registered Users 2 Posts: 23,212 ✭✭✭✭Tom Dunne


    As mpeg is a lossy compressor (i.e. it throws away data during the encoding process), I would be very doubtful you could improve the quality.


  • Registered Users, Registered Users 2 Posts: 1,393 ✭✭✭Inspector Gadget


    ...the only way you could realistically improve the quality of your MPEG, say, is to re-encode it at a higher bitrate from a losslessly compressed/uncompressed (i.e. bloody big), or if that's not possible, higher bitrate but lossily compressed copy of the source. As it's unlikely you have a higher quality copy of the file lying around anywhere, you can't do that much with the video you've got, short of things like colour/brightness/contrast/gamma correction, softening/blurring it slightly (to make blocky MPEG artefacts less obvious) or stuff like that.

    One point that may be interesting is that MPEGs that can look pretty rough on a computer screen can look better when burned onto VCD and viewed through a TV with a DVD player or something - I think it's a combination of the low (but fixed) -res display, the much higher contrast ratio a TV offers over a typical monitor, and the fact that you've stepped back from it a bit...

    HTH
    Gadget


  • Closed Accounts Posts: 2,525 ✭✭✭JustHalf


    One thing you could do is vectorise the feed and re-encode it at a higher resolution. This was going to be the basis for my fourth-year project.

    You're welcome to give coding that up a try! :)


  • Registered Users, Registered Users 2 Posts: 1,393 ✭✭✭Inspector Gadget


    :eek: x 10^100?

    Gadget


  • Closed Accounts Posts: 2,525 ✭✭✭JustHalf


    What's that for?


  • Advertisement
  • Registered Users, Registered Users 2 Posts: 1,393 ✭✭✭Inspector Gadget


    Originally posted by JustHalf
    What's that for?

    The idea of trying to faithfully vectorise a blurry, low-res (= low detail) bitmap image from an interlaced video source with inconsistent colouration (as analogue video is wont to do) that's already been quantised/resampled from a larger-format, already less-than-ideal source into one with accurately described vector shapes you can enlarge? (of course, you'd also have to be able to do it for 25 or 30 frames every second for a relatively large number of minutes - I'd imagine this would give a whole new meaning to "processor intensive") ;)

    Gadget
    P.S.> I know you were only kidding - so was I...


  • Closed Accounts Posts: 2,525 ✭✭✭JustHalf


    Originally posted by Inspector Gadget
    The idea of trying to faithfully vectorise a blurry, low-res (= low detail) bitmap image from an interlaced video source with inconsistent colouration (as analogue video is wont to do) that's already been quantised/resampled from a larger-format, already less-than-ideal source into one with accurately described vector shapes you can enlarge? (of course, you'd also have to be able to do it for 25 or 30 frames every second for a relatively large number of minutes - I'd imagine this would give a whole new meaning to "processor intensive") ;)

    Gadget
    P.S.> I know you were only kidding - so was I...
    I wasn't kidding. :D

    If you think about it, you'll see that this is not as hard as it looks. Cartoons tend not to change much every frame. Now, if you have a really low-quality MPEG, there WILL be a much greater change per-frame due to compression artifacts. It *should* be possible to reduce this substantially by means of applying a filter which assumes that minor changes in colour value are merely the result of compression artifacts, and were not in the original feed. An average colour value could be derived and applied. After a cleaned-up feed has been created, this could be vectorised and resampled at a higher resolution.

    This uses certain features of cartoon video that are not present in live video, to work around some of the problems with using a codec designed for something (live video) other than what it has been used for (cartoons).

    I might develop this sometime in the future. For now, it's one of those up-in-the-air ideas.


  • Closed Accounts Posts: 1,571 ✭✭✭Mailman


    Suggestion: Play them through a hardware decoder e.g. a good board version of the DXR2 or DXR3 and/or look at them on TV rather than on a monitor.
    Hope that the pass through cable doesn't disimprove the picture on monitor for other applications.

    With software decoder you'd have to look at tweaking largely undocumented registry settings to set the trade-off in the player between quality\performance more toward quality than performance.


Advertisement