VideoHelp Forum




+ Reply to Thread
Results 1 to 7 of 7
  1. Member
    Join Date
    Jul 2002
    Location
    United States
    Search Comp PM
    I have a winTV2000 pvr pci, when i capture to mpeg2 the res is 704x480,
    when i capture to .avi i use a resolution of 352x480.
    My goal is to make a vcd ntsc (352x240).
    well my question is which method gives the best quality on down sizing to 352x240? reduceby2, bilinearresize,simpleresize?? I have tried them all and I really can't seem to distinguish a difference. Bicubic resize, I think is really just for upward resizing.
    anyone know the real technicalities between all the different methods of resizing with avisynth???

    Thanks
    "The software said Win XP or better, so I Installed Linux"
    Quote Quote  
  2. Member
    Join Date
    Nov 2001
    Location
    Raleigh, NC
    Search Comp PM
    use BilinearResize(352x240)-bilinear is best for sizing down.
    End of Line.
    Quote Quote  
  3. It depends on bitrates you're using. If you go for very high bitrate, such as over 2.5Mb/s, Bicubic is better.
    Quote Quote  
  4. Member DJRumpy's Avatar
    Join Date
    Sep 2002
    Location
    Dallas, Texas
    Search Comp PM
    I've always been told bicubic is better, all the time. Which is it?
    Quote Quote  
  5. Member
    Join Date
    Mar 2002
    Location
    chicago
    Search Comp PM
    Originally Posted by DJRumpy
    I've always been told bicubic is better, all the time. Which is it?
    easy way to find out:
    encode the same small clip using both for increasing and decreasing size, and compare them yourself. just make sure you change nothing else about the encode except the resizing methods so that you can compare just those.
    what are you askin' me for...
    I'm an idiot!
    Quote Quote  
  6. Member DJRumpy's Avatar
    Join Date
    Sep 2002
    Location
    Dallas, Texas
    Search Comp PM
    If you use the resize option in virtual dub, and use the preview option, you can see immediate changes to what your resizing. The 'Nearest Neighbor' looks bad (jaggies). The bilinear seems to blur the picture a little. The bicubic almost seems to do the reverse, and sharpen it a bit too much.

    There are three different bicubic resize modes
    =-0.75
    =-0.60
    =-1.00
    Can you tell me the difference? I always just pick the last one. Using the preview, I can't tell a difference between these 3 modes.
    Quote Quote  
  7. Member
    Join Date
    Feb 2001
    Location
    Berlin, Germany
    Search Comp PM
    As usual it depends on the source, bitrate, the other used filters etc. etc.
    The widely accepted rule of thumb is, that bilinear should be used shrink the image and bicubic to up-scale the image.
    Bilinear Resizes blurs the image, Bicubic sharpens the image more or less, it depends on the b/c parameters. Bicubic 0,1 will probably sharpen the most images too much. It may look nice in the preview, but if you convert at a low bitrate, the encoder has a dirty job. Very sharp edges require a high bitrate to encode it properly. Sometimes you will even end with ghost images.
    For progressive sources (DVD to SVCD) usually I use Bicubic 0,0.5.
    I agree with fantomlord, just do some tests.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!