jagged edges when converting captured avi to dvd.

[nick101181]

Here is a pic with an example. I think this has to do with an interlacing issue but I'm not sure.



I was givin this script by fulci lives and it proved to definatly clean up a lot of my vhs sources when I convert the avi's

LoadPlugin("C:\Program Files\avisynth v208\mpeg2dec.dll")
LoadPlugin("C:\Program Files\avisynth v208\convolution3d.dll")
AVISource("D:\Captured\Summerslam 1991.00.avi")+AVISource("D:\Captured\Summerslam 1991.01.avi")+AVISource("D:\Captured\Summerslam 1991.02.avi")+AVISource("D:\Captured\Summerslam 1991.03.avi")+AVISource("D:\Captured\Summerslam 1991.04.avi")
SeparateFields()
Convolution3D (0, 6, 10, 6, 8, 2.8, 0)
Weave()
LanczosResize(352,480)
Trim(318,283804)


I read not to de-interlace it cause you loose quality. A test clip I made using this script and tmpgenc with the noise filter on @20,1,20 and I think de-interlaced double field came out crystal clear. If I'm doing something wrong I would like to know since I don't have to de-interlace it. the above pic was not de-interlaced and it was encoding in CCE with a average bitrate of 3550 kbps and a max of 9000 kbps. thanks

[johns0]

I get those lines too when i encode at 352x480,doesnt matter the source or bitrate,if anybody knows how to get rid of that horizontal resize effect then i like to know also.

[nick101181]

*Bump* Anyone have any ideas to help myself and John out?

[ronnylov]

Convolution3D is a filter that use temporal information (differencies between frames) and spatial information (nearby pixels in the same frame). When you use separatefields you split the fields so it plays upper field, lower field, upper field, lower field and so on. But the pixels of the upper field is not aligned with the pixels of the lower field (they differ one pixel on the full frame).

There are two solutions depending on if you want to deinterlace or not.

1) Deinterlace first and use convolution3d on the deinterlaced video

2) Separate the fields and make one video with only upper fields and another video with only lower fields. Apply convolution3d on both videos and then reconstruct the interlaced frames again. This method is best if you want to keep it interlaced.

You wrote that you wanted to keep it interlaced so I would modify the script as follows:

LoadPlugin("C:\Program Files\avisynth v208\mpeg2dec.dll")
LoadPlugin("C:\Program Files\avisynth v208\convolution3d.dll")
SegmentedAVISource("D:\Captured\Summerslam 1991.avi")
SeparateFields()
odd=SelectOdd.Convolution3D (0, 6, 10, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D (0, 6, 10, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
LanczosResize(352,480)
Trim(318,283804)