Interlaced DV to MPEG2 basic AviSynth

[newbie04]

Hello,

I usually record a tv show 5 days a week and so I use mode EP, but I recorded one in SP mode and it looks just fantastic and so I don't want to use any filters. I want to see if it looks alright when I burn and play on DVD.

My usual AviSynth script uses the Convolution3D filter and looks like this:

LoadPlugin("Convolution3d.dll")
AVISource(E:\CVD\Monday.avi",audio=false)
ConvertToYUY2(interlaced=true) 
crop(8,4,-8,12)
SeparateFields()
odd=SelectOdd.Convolution3D(0, 6, 12, 6, 8, 2.8, 0)
evn=SelectEven.Convolution3D(0, 6, 12, 6, 8, 2.8, 0)
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()
AddBorders(8,4,8,12)
LanczosResize(352, 480)

I use Panasonic DV Codec and output is RGB, so I need to convert to YUY2 first.

Overview of the chain of events:
S-VHS Tape -> ADVC-100 -> Scenalyzer -> DV file -> AVI2SVCD -> CCE 2.5 -> MPEG2

I want to try the following on an SP recording:

AVISource(E:\CVD\Monday.avi",audio=false)
ConvertToYUY2(interlaced=true) 
crop(8,4,-8,12)
AddBorders(8,4,8,12)
Lanczos4Resize(352, 480)

But when viewing it in PowerDVD there is weird jerky movements in places. Like if someone moves their hand it has almost a vibrating type motion. Do I need to separate fields again? I'm not sure what to put since there are no filters involved.

[manono]

Hi-

If you use the Cedocida codec, you can output YUY2 or YV12:

http://www-user.rhrk.uni-kl.de/~dittrich/cedocida/index.html

...so I don't want to use any filters

C3D is a filter.

Do I need to separate fields again? I'm not sure what to put since there are no filters involved.

Crop, AddBorders, and Resize are all filters. However, it sounds more like a field order problem to me. You can stick Info() into the script to find the field order at any place in the chain. The script is probably BFF, but you're keeping CCE at its default TFF. Just a guess.

Separating the fields isn't the best way to filter interlaced sources (although much better than filtering whole frames). Better is to (Smart)Bob, filter, separate the fields and reinterlace:

LeakKernelBob(Order=0)#if BFF
Convolution3D(0, 6, 12, 6, 8, 2.8, 0)
SeparateFields()
SelectEvery(4,1,2)#if BFF
Weave()

and encode for TFF.

[newbie04]

I've forgotten why I chose Panasonic DV Codec, I think its because VirtualDub needs RGB? I can't remember. I'll checkout Cedocida.

Thanks for explanation about using a Bob filter. You made me go and do more research.

So now I am looking at bobbers like SecureDeint, MvBob, MCBob, etc..

But anyway, this is what I used and it worked fine:

# S-VHS SP mode script
LoadPlugin("E:\Program Files\AviSynth 2.5\plugins\LeakKernelDeint.dll")
AVISource("E:\CVD\interlaced_DV.avi", audio=false)
ConvertToYUY2(interlaced=true) # for filters and CCE
LeakKernelBob(Order=0, threshold=3) #if BFF, Order=1 if TFF
#----filters go here-----
crop(8,4,-8,12)
AddBorders(8,4,8,12)
Lanczos4Resize(352, 480)
#------------------------
SeparateFields()
SelectEvery(4,1,2) #if BFF, (4,0,3) if TFF
Weave()


As for LeakKernelBob is threshold=3 good for a clean source?

[manono]

Hi-

So now I am looking at bobbers like SecureDeint, MvBob, MCBob, etc..

Hehe, you're a lot more patient than I am. I recommend LeakKernelBob because it's the fastest decent one, although I recognize that if you have all the time in the world, the results may be marginally better with other bobbers.

I could be wrong here, but I believe the threshold only comes into play when doing a regular deinterlace, and not a bob. In any event, I don't use anything but the default when bobbing. But I don't think setting it for 3 hurts anything. But like I said, I could be wrong.

But anyway, this is what I used and it worked fine

So, no more jerkiness? Then maybe my guess about having the incorrect field order was correct. That script turns the field order to TFF, no matter what you start with, so nothing has to be done with the field order in CCE. If you prefer BFF for some reason, a slight change in both the script and a CCE setting will allow that as well.