Log in

View Full Version : Trouble IVTC/de-interlacing


NeonEva
16th July 2003, 04:19
Hello, I am trying to find good settings for this VOB segment (Gasaraki - Anime) but I am having no luck with any of the AviSynth plugins for 2.5.2 ;[

I never used ForceFilm since it was NTSC format, I tried ForceFilm anyways but still get many interlaced frames with it and it does not give ...01230123... either way in the .d2v file

I have tried many settings for
Telecide().Decimate(Cycle=5)
FieldDeinterlace()
IT(fps=24,ref="TOP",blend=true)
and a few others I seen named on this forum and on the avisynth.org plugins page.

no matter what I used I would still get ghosting/interlaced frames.
the time segnments I would mostly check for these were @ times of
1:52 - 1:18 - 2:02

here is a 92.5MB rared segment of the VOB for anybody willing to help me out

ftp://doom9:doom9@neoneva.no-ip.com:9696/test.rar

sorry for any slow speeds, my upload cap is 50k

Thanks in advance ^_~

*edit* should probally also add that I am converting this DVD to XviD for myself ^_~

sh0dan
16th July 2003, 15:56
Thanks for the detailed report and the extensive test material - this is how it should be done!

Seems like there is more than me trying to dl it right now, but I'll give it a shot. :)

NeonEva
16th July 2003, 16:20
greatly appreicated sh0dan ^_~

I was looking over the ep2 today and there is jumpiness and merged frames all over the place from the d2v, will have a more extensive look @ it later when I get home from work and if I still runn into errors whit that then I will make a VOB sample of that also ;]

DoW
16th July 2003, 19:44
Have you tried Neuron's decomb 5 beta 15 recently? I found that it worked wonders on ZoE (one of those animes that someone DREADS trying to encode).

Lefungus
16th July 2003, 20:22
With Decomb 5b15, i've found that default parameters works great!

mpeg2source("C:\Vobs\test\test.d2v",CPU=0,IDCT=5)
Telecide(order=1,guide=1,post=2,blend=true)

You could add decimate() after, but all the intro-credits are progressive so they'll be a little jerky.

sh0dan
16th July 2003, 20:57
I agree! Your material is Hybrid, with a blend of 3:2 FILM and NTSC material. Using decomb 5 (http://forum.doom9.org/showthread.php?s=&threadid=54456) I get no seriously interlaced frames, using "Telecide(order=1,guide=1,post=2)".

Considering the amount of 30fps material I'm not sure I would choose to decimate this material - it could create jerkiness in the 30fps section. If there is much more 3:2 material in you might want to try "decimate(cycle=5, mode=1)" - it seems to work ok on the material you supplied.

I'm experiencing some weirdness in telecide, that creates a stutter in the short pan scene around (undecimated) frame 3111. I'm contacting neuron2 with the problem. (Disabling post-processing will make the pattern detection for this section work - that doesn't really make sense)

NeonEva
16th July 2003, 20:58
Yeah guys I have tried neurons decomb5b5

Using "Telecide(order=1,guide=1,post=2)" on that sample I posted gives better results, but then when I use the same command on the ep2 intro, on same DVD, it gives way different results. get frames that have jagged edges all over the objects in the frame ;[

same settings, same intro(ep1 vs ep2), and different results 0_o

ftp://doom9:doom9@neoneva.no-ip.com:9696/test2.rar

gizmotech
17th July 2003, 02:49
With the second episode ensure that the field order is correct.

I've found with some dvd's depending on where the dvd is ripped it can resulting in a change between the initial field order and the current.
Example:
BGC2040, disc 1, title1, chapters 1-4

If ripped on chapter 1, the clip is order=1
If ripped on chapter 2, the clip is order=0.

Check and ensure you are using the correct order each time you rip. It can change (the evul is among you :P)

GizmoTech

NeonEva
17th July 2003, 02:56
Thanks for the info Gizmo ^_~

I will make sure to check for it.

If any of you where having troubles downloading a file or getting cut off earlier I apoligize for the inconvience, was having troubles with my connection, all is fixed now and should work fine :D

Lefungus
17th July 2003, 10:56
Gizmotech is right, the field order is swapped on your second sample. So use order=0 and you'll have similar results as with sample 1.

bilu
17th July 2003, 13:59
This way you'll have no problems with MPEG-2 sources:

function FOrder(clip c) {
fo= GetParity(c)? 1:0
return fo
}

Usage example: Telecide(order=FOrder(),guide=1,post=0)


Bilu