Horizontal lines after use telecide

[Ryo94]

I'm using staxrip (which use DGIndex) to encode a music dvd (absolute garbage)

The clips are progressive, most of them TTF an the one in question ntsc BFF 3:2 (or I think so)

Screencaps of the source
http://img218.imagevenue.com/loc165/th_74035_screencaps_source_122_165lo.jpg (http://img218.imagevenue.com/img.php?image=74035_screencaps_source_122_165lo.jpg)

I used Telecide(post=0).Decimate(cycle=5) and I obtained very nice results but still getting some horizontal lines in some parts of some frames.

See the lines in mouth in the 1st pic and in the hat in the 2nd
http://img184.imagevenue.com/loc113/th_74246_output_122_113lo.jpg (http://img184.imagevenue.com/img.php?image=74246_output_122_113lo.jpg)

I'm wondering if there is any I can add to the line above in order to avoid these lines.
Because it's ntsc 3:2 material I read that should use telecide(guide=1) but after tried it I see no difference in the results.

Source sample (.vob, 16.4MB)
http://uploaded.to/?id=x0g1ks

Thanks in advance for any help.

[Guest]

If you step through the fields you'll see some pathology. Foe example, the face moves separately from the hat! That is bizarre.

One simple thing is just allow postprocessing to fix the residual combs caused by that pathology:

assumebff()
telecide(vthresh=30)
decimate()

BTW, there is no combing of the mouth in your first picture as you allege. Your second picture does show combing in the hat, which is fixed by postprocessing.

[Ryo94]

Thanx for reply neuron2,

Well, I tried that line and as result combed frames disappear but I get 2 side effects

1. Some frames looks worst, see:

Assumebff().Telecide(vthresh=30).Decimate()
http://img178.imagevenue.com/loc428/th_81394_pic_1_5postprocessing1_122_428lo.jpg (http://img178.imagevenue.com/img.php?image=81394_pic_1_5postprocessing1_122_428lo.jpg)

AssumeBFF().Telecide(post=0).Decimate(cycle=5)
http://img241.imagevenue.com/loc8/th_81436_pic_1_7no_postprocessing3_122_8lo.jpg (http://img241.imagevenue.com/img.php?image=81436_pic_1_7no_postprocessing3_122_8lo.jpg)


2. See this 15 pic sequence, the hat is alive now!!! :D
http://uploaded.to/?id=cfb1ic

[Guest]

You can't expect perfection from such a bad source. Anyway, you'll never notice that when played at normal speed.

[Ryo94]

True, thanks for you help and best regards :)

[scharfis_brain]

this most likely is a standards converted video.

provide an unprocessed sample.

[Ryo94]

this most likely is a standards converted video.

It's fragment of a large vob taken from the original dvd-9 and cutted with mpeg2cut.

[scharfis_brain]

that doesn't negate that the video itself may be standards converted.

provide a sample, please.

[Ryo94]

I don't undertstand what do you mean by "sample", I've the dvd, tell me what to do to get "that video sample" and I'll do it (as long as don't imply upload large files, I've a pour upstream), then I can upload it like I did with the 16Mb fragment that I posted above, the first thing that came to me was cut a pice of the large vob where is the video with mpeg2cut, if there is another way to get a sample from this dvd please let me know it.

[scharfis_brain]

sorry I missed that you actually provided a sample.

it was due to the code-formatting.

[scharfis_brain]

okay it seems to be some weird compositing error.

with scene by scene manual ivtc I was able to get rid of the artifacts. but that will be a though way to go...