analog deinterlacing (some help requested)

[Wilbert]

I captured a pal source. The main problem is that I don't know how to deinterlace it.

1) It has horizontal lines. I'm not entirely sure whether they are interlace effects, because I'm not able to remove them (tried FieldDeinterlace).

2) Contains some ghosting. I can get reasonable results using a smartbob (smoothdeinterlacer) with selecteven. The problem is that it can't be a messed up ntsc-pal conversion, since the source is a capture of tmf-awards (shoot here in the Netherlands). So I don't know the origin of this ghosting. Another strange thing is that I captured something else from the tmf-awards, which could be properly deinterlaced with "telecide".

I saved a small part as xvid (3 MB/s). If you need a larger part you will have to explain me how to makes zip files of 1 MB :) Does someone have good ideas?

http://www.geocities.com/wilbertdijkhof/Atomic.avi

[mf]

This looks like resized interlacing. Is this exactly what your capture card delivers or did you alter it in any way?

[Wilbert]

No, I didn't alter it. (capture was at 704x576)

[mf]

Edit!: Oh wait, no. My video player was being weird, my video window was too small so the playback was resized. Nevermind my previous post, I'll have a look now.

[Guest]

Post it in HUFYUV. I don't have Xvid and don't intend to.

[mf]

Originally posted by neuron2
Post it in HUFYUV. I don't have Xvid and don't intend to.
It's P-frames only, so just force the fourcc to DX50 :).

[sh0dan]

Strange clip.

It is pure interlaced. It contains blended chroma fields. It looks like somethng like a bad chroma upsampling happened somewhere.

Please to a huffyuv, unless this is actually the captured source.

Anyway - fielddeinterlace() works fine - it doesn't expose the blended chroma - however blend=false will.

[Wilbert]

Anyway - fielddeinterlace() works fine - it doesn't expose the blended chroma - however blend=false will.
I didn't try blend=false. I will look at it again.

Please to a huffyuv, unless this is actually the captured source.
The problem is that I can't upload files > 1MB (university), so you will have to wait until the weekend.

[Wilbert]

I didn't try blend=false. I will look at it again.
That helped on that part of the clip. But I uploaded another part of the same capture, where it didn't help: (again unprocessed XviD 3MB/s)

http://www.geocities.com/wilbertdijkhof/Atomic2.avi