Seeking avisynth function for "blend-a-field" idea

[vhelp]

I need to blend fields from two adjacent fields (using avisynth selectevery as the field selector) if there *is* such a function in avisynth.. anybody know how ?

I have a few Disney Channel shows that sometimes come in Interlace and sometimes Progressive. The show I'm talking about is "The Suite Life of Zack and Cody". It seems that the network can't make up their minds how they want to broadcast this show--interlace vs. progressive. I figured out how to restore it back to progressive but it leaves behind "interlace artifacts" and I think it has to do with the way they apply their special *interlace-field-blend* (IFB) function to the video.

The idea (as seen below) is experimental because I'm not sure it is IFB or just a -dropped- field. So I'm playing around with different ideas to blend fields if possible but selectevery() is not the blending kind of tool -- its not a blender. Otherwise, this show (the one I have that is interlaced) is prob screwed up too bad to bother with or I will consider other alternatives. Its not that important. But you know how it goes when you start something you find yourself knee deep into debugging it.

I also have a bunch of other shows that containe similar field applications that *are* interlaced, however.

** Hannah Montanna
** Cory in the house
** that's so raven

The Theory:

T0 T2 T4 T6 T8 T10
B1 B3 B5 B7 B9 B11

The restore to 30p is: T0B3-T2B5 .. .. ..
And the theoreticle blend-restore might go something like this: B1T4

And then figure out where to apply this inside the fields, above. Mind you, its just a theory I've been conjouring around with. I have a feeling that the B1T4 fields might have to be dropped all-together, and the blending part might be something more or less like: B3T2 and then insert them back into the t0b3t2b5 group, maybe something like t0b3BFt2b5 or t0bBFb5 or some other ideas to ponder over.

Any help on an already existing avisynth plugin or filter (that I'm not aware about) that just blends fields would be appreciated.. else I will prob have to code something raw to do this, but I rather not. And it may not even work anyway.

edit -- oops. I think my logic backfired on me. It looks like I actually need an un-blend function for those fields that have the interlace artifacts. fwiw, I'll leave the remaining topic un-touched.

Thanks guys,
-vhelp 4900

[jagabo]

Applying:

Blur(0, 1.0)

to a woven pair (frame) is the equivalent of VirtualDub's Blend Deinterlace. I like to sharpen up again afterwards:

Blur(0, 1.0).Sharpen(0, 1.0)

[vhelp]

Thanks for your suggestion.

Yeah, I almost forgot, my source is analog noise. So it makes it that much difficult to deal with. But I think that the source is too problematic to fix, though something like virtualdub's PAL filter did help a little since that function has a blending algorithm built in. It also can act as a un-blender too depending on how its used. But, in the case of this source in question, I think I will need an adjustable blend/deblind function. The bob in vdub didn't do much if anything though its adjustable. I think that the blend/deblend is the criticle part here.

I think that what I will do is take apart one of my image compare routines and alpha-nize some routines to blend/deblend and see how that goes. At this point, its more of an educational quest now.

Next week (this friday) Disney kicks off their new-er (and now called) "The Suite Life: on deck". I am curious to see if they will broadcast it as interlace or progressive. If interlaced, then I'm pretty sure that it will entail the same consistant T0B3-T2B5 type interlace nonsense as is in the other shows that I mentioned, though they do seem a bit more cleaner at this Restore-30p application.

That Disney channel shore can confuse us. Some shows are interlace and others are progressive. I just don't get it why they make such nonsense decisions like that. Same casts, same sets, same camera equipments, same everything -- almost. Makes no sense.

I'll post some pics anyway, in case anyone is curious what all the fuss is about

If I find an acceptable solution to my current source, I'll post an update to that.

I will keep your Blur(0, 1.0).Sharpen(0, 1.0) tip in mind for other projects. Thanks..

-vhelp 4901