doubleweave and pulldown

[fmctm1sw]

I read in my avisynth documentation that my capture card is combining fields incorrectly (for telecined material) and doubleweave was a step in correcting the problem. I tried to follow the steps under doubleweave and pulldown, but when I get my 60fps video from doubleweave, I see no pattern of 5 (0,2/1,3/2,4/etc.) that recovers progressive frames. I seem to have a pattern of 10 frames. I picked 3 progressive frames out of that using the selectevery() command and ended up with 18fps material. My source is 352x480 VHS movie. Anyone have any guesses?

[scharfis_brain]

ever tried telecide and decimate provided with the decomb-package of neuron2?

[fmctm1sw]

Yes, but if my card puts fields in the wrong order, it will still be messed up right?

[Guest]

You have to be more precise about "wrong order". Do you mean with a one-field phase shift, or do you mean with incorrect field dominance? These two are not the same.

[stickboy]

Originally posted by fmctm1sw
Yes, but if my card puts fields in the wrong order, it will still be messed up right?AFAIK, the common problem is that some capture drivers output top-fields as bottom-fields and vice-versa. If that's the case, then a simple SwapFields() call should fix it.

[Inc]

Also some decoders like mjpeg via picvideo do got an internal decoding option "swap fields". For those who do capture using mjpeg for example.

[fmctm1sw]

Thanks for the replys. This is what I'm talking about:


The reason you need to use DoubleWeave first is that capture cards combine fields in the wrong way. In terms of fields, the 3:2 pulldown sequence is simply "A A B B B C C D D D ...", where "A" through "D" represent the original film frames. But the capture cards combine the fields into frames with no respect for the pulldown pattern, and you get this:

A B C D D (30fps)
A B B C D


When I do a doubleweave, no combination of pulldown (using the showfiveversions function) will give me all progressive frames.

[Guest]

Post a link to an unprocessed source clip. Then we can answer you without guessing.

[fmctm1sw]

Originally posted by neuron2
Post a link to an unprocessed source clip. Then we can answer you without guessing.

I'm trying desperately. I'm not much on websites...

edit: Ok, I need help. I do not know how to upload a 10 and 20MB avi file to a website. I have a 15MB limit on yahoo (and it won't take and avi extension) and 50megs.com has a 250k file size limit.

[Abond]

Pack it in .zip

[Guest]

As Abond says, pack it in a zip. You may have to trim the AVI a bit bit before zipping it so that the final zip file is less than 15Meg.

[fmctm1sw]

Originally posted by neuron2
As Abond says, pack it in a zip. You may have to trim the AVI a bit bit before zipping it so that the final zip file is less than 15Meg.

Ah, finally I got it up there. What a pain n the a$$....

http://www.geocities.com/mattlucy2004/page.html

[Guest]

Thanks. I've been away at an embedded systems show. When I return this weekend, I'll have a look. Please leave it up until then.

[Richard Berg]

Is it possible to put up something with more motion? I have a hard time distinguishing where there's weave artifacts on some frames. Then again, my sources are clean enough that I've never had to obsess over decombing, so perhaps the watchful eyes here can be more help.

[Guest]

There's enough movement. Blow it up by 4 times and look in the trees. :)

You've asked in the context of the pulldown filter, so I'll answer that way.

Here is the script to correctly render the video:

assumetff()
doubleweave()
pulldown(0,2)

The assumetff() is essential. Determining the field order of a video and ensuring that Avisynth treats it that way is essential when using filters that are field-order sensitive. One way to determine a clip's field order is to apply this filter:

assumetff()
separatefields()

and step through the fields. If motion is always forward and not back and forth (moving first forward and then backward), then the clip is TFF, else it is BFF. I used this method to determine that your Buddha clip is TFF, yielding the script above.