Confused about dealing with interlaced video?

[Yobbo]

I know when processing interlaced footage it's important to separate the fields first, then reinterlace afterwards.

Is it advisable to do something like this to maximise quality -

AssumeBFF()
yadifmod(order=0, mode=1, edeint=nnedi3(field=2))

## (processing/denoise etc) ##

AssumeBFF()
SeparateFields()
SelectEvery(4,0,3)
Weave()

OR, is this redundant and a simple bob or separate fields suffice? Also nnedi2 instead of nnedi3 perhaps? I don't want to cut corners, I need maximal quality for this particular DV footage I have.

Thanks in advance!

[IanB]

If you are not changing the vertical dimension (height) then a simple SeparateFields() may suffice. If your filter chain has temporal dependencies then you may want to process the top and bottom fields individually. Use SelectEven() and SelectOdd() to divide them into 2 distinct clips, then Interleave() and Weave() to reassemble the processed whole frame.Function ProcessIt(Clip clip) {
clip

## (processing/denoise etc) ##

return last
}

...Source(...)
AssumeBFF()
SeparateFields()
E=SelectEven().ProcessIt()
O=SelectOdd().ProcessIt()
Interleave(E, O)
Weave()If you are manipulating the height or need field rate temporal filtering, then use the best bob your time budget can afford. Process the double rate clip then optionally re-interleave. Note :- the best bob is dependent on the source material and your personnel taste.

[2Bdecided]

It depends on the nature of the noise and the nature of the filtering.

That's not a very helpful reply, but it's a truthful one!

The most obvious visible differences between methods (if there are any) happen when either the chosen deinterlacing or the chosen denoising don't work too well.

Cheers,
David.