Log in

View Full Version : MCbob Weakness.


Terranigma
12th February 2007, 00:03
Is it just me, or is there's a weakness when using mcbob on a 25.000fps or 23.976fps source when deinterlacing multiple frames in a row that's duplicated? It doesn't deinterlaces all of them, just the first and last duped frame. Weird. :confused:
I don't seem to have that problem using another bobber such as eedibob. :devil:

Edit: Nevermind. I found a way around it. I had to change the mtnmode to 1

Didée
12th February 2007, 01:47
Your description of the problem isn't fully clear to me, however ... 'in that departement', it's not a weakness of MCBob, it's an inability. :D

A case of "quasi-duplicates" happens e.g. when there are progressive sequences in the interlaced stream. MCBob doesn't check at all for that case. That means when searching/constructing the field X in the fieldsequence AaBbXcDd, MCBob assumes that X is temporally in-between the fields b and c. But when the sequence is in fact is progressive, then X is at the same temporal location as field c. Therefore, the created compensation in this case is necessarily wrong, and one is at the error correction's mercy to filter the mistake out.
(And here, mtnmode=2 is stricter than 1, and 0 is the weakest.)

Still, I don't fully understand your description:
... on a 25.000fps or 23.976fps source when deinterlacing multiple frames in a row that's duplicated? It doesn't deinterlaces all of them, just the first and last duped frame.
For one, a 23.976fps source never should be interlaced. :)

Then, where do those multiple duplicates come from, when the source is interlaced?

In any case, it applies what I said above: if there are duplicated (and interlaced) frames (caused by frame drops, or Anime related?), then the sequence of fields is not temporally consistent.

A contrary example: when trying to process a telecined source by bobbing + decimating, then MCBob is about the worst possible choice, because the flow of fields in a telecined source is not temporally smooth.

MCBob is only suited (if at all :) ), when the sequence of fields is a straightforward, not-interrupted flow in time.