Help with proper decombing or deinterlacing

[killerteengohan]

After checking and comparing frame by frame mode0 vs mode1, they are about 99% the same, but those leftovers in that one particular scene came out so much better in that one part for whatever reason.

https://forum.videohelp.com/images/imgfiles/x7GG6FH.png

Apparently that is also enough for srestore to change it's mind about what frames to pick for that scene.

I'm beginning to wonder if it's been a mistake all these years to use the default mode=1 like I have been. It barely came out different everywhere else, but that one scene that was bothering me came out slightly different, and for the better on only those few frames that were an issue.

Thanks for the ideas and help.

[jagabo]

I'm beginning to think that there's a bug in TFM(mode=1, field=1). Consider the following:

Mpeg2Source("VTS_07_1.d2v") 
Crop(16,16,-16,320)

src = last
t0 = TFM(field=1, mode=0, pp=0)
t1 = TFM(field=1, mode=1, pp=0)

StackVertical(src.Subtitle("original"), t0.Subtitle("TFM(field=1, mode=0, pp=0)"), t1.Subtitle("TFM(field=1, mode=1, pp=0)"))

frame 39607:

[Attachment 67210 - Click to enlarge]

I used pp=0 so we could see the output of the field matcher without any post processing (deinterlacing).

The original frame has no comb artifacts. Using TFM(field=1, mode=0, pp=0) doesn't change the frame, it doesn't need to. But TFM(field=1, mode=1, pp=0), allowed to look for one more combination, decides that combination is better -- but it's obviously full of comb artifacts. That choice makes no sense. The problem happens several times in that shot.

[killerteengohan]

Makes me wonder too. Is mode=1 supposed to be the better option?