Is there anyway to properly frame match this video so there is no blending.

[LaserBones]

I've tried tfm() with srestore(). I'm just not finding anything that works. It's the only episode of the entire DVD series that seems to be poorly encoded.

[Sharc]

I've tried tfm() with srestore(). I'm just not finding anything that works. It's the only episode of the entire DVD series that seems to be poorly encoded.

Double blended fields. I don't think there is a real fix for this.
Maybe try

qtgmc()
srestore(frate=23.976)

It's about the best I got. No real fix though.

[Selur]

The blending is already in the fields.
Sometimes the bottom and sometimes the top field and sometime both has the blending in it.

=> decimation won't help and I doubt there is a way to remove the blending

Cu Selur

[poisondeathray]

Yes, there is only 1 clean field in 5, so you would expect to have residual blends

If pattern is consistent in this clip for the whole episode, you might be able to improve srestore results by subtracting an offset frame by some %, then adjusting the colors/levels to match . After srestore on this clip, every 2nd frame is still slightly blended

Depending on how the sample was cut, and if the pattern is consistent - you might have to adjust it and possibly include some scenechange logic. You might have to use other types of color or levels matching such as matchhistogram, colourlike etc... you have test it out

BWDif(field=2)
Srestore(23.976)

o=last
t=o.trim(1,0)

overlay(o, t, mode="subtract", opacity=0.15)
levels(0,1,218,0,255,false,true)
ov=last

interleave(o.selecteven, ov.selectodd)

[LaserBones]

Yes, there is only 1 clean field in 5, so you would expect to have residual blends

If pattern is consistent in this clip for the whole episode, you might be able to improve srestore results by subtracting an offset frame by some %, then adjusting the colors/levels to match . After srestore on this clip, every 2nd frame is still slightly blended

Depending on how the sample was cut, and if the pattern is consistent - you might have to adjust it and possibly include some scenechange logic. You might have to use other types of color or levels matching such as matchhistogram, colourlike etc... you have test it out

BWDif(field=2)
Srestore(23.976)

o=last
t=o.trim(1,0)

overlay(o, t, mode="subtract", opacity=0.15)
levels(0,1,218,0,255,false,true)
ov=last

interleave(o.selecteven, ov.selectodd)

This is amazing. Thank you so much. It's going to take me a minute to wrap my head around this. So this is actually removing blends by subtracting with an overlay. Why is the color/levels adjusted in that process? Also, is there a specific reason to use BWDif as a bob deinterlacer compared to another deinterlacer?

Again, thank you. This is why I love this forum.

[poisondeathray]

Why is the color/levels adjusted in that process?

When you subtract something, the value becomes lower - so you expect the picture to become darker in the Y channel.
A compensatory adjustment is required for the adjusted frames, otherwise there will be a brightness change between adjusted frames and non adjusted frames, producing "flicker" . You can comment out the line and see difference.

Also, is there a specific reason to use BWDif as a bob deinterlacer compared to another deinterlacer?

You can use any double rate deinterlacer. I often prefer BWDIF prior to deblending operations over QTGMC in many cases, because QTGMC can produce additional blending and ghosting.


Also, if the pattern is constant for the whole episode, it's probably better to use a SelectEvery call, than Srestore - because there is a chance SRestore might pick the wrong frames in some cases