Fixing screen shake on part of a video with Avisynth

[VideoFanatic]

I have 2 parts of an interlaced video where only the 2nd part has screen shake (I've uploaded a short test clip where the whole thing has screen shake but for the purposes of this script just pretend the 1st part doesn't have screen shake). The 2nd part needs to be progressive while DePanStabilize is working to fix the screen shake so I re-interlaced it again after that. However I noticed that the filter has no effect unless the weave line is before DePanStabilize but if I do that then it won't work properly. It's a catch 22 situation.

I want QTGMC used on the whole video and then re-interlaced and I would like the screen shake fixed on only the 2nd part of the video. How can I do that?

The tweaks are not important - it's just to show the layout of my script.

The following doesn't work when fixing screen shake on the 2nd part:

AssumeTFF() QTGMC(Preset="Super Fast") 

filtered=Tweak(Hue=10, Sat=0.7, Bright=10, Cont=0.95, Coring=False) ReplaceFramesSimple(last,filtered,Mappings="[0 10]")

filtered=Tweak(Hue=0, Sat=0.6, Bright=10, Cont=0.95, Coring=False) ReplaceFramesSimple(last,filtered,Mappings="[11 124]")

mdata=DePanEstimate(dxmax=0)
filtered=DePanStabilize(last, data=mdata, initzoom=1.05, dxmax=0, dymax=4) ReplaceFramesSimple(last,filtered,Mappings="[11 123]") 

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

[Selur]

split the video in two parts using trim(..), filter both parts separately

[VideoFanatic]

I tried that before and it didn't work. Can you give an example please?

[Selur]

here an example:

#load video
video = DGSource(...) #or whatever you use
#deinterlace video
video = video.AssumeTFF()
video = video.QTGMC(Preset="Super Fast") 
#split video
videoPart1 = video.Trim(0,100)
videoPart2 = video.Trim(101,0)
#filter second part
videoPart2 = videoPart2.Stab(range=1,dxmax=4,dymax=4,pixaspect=1)
#join videos
video = videoPart1 + videoPart2
#reinterlace
video = video.SeparateFields()
video = video.SelectEvery(4,0,3)
video = video.Weave()
#return video
return video

[VideoFanatic]

OK but how do I get this working for videoPart2?:

mdata=DePanEstimate(dxmax=0)
filtered=DePanStabilize(last, data=mdata, initzoom=1.05, dxmax=0, dymax=4)

[Selur]

mdata=videoPart2.DePanEstimate(dxmax=0)
videoPart2 = videoPart2.DePanStabilize(last, data=mdata, initzoom=1.05, dxmax=0, dymax=4)

you might want to re-read the avisynth input tutorial on the avisynth.nl side, it seems like you did not understand how to apply filters,...

[VideoFanatic]

OK so I've got this:

video = video.AssumeTFF()
video = video.QTGMC(Preset="Super Fast") 

videoPart1 = video.Trim(0, 10). Tweak(Hue=10, Sat=0.7, Bright=10, Cont=0.95, Coring=False)

videoPart2 = video.Trim(11, 249). Tweak(Hue=0, Sat=0.6, Bright=10, Cont=0.95, Coring=False)

mdata=videoPart2.DePanEstimate(dxmax=0)
videoPart2 = videoPart2.DePanStabilize(last, data=mdata, initzoom=1.05, dxmax=0, dymax=4)

video = videoPart1 + videoPart2

video = video.SeparateFields()
video = video.SelectEvery(4,0,3)
video = video.Weave()
return video

DePanStabilize is still being applied to the whole file instead of just part 2. You can see this by enabling and disabling DepanStabilize - you will see that the screen is zoomed in when it's activated.

[Selur]

videoPart2 = videoPart2.DePanStabilize(last, data=mdata, initzoom=1.05, dxmax=0, dymax=4)
->
videoPart2 = videoPart2.DePanStabilize(data=mdata, initzoom=1.05, dxmax=0, dymax=4)
or
videoPart2 = DePanStabilize(videoPart2, data=mdata, initzoom=1.05, dxmax=0, dymax=4)

[VideoFanatic]

I tried each of those line instead of the red line in my previous post. DePanStabilize is still being applied to the whole file instead of just part 2

[Selur]

you can try:
videoPart2 = videoPart2.DePanStabilize(videoPart2, data=mdata, initzoom=1.05, dxmax=0, dymax=4)
if that also still applies to the whole script I don't know either