FFT3dfilter script suggestions for this clip?

[Brassplyer]

I've been playing around with the FFT3dfilter. Tried various parameters, copy/pasting scripts I've found. I've gotten it to soften the noise a bit but so far haven't gotten what I'd call really amazing results. I don't claim to have a comprehensive grasp of how all the available parameters affect the results.

Below is a link to a 3-second DV avi clip, interlaced, about 11 megs. Lots of noise. I've been working on it deinterlaced with qtgmc but am linking to the interlaced version just to minimize the download.

Is it be possible to get results comparable to what Neat Video does?

https://docs.google.com/file/d/0B0zvAZXgfLgidm1QWm9sTUVueUk/edit?usp=sharing

Screen shot from the video

http://forum.videohelp.com/attachment.php?attachmentid=18730&d=1373069396&thumb=1

All input is appreciated.

[creaothceann]

00.avs:

DSS2("00.avi") # example clip
AssumeBFF
QTGMC(Preset="Placebo")
ConvertToYV12
(save losslessly via HuffYUV/Lagarith/x264(_vfw)/ffdshow/...)

01.avs:
DSS2("01.avi")
TemporalDegrain

[Brassplyer]

00.avs:

DSS2("00.avi") # example clip
AssumeBFF
QTGMC(Preset="Placebo")
ConvertToYV12
(save losslessly via HuffYUV/Lagarith/x264(_vfw)/ffdshow/...)

01.avs:
DSS2("01.avi")
TemporalDegrain

Looking at DSS2 it appears to be a way to open the file. Why would you use that instead of Avisource or FFMpegSource2?

Do you consider TemporalDegrain to be superior to FFT3dfilter?

[creaothceann]

Looking at DSS2 it appears to be a way to open the file. Why would you use that instead of Avisource or FFMpegSource2?
AVISource is quite old (and didn't work with the example clip), and FFMpegSource2 is too much to type. ;) DSS2 is basically DirectShowSource but more frame-accurate and without audio. It's in the Haali Media Splitter (included in k-Lite / CCCP) file avss.dll.

Do you consider TemporalDegrain to be superior to FFT3dfilter?
I never used FFT3dfilter directly; though some scripts/plugins I used may have used it.

[Brassplyer]

I've been playing with TemporalDegrain some. The problem I see is that the grain that's there doesn't track or follow the image the way it does in unprocessed video - it creates an unnatural-looking "screen" of grain that stays in one place, i.e. like you're watching the video through an overlay of noise.

Do you think it's possible to get results similar to Neat Video with AviSynth?

[StainlessS]

Sounds to me that you need a Spatial filter if grain is temporally constant, a Temporal degrain filter would
see the constant positioned grain as detail. Suggest maybe try VagueDenoise on gentle settings (~1.4 -> 1.8),
and then perhaps a little gentle Temporal or Spacial/Temporal filtering, eg FFT3DFilter() or other.

[feisty2]

tnlmeans is good for this kind of terrible grain