Best point to deinterlace when going from mpg to DV avi

[brassplyer]

Going to go from an .mpg to DV avi, using Virtualdub. I also anticipate using the Thalen Deshaker in Virtualdub and it seems the best results are gotten with deinterlaced video.

Any problems with deinterlacing going from mpg to .avi? Should I go to avi first and then deinterlace? Will it make any difference? Using the MainConcept DV codec.

Any advantage going to uncompressed as an intermediate step?

Thanks.

[jagabo]

You should use AviSynth to bob deinterlace with QTGMC(). Then deshake in VirtualDub. Finally reinterlace with VirtualDub for DV. Or you could do all the filtering in AviSynth:

Mpeg2Source("filename.d2v")
QTGMC()
mdata = DePanEstimate()
DePanStabilize(data=mdata)
SeparateFields()
SelectEvery(4,0,3)
Weave()
ConvertToYUY2(interlaced=true)

[guns1inger]

DV AVI is interlaced, so de-interlacing is probably going to make things worse. If you go to Lagarith or Huffyuv in an AVI container you might get better results.

What will your final format be ? If it is not DV then I wouldn't go there at all. Use a lossless codec (again, lagarith or huffyuv) as your intermediate. Open the mpg in virtualdub directly, or through an avisynth script that does your deinterlacing for you. Reducing re-encoding to an absolute minimum.