Best way conversion interlaced DV to interlaced MPEG-2

[yup]

Hi people!
I have as source PAL DV 720x576 25fps interlaced. I have some expirience with Avisynth and CCE and TMPGEnc. Advise me way for making interlaced video from DV to MPEG-2 (I prefer CVD 352x288 for PAL). I am very sceptical to deinterlacing. Why if possible make from interlaced source interlaced target.
yup.
My software Canopus DV decoder, Avisynth 2.5, CCE SP 2.62.

[Xesdeeni]

352x288 would not be interlaced. If that is your target, then you can either throw out every other field and scale horizontally (720x288 -> 352x288), or use a smart deinterlacer and scale (720x576 -> 352x288). The latter should have slightly better quality and less noise, so it should compress a bit better, but it will take longer to process.

If you actually want 352x576 or 720x576, then it is usually best to leave the video interlaced. In this case, you should be able to feed the encoder with this directly (scaled for 352x576). CCE has a checkbox for progressive, which you would NOT check. The resulting MPEG will be interlaced.

If you really want 352x576 or 720x576 @ 25 progressive Fps, then you can use a smart deinterlacer and then check the progressive box in CCE. I would only recommend this if you are doing very heavy compression (<3Mbps), and then I'd recommend the 352x576 resolution only.

Xesdeeni

[yup]

Hi Xesdeeni!
Thank You for reply. I want make from DV PAL 720x576 25fps interlaced source MPEG-2 CVD interlaced target 352x576. But when I use Avisynth and try SeparateFields() it is not work I get 720x576 (not 720x288). I want make folowing, separatefields, spatialsoften, weave and at end temporalsoften. I try some deinterlacer for Avisynth (Donald Graft decomb and Gunar Thalin Smartdeinterlacer) and not satisfied (i try some parameter, not default). I know that interlaced video need larger bandwidth than progressive, but my video without fast motion.
yup.

[Xesdeeni]

I'm not using AVISynth 2.5 at this point, only 2.07, but it should work the same. When I load a PAL DV video (the only PAL one I have is from http://www.tecoltd.com/enctest/enctest.htm) and SeparateFields(), I get 720x288. I always load the script into VirtualDub to test it out. If you aren't seeing the same thing, you could try adding an AssumeFrameBased() before the SeparateFields().

For deinterlacing, it shouldn't surprise you that I recommend my port of Gunnar's SmoothDeinterlacer (http://home.bip.net/gunnart/video/AVSPorts/SmoothDeinterlacer/). But remember, just because you don't think there is much motion, doesn't mean there really isn't any. A hand-held camera almost never has any shots without motion. Only if you shot from a tripod can you expect a deinterlacer to be able to do much.

If you can make a short clip available for test, it would probably be easier to troubleshoot your problems.

Xesdeeni

[yup]

Thank You Xesdeeni.
Now I need Avisynth 2.5 for some DVD to XVID backup. After I plan reinstall 2.07 and continue my expiriment. I try Your port Gunnars during I have Avisynth 2.05 and IMO this work little better than decomb Donald Graft for content with which I work.
yup.

[yup]

Thank You Xesdeeni Once more. Using AssumeFrameBased work. But I little missunderstand, why for separate field I need used progressive frame (FrameBased). It is Avisynth feature? Why AssumeFieldBased not make same thing.
yup.

[Xesdeeni]

AssumeXXXBased() tells AVISynth what format each unit (field or frame) is. A 720x576 unit is actually a frame, so it is "frame based." A 720x288 unit is actually a field, so it is "field based." Normally the decoder should provide this information, but not always, which is why AVISynth has the AssumeXXXBased() commands.

Xesdeeni

[yup]

Thank You Xesdeeni! Now understand. Before I deal with progressive video.
With kind regards yup.