deinterlace before or after capturing

[codemaster]

Source is PAL, analog, and I will capture using:
- a Compro VideoMate C200 Pro usb analog video/audio analog capture stick;
- composite cable;
- CyberLink PowerDirector 9;
- mpeg2 codec.

Does PowerDirector 9 have the capability to both deinterlace and capture in real-time ?

Deinterlacing is done by:
- the capture device's driver ?
- the capture device's software ?
- or it cannot be done with the capture device's driver and/or software, and must be done after the recording, in avisynth or compressor software ?

Yadif deinterlaces better than the capture device's driver and/or software deinterlacing filter ? If true, then it is better to capture at 25i, instead of 25p... but how will every field from the analog source signal be synchronized with every field in the digital mpeg2 file that was captured ?

If I have a digital file, that is TFF, and I encode it to another digital file that is also TFF, then motion looks normal. If I have a BFF digital file, and I encode it to a TFF digital file, motion will not look normal. But if the source is not digital TFF file, or digital BFF file, but instead it is analog PAL signal, and I digitize it with the capture stick and I encode/record it with PowerDirector directly to mpeg2 TFF, is it possible that motion will not look normal, because the field order of the analog signal was not synchronized with the field order of the recorded digital file ?

[2Bdecided]

Why do you need to deinterlace? If output is MPEG-2, you don't need to.

If you want to deinterlace for something else, do it after capturing.

[jagabo]

Generally, real time hardware and software deinterlacers aren't great, and should be your last resort. Yadif is better. QTGMC is best. But, as was pointed out, there's no need to deinterlace if you're going to a format that supports interlaced video. And you can easily test all this for yourself.

[PuzZLeR]

Deinterlacing can be done at all levels you mentioned, but I too will say don't bother to deinterlace, at least not at the capture level, and not for what you retain as your source. Deinterlace - and only after capture - if you wish to use it for other formats, decoders, players, etc, that would benefit from it, such as MPEG-4 formats on a supported product.

Deinterlacing is destructive. And given that capturing directly to MPEG formats is made more possible by many capture apps by disabling some higher quality, better efficiency, features of MPEG, I can see deinterlacing as being even worse at this level. I haven't tested it, and didn't bother, but I have faith in the fact that you can test it for yourself and see as per Jagabo's post.

Yes, I too would peg deinterlacing quality in ascending order: capture level, playback level, Yadif, QTGMC.

And as for TFF/BFF, you can go from one to the other, but you'd have to know what your source is, and let your encoder know what to do. For example, I capture in HuffYUV at BFF, but when encoding to MPEG-2 DvD TFF, I make sure the offset line is 0 (instead of 1), otherwise the motion would get funky.