WDM + segmented capture?

[reepa]

I'm in a serious need for one. I'd like to record a movie tonight and I'm stuck with a FAT32 file system. I have a WinTV Go tv tuner card (with the 881 chip), for which I cannot find VfW drivers. I can't use VirtualDub or AVI_IO for capturing because of that. I can't find a program that supports WMD drivers and segmented avi either. Any help would be appreciated. The movie starts in 1 and a half hours and it doesn't have commercials so I can't cut it in a clean way manually :(

Thanks

[ppera2]

Ahh... this is first time that someone complains when no comercials :) .

Iuvcr can record to segmented files. It has WDM support. Shareware, but it will work enough for you that record that show tonight :D

http://www.iulabs.com

[reepa]

Thank you so much! Too bad I had to record Star Wars on vhs... but next week I can record Empire Strikes Back with full resolution :D Anyway, I captured the vhs record on computer. It's pretty noisy and stuff so I need to figure out good postprocessing settings.

[ppera2]

You can remove noise, but never will get sharp picture from VHS.

[reepa]

I experimented a bit. I compressed the same part of the movie with different filter settings with divx5 (1300kbps). Since the movie was not interlaced, I assumed it was sent with half the resolution of PAL, since both the fields are identical. My VCR manual says its horizontal resolution is around 200 lines.

The most logical thing to do would be to just use 2:1 high quality resize. It effectively removed noise and there were very little artifacts.

I also compressed the clip with full resolution without any noise reduction filters. The result was not that good with the noise causing artifacts.

The next combination was full resolution + blend deinterlace. It effectively removed all noise while still keeping the resolution.

I tried a bunch of other combinations too, such as halving the height, halving the width, halving the width plus discarding the other field, discarding the other field while keeping the horizontal resolution, and halving the width combined with blend deinterlacing.

Discarding the other field ended up looking the worst. It made the noise and grain very coarse. Halving the width made the artifacts more visible when it was played back. Also, the noise caused artifacts.

2:1 high quality resize, half width, full resolution + blend deinterlace, and half resolution + blend deinterlace ended up looking the best. 2:1 resize had the least amount of artifacts but the playback resizing was quite noticeable. Full resolution + blend deinterlace smoothed the noise out almost totally and it looked very good when compressed in divx. I'll post some pics of the tests I made. The black bar on the bottom is for the subtitles. Resolution: 768x384

Original lossless compression pic:

Full res + blend:

One quarter size:

Half width + blend:

Half height:

Full res with no processing:

Half width

Discarded other field:

Half width + discarded field:

Now I have to figure out how to append all the parts of the movie (they all have a little different framerate, something like 24.9992 or 25.0011 so I have to convert all the parts into 25.000 fps) without having a visible seam where the joint is. I was wondering if I should demux the sound, combine the all the sound files, then compress them separately, then fix the videos.