Direct MPEG capture vs. Conversion

[lantern]

I have been experimenting recently with capturing VHS directly to mpeg in ATI MMC (I follow LordSmurf's guides pretty closely) and capturing to avi with VDub (352x480 Picvideo codec setting of 20 for quality, I get about 2.3:1 compression). I then convert using TMPGenc to convert to mpeg.

I have noticed that in my opinion, the avi -> mpeg route provides clearer and sharper captures. When I capture directly to mpeg (MMC), I sometimes end up with macro blocks and less clarity (I have played with the videosoap settings and have also turned off videosoap).

I was wondering about other opinions concerning direct to mpeg capture and avi->mpeg conversion. Are the MMC codecs and the TMPGenc codecs different? Is it just an issue of direct to mpeg capture? Is 2-pass VBR conversion that much better than direct capture to mpeg?

[admin]

MMC uses Ligos GoMotion, which is entirely different from the Tsunami engine in TMPGEnc (the "T").

Two pass VBR will yield better quality than a direct MPEG capture, in most cases. It first analyzes the video, then encodes it based off the analysis. For an active MPEG encoder to do this, it would need a huge buffer, or some other means to see into the future.

Ligos is actually much better encoder, and is part hardware, part software on the card. The LSX encoder for the PC also uses Ligos as the engine, an older encoder, but still quite nice.

The easy way to fix MPEG compression artifacts on live capture is to pump up the bitrate a bit, and then to experiment with the IPB settings.

In actuality, the Tsunami has a way of softening the video, so it is probably just glossing over the same errors you saw from MMC files, just that MMC chose detail over softness.

[lantern]

Would you suggest capturing with MMC at a high bitrate, then running through TMPGEnc with two-pass VBR to fit to disc?

Even with the higher bitrate, the file would still be considerably smaller then an avi with 2.5:1 compression

[admin]

If anything, if space is a problem, I would capture at 704x480 MPEG-2 at 15MB/s with one I and two P (no B) frames, and then re-encode with Procoder to lower bitrate/resolution ... that's if you're not happy with MPEG-2 captures and do not have room for AVI captures.