_,,,^..^,,,_ said:
CapableMetal, don't worry, I didn't take it as a negative critic, but a comment about the sound you have the right to do.
No problem, you have the right to ignore my suggestions, hehe. At the end of the day this is your project, you go with what you feel is right and I'll enjoy the results ;)
I also thought to compress the tracks at 640kbps, but I know there are few DVD players that "don't like" AC3 tracks at that bitrate... maybe the next time I'll convert them so, but for the moment, 448kbps is the right compromise - in a dual layer DVD, thay take about 1.8GB, and leave just the empty space for a decent video compression for AVCHD...
If I remember correctly, which I sometimes don't, 640kbps AC3's will only work with AVCHD/Blu-Rays. The DVD specification only allows up to 448kbps audio. That said, AVCHD discs don't work on regular DVD players anyway, but 448kbps will work on either, so I'd agree you've made the right choice there, giving more space for the video. ;)
For the DVD, I don't know if I must use the tracks as they are, or use a lower bitrate to leave more space for the MPEG2 video - but maybe compressing at 7400kbps, the quality of MPEG2 @ 720x432 will be the same than I used for the AVC @1280x544, if not better, don't you think?
On paper, no. AVC is a newer, more complicated and efficient compression algorithm than MPEG2 (which wasn't the most efficient video compression when it was introduced!). An AVC file should always look better than an MPEG2 file encoded at the same bitrate.
That said, you're working with a letterboxed laserdisc rip that is lower resolution than an either an anamorphically squashed to 4:3 MPEG2 or a 720p (after you've added the borders) AVC file. I guess it depends on what filtering you've done and what image quality you've ended up with. Maybe its best to do some sample compressions of high-motion/low-motion clips and compare the results and see which you think is best.
I'll watch your results whichever you decide to go with!
I viewed briefly the test video done yesterday, and - as I suspected - some dropouts are still present... next version, I'll capture any laserdisc two more times with different laserdisc players (I have only the benefit of choice); then use the TOOT to eliminate dropouts... maybe as I have four captures, I could do this right now: stating that a,b,c,d are the four aligned captures, I could TOOT a,b,c, then TOOT b,c,d, then TOOT a,b,d, THEN a,c,d, and use them as the four starting point for median and average instead the pure captures...
What kind of drop-outs are you experiencing?
I ask because I usually lose 1 or 2 frames from each capture I do, and they don't get reported as dropped frames. I wonder if you're having a similar problem?