Hi again, Howell --
Sorry. I only skimmed your last note and forget to reply.
> You haven't really commented (at least I haven't seen it), though, on the MainConcept codec or its effect on the final output.
No, I haven't really used it enough. From what I've seen, though, it seems competent enough as a compressor. The sad thing is that we're limited to the (very few) encoding parameters provided by Adobe Media Encoder. Personally, I like CCE. I like options :)
> Like I said, we can get a good AVI file from several different methods. It seems to me that it is the final transcription that is giving the most trouble.
I can't really agree with this in principle (though I've seen neither your .AVI nor MPEG output, so I really don't know). I think the hardest part of the process is getting a good interlaced 720x480 avi (or 720x576 for our PAL friends).
Sure, there's a big difference between good and bad MPEG encoding but to me the larger problem is the pre-processing of the HD material. High-quality deinterlacing/bobbing, scaling and weaving are probably the most important elements in the HD->SD process and, incidentally, some of the weaker areas of PPro.
There are other important things, too -- such as noise reduction, sharpening, softening -- which are situation-dependent. It's also important to limit luma/chroma values to "legal" range.
My position is that it is not only important to apply these to your video, but to maintain control over the exact order in which they are applied. Some things (like noise reduction and sharpening) I find best to apply to the HD resolution image -- but sharpening can only be applied successfully to a progressive frame (noise reduction, too, well maybe... that's another topic).
Other things are best applied to the final resolution image, such a mild vertical blur to reduce interlacing artifacts when needed. Again, this will work best on a progressive frame -- after scaling but before weaving. Chroma/Luma limiting is best applied at the last stage -- if only to save on system resources, as there's not really much benefit to doing it earlier in the process.
This, of course, is my process. Others may (and doubtless will) argue otherwise.
I believe that, given great looking input, most MPEG2 encoders will provide "decent enough" results. Have you ever seen some "ripped" DVDs made with DVDShrink or similar programs? All of my insticts tell me that these should look absolutely horrible, but I've been quite surprised at times. The reason they look acceptable is that the orignal material was spot-on. Sure, there's some blockiness on fades and mosquito noise on flat surfaces but it's not really that distracting. I've even seen surprisingly good results from DivX to DVD transfers.
I'm not suggesting that we set the bar as low as bootleg DVDs, but I think it's a good example of how -- if all else is right -- our eyes can forgive a little crappy compression :)