For the years when I have done projects in DV, I've used the excellent CinemaCraft MPEG2 encoder to make DVD-compliant m2v files which I would then import to either Maestro (an excellent but ancient authoring program) or more recently to DVDlabPro. For the past year or two I've migrated into shooting HDV, which adds another layer of complexity.

For the past few days I've looked at a couple of paths that all start with HDV and end with a DVD. Here are my initial observations:

(1) I have edited in m2t, in Mjpeg, in Cineform, and in DV. On my fairly powerful machine (HDV rendertest in 6:21) I much prefer the smoothness and fluidity of working in the DV environment. Multiple layers and color corrections don't phase the playback one bit. I have not played with the purchased version of Cineform, but the default Cineform does not run all that well on my machine and file size is also a minor consideration. Mjpeg works okay but shows weird artifacts in the preview window if the project is set to HDV. Working directly in m2t can be done, but there are times the system really seems to be laboring, and the frame rate slows down considerably with the addition of various effects. So my present preference is to pull in the raw m2t clips and immediately GearShift them into DV proxies, and then work with the proxies.

(2) I've also learned (as no doubt many of you have) that you can encode into MPEG2 from DV, but you pay a price doing so if you live in an NTSC world. NTSC DV is 4:1:1, meaning that color sharpness is excellent in the vertical direction but horrible in the horizontal direction. Okay, fine. But when you encode to MPEG2 the vertical color resolution gets cut in half, leaving you with medium color resolution vertical and horrible resolution horizontal.

If, on the other hand, you encode to MPEG2 with any of a number of other formats (raw m2t, mjpeg, huffYUV, uncompressed, ...) then you start with excellent color resolution and wind up with medium color resolution in both vertical and horizontal directions. I didn't think it made much of a difference until I encoded red objects, which look pretty horrible in 4:1:1. So this means that while I might like to edit in a DV proxy world, I need to render to MPEG2 from a different medium.

(3) I've always enjoyed using CinemaCraft as my MPEG2 encoder, because (a) the results are terrific (b) the encoder is blindingly fast and (c) you have the ability to graphically fine-tune the bitrate for any segment of the render. In my various tests performed over the past few days, however, I've come to the conclusion that the MainConcept encoder produces results virtually identical to CinemaCraft at bitrates down as low as 3Mb/sec (CBR). In fact, at the extreme bitrate of 2Mb/sec (that no one would ever use) the MainConcept CBR output looked BETTER than that from CinemaCraft.

The blazing speed of CinemaCraft is great. I routinely got 3.8x realtime when encoding from DV. However, if I use CinemaCraft to encode from an m2t master using frameserving, that speed advantage is largely lost, since it appears it can go no faster than what Vegas is able to hand over to the frameserver. So the speed advantage is largely gone. Still, CinemaCraft gives much more encoding flexibility than MainConcept.

Anyway, at this point my muddled brain has concluded this workflow.
--Shoot in HDV
--bring m2t clips into the PC
--make DV proxies with GearShift
--edit in DV on an HDV project timeline
--for final render, GearShift back to m2t clips and render with CinemaCraft via frameserver (render with MainConcept from m2t would be nearly as satisfactory)

I think it makes more sense to render from the m2t source directly because while I like to edit in the DV environment, DV can't be used as a source for the final render, so I might as well go to the raw m2t source. Also, I can avoid any generation losses this way.

There are many other ways to get to the final destination but this is one way I'll try on my next project.