Following Red-M's KVCDx2 guide. When I get to point of running BBMPEG with MPG from TMPGEnc and MP2 from Headac3he, I get "FILE d:\temp.mpg is not a 11172-2 or 13818-2 Video Stream".

Where did I screw up?

You're supposed to multiplex an m1v file and an mp2 file to get an mpg file. It seems that you've accidentally encoded the video straight to an mpg file. Demux the m1v file out of the mpg file with TMPGEnc's MPEG Tools and the mux the correct files.

Thanks.

How should I have encoded to produce the m1v file directly?

If you encode the video in TMPGEnc, choose ES (Video Only). If you encode both the video and audio in TMPGEnc, choose ES (Video + Audio). You can see that the output file will have the extension .m1v or [.m1v+.mp2] in the TMPGEnc main window.

Gotcha. Two questions:

When I use the AVS file from FitCD the second time to produce the the M1V file in TEMPGEnc (ES Video Only), do I modify the GOP settings again?

Firstly, click Setting, then the Advanced Tab and change the GOP to 1,15,2,1,15 for PAL & 1,18,2,1,18 for NTSC, depending on your DVD.

My film is 16:9 NTSC. I used Force Film in DVD2AVI. Do I change the advanced screen in TEMPGEnc to 16:9 and Full Screen (keep aspect ratio 2) for both the sample and final runs?

Firstly, you should use the new GOP developed recently, it's 1-12-2-1-24 for both PAL and NTSC. Also uncheck 'detect scene change' as it's currently recommended. You'll have to use the new file size prediction formula which you'll find in the prediction forum.

The GOP settings stay the way they were the last time you used the program. They will only change if you load a template that has a different GOP or if you change the values manually. I never load any templates as I enter all the changes manually so I don't have to touch the GOP settings unless Kwag comes up with an optimization once more

What comes to the second question, I'm not sure. However I think that the source aspect ratio really doesn't matter as you won't be resizing in TMPGEnc, you're doing that in the Avisynth script. Personally I would leave the field as it is after you load the script in TMPGEnc and select 'full screen' as the arrange method. This is what I've used. You'll see if it's all wrong when the preview is shown while encoding.

The sample and the final encode should naturally have the same settings.

So I use 1-12-2-1-24 for both PAL and NTSC for TMPGEnc for both the size prediction runs and the final run?

Yep, that's the correct procedure.