It's a bit more complicated, but still manageable.
We use DGIndex (part of the DGMPGDec package) to do the inverse telecine index and to convert the ACC audio from the vob into a wave file. I used DGMPGDec 1.2.1 from
http://www.videohelp.com/tools?tool=622 .
Then we use tMPGEnc 2.524 (from
www.tmpgenc.net) for the encoding, using the .d2v index file as the video source, and the wav file as the audio source.
I found the best quality is obtained using the rate-control mode CQ_VBR with Quality set to 9 and Maximum bitrate set to 2000 (Minimum set to 0). In general this get VGA at an average bitrate of about 1.4 to 1.7 Mbps (including 64 Kbps audio), with very good quality. Always use "Highest quality" motion search if your PC is fast enought, or "Motion estimate" otherwise. The CQ_VBR parameters can be adjusted, e.g. Quality set to 8 and Maximum bitrate set to 1200 will get you VGA at 1.0 to 1.2 Mbps.
Now, there is one small issue when using VBR: the video is encoded with variable bitrate and the audio is fixed bitrate, so the audio and the video packets can (sometimes) get very far apart in the multiplexed stream. I have not tested what happens with the various players (including Betaplayer and PocketTV) when the audio and video packets become VERY far apart, but this could cause problems if the distance between them is larger than the buffer size. This is something we are looking at. On small clips, or contents which is more or less homogenous in complexity, it should not be a problem.
Unfortunately, for some reason I cannot read AVISynth source with tMPGEnc (i.e. using .avs scripts). Not sure why. Luckily tMPGEnc can read .d2v index files!
DGIndex does a very good job at inverse-telecine. I was never able to get correct inverse-telecine results with the (very complex) inverse-telecine feature buitt in tMPGEnc.
To do it with DGIndex: Start DGIndex.exe, load the .vob files with File > Open. Select Video > Field Operation > Forced FILM. Select Audio > Output Method > Decode to WAV. Select Audio > 48 -> 44.1Khz > Mid to convert audio to 44.1 Khz (it's important, because the Pocket PC audio drivers are less efficients when dealing with 48 KHz audio, they use more CPU cycles). You can select the Audio conversion quality, but higher quality will make the DGIndex process slower. Then select File > Save Project, and DGIndex will generale the .d2v index file and the .vaw audio file that you will encode with tMPGEng.
Other interesting tMPGEng settings that work:
Setting > Advanced > Noise Reduction works well by removing spacial and temporal noise. The "high quality mode" of the filter is very slow and may not be that important.
The Setting > Advanced > Sharpen Edge works well by increasing subjective quality. I use settings Horizontal 64, Vertical 64.
If necessary, Setting > Advanced > Custom Color Correction can be used to adjust the gamma etc.
To set the parameters of eacj of those advanced settings, bouble-cluck on the setting in the list Setting > Advanced, and make sure the check box is selected to enable it.
A small detail not to forget: very often, audio from DVD must be normalized, otherwise the level is loo low. tMPGEnc does that very well and automatically: In Settings > Advanced > Audio, check "use audio edits" and click on "Setting". Then check "Change Volume" and click on "Normalize". Type 100, click OK, and when the audio has been analyzed, the normalization factor will be set for you. Click OK to close the audio settings. Generally I encode the audio at 64 Kbps in stereo or joint-stereo, but you can select a slightly higher bitrate if you have very good ears and very good earphones.