[djenyc]

Hmm. That's interesting. Can you upload a sample clip? 2 minutes is long enough.

By the way I sometimes use CQ 1 pass. For most of the Bluray rip so that way I could keep resolution high and still have nice looking video. I don't give a crap what filesize I get at end. I'm just glad with quality I will be getting. (Not only that, but it's also faster too, just like you stated.)

I think I should figure out what bitrate each CQ number would spit out.

Yes, I think Average Bitrate (2pass) is for burning discs, where as Constant Quality is more appropriate for a media server where you don't need to hit an exact file size. In addition to faster encodes, Constant Quality will use bitrate appropriate to each file (e.g. high motion action material will get more, talking heads in front of a wall will get less) and also take video resolution in to account.

I've uploaded a zip file with several samples at different quality settings here:
http://www.sendspace.com/file/o20ueu

They are only 10 seconds long, but hopefully will give at least some idea on what to expect:

For file size vs quality, for the 10 second clip with no sound:
Q20 - 1967kb
1278kbps (2pass) - 1600kb
Q21 - 1426kb
Q22 - 1015kb
Q23 - 728kb
Q24 - 540kb

A 1:28 movie encoded with x264/Q22 and AC3/6ch@224kbps ended up at 600mb. Originally I encoded it with x264/2pass 1278kbps and AC3/6ch@448kbps and it was ~1250mb.

I've seen several people recommend Q22 for high quality encoding, while others were using Q24-Q26. I found no difference between Q26 and Q24 while watching movie on 27" CRT TV, but could see slight difference between Q24 and Q22 when watching it on a notebook screen (Q26 was bad), so I decided on Q22.

For sound - I agree with the guide - AC3 is the only way to go. I tried Nero AAC-LC first, and while it provided excellent compression, decoding was using up 20% of XBOX CPU and caused major frame dropping. Where as, I got no dropped frames with 6ch AC3 (75-95% cpu utilization total).