Why is the compression so much better?

[silver_cpu]

When I run a compression test @512X384 (16 W-modul X 16 H-modul), I get about 45% I think (I'm using GK). However, when I do the exact same test encode with the same filters, but at 576X416 (32 W-modul X32 H-modul) I get 104% quality? I know that 32X32 is supposed to give you fewer macroblocks, but this is an enormous difference in compressability, since I'm able to encode at twice the quality at an even higher resolution.

Also (and I hate to be off-topic), does anyone know a link to a two-pass encoding guide? I can't for the life of me get the first run to work. VDubMod starts, but can't seem to open the .d2v file or something, and the entire program hangs. This does *not* happen with DivX. I would, however, much prefer to use Xvid, as it offers very high quality, better compression, and frankly it leaves DivX in the dust, encoding 4X faster on my machine. Any ideas why this might be happening?

[mf]

ME might be better at that higher resolution. Try using qpel on that 512 encode and see if it helps.

[silver_cpu]

I am using Qpel, on both encodes. I am not using GMC on the Xvid because of the lovely VHQ feature, but I am using it on the DivX encode.

Also, if anyone has any ideas about my Xvid hangup, this would be greatly appreciated. I know that 1-pass encodes work great. Would it help if I posted exactly what I was doing, and what settings I was using?

[powerslave]

Yes, its always good to post settings. I use Gknot and so far the only problems ive had with xvid have to do with using custom matrices. Other then that, i havent had any problems doing 2 passes. Have you read snowbeach's great guide? The link is located in the newbie settings thread if your interested. Also, ive found better compressability using bilinear resize and using some filtering in your avisynth.

[silver_cpu]

I've gotten it to work now, so I don't think I'm going to go through the trial of posting all my settings :) However, it had something to do with one of the running of a compression test. This is odd, I think. I normally am able to run GK just as is suggested. However, when I ran a compression test, it output a "Movie.stats" file, which it was automatically selecting and using in the first pass of the two-pass encode. My guess is that was causing the conflict. I'm not sure, though, I'll do some more tests as soon as the current encoding is complete. Thanks for the help, I'll check out the Newbies thread.