The N810 has a screen resolution of 800x480 and can display 65.536 colors only.
So first thing to do is check results with this setting, no million colors needed for this device, they'll get discarded/interpolated to what comes closest within the 65.536 colors range by the GPU. Save bandwith to what is really displayable.
As for best resolution, I'd use 800x450 for 16:9 content and 640x480 for 4:3 content with square pixels (Pixel-Aspect-Ratio=1). This will give you small black bars at top and bottom for 16:9 or at the sides for 4:3 content though, but should give best performance on device.
If PAR is 1 it's "native" to the display so 1 clip-pixel matches exactly 1 screen-pixel.
Your 720X480 footage has a PAR of ~1,185 if it's 16:9 content (Screen-Aspect-Ratio=16:9=1,778) which means pixels get stretched which is not native and needs interpolation done by device.

Also, why you want to use xvid.avi, it's not even listed in the specs?
Expect best performance and hardware-acceleration on any device with the codec the videocamera uses.
For the N810 that's H.263, MPEG-4.