How do I capture SL footage to edit in with RL footage?

I'm creating a RL movie using digital video. I'd like to use SL/Machinima for a dream sequence. Does anyone have any tips on how I should capture the SL footage to help ensure that the resolution/quality can be integrated with my RL movie and also look good in a theater screening?

You need a really, really good computer. Crank all the graphics all the way up. Etc.

If you are showing it in SL, then you really don't need better quality than a standard monitor resolution. If for RL broadcasts, you only get 32bit colour depth from SL, and the colour pallets lack full luminance and even bump mapping abilities. I grab footage at 1600x1200 at 30fps and apply bumpmapping and luminance using masks (see below for more on masks). You will never match RL footage.

If you take the SL created machinima into your editing software (I prefer Sony Vegas) you can use overlay masks. Asign your RL movie as the main movie, and apply subsequent layers of movies as overlays.
Apply a 'bounding' box mask around your SL movie, use a 'Chroma Key' colour for the mask, then when you apply the Chroma Keyer to the mask, it will remove anything created by the bounding box mask. Use a feather edge, to blend it, or apply multiple bounding boxes that overlay to simulate say clouds.
Then using the pan/zoom facilities for the overlay, you can size it and move it to the location you need it, you can also use 'morphing mask' to fade, move and resize that overlay.

It sounds complicated, but a very useful technique to use and learn. If you move the position of the overlay, say to make it rise up from the head of the 'sleeper', then you will need to use the Ken Burns technique to move each mask.

In Vegas, all overlays appear in a single window, so you can not only see the keyframes for each mask, but switch between them, to make adjustments, the 'onion skin' markers of each mask, can be seen clearly in the mask window timeline, so timing is easy to align.