Worked out most of the painful details, here they are:
-Use the same camera and rendering setup as that used for making your cube-faces.
-Remember the direction your camera is facing when selecting a spot to animate. You have to put ‘NORTH’, ‘SOUTH’, etc in the spot setup.
-Once you’ve set up your animation, check the ‘border’ option, but don’t check the ‘crop’ option (yet).
-Hit Shift-B and draw the region you want to spot. Here’s the hard part. You only have visual control for selecting the region, but it needs to be in dimensions that are codec friendly. Not sure what those are, but so far any even-number seems to work.
-Render one frame. In the render output window, click-drag with the mouse to locate the exact x,y location of the upper left corner of your spot image. Remember those values for your script later. Also note the full-size render vertical resolution for later. (ie 2048 usually)
The fun part:
-Now check the ‘crop’ button and render one image. Hit ‘N’ in the render window to view the properties palette. It will have the image dimensions. If it’s not as even-numbered, you will have to draw the frame again. Yeah, this SUCKS!!! I zoom the frame as much as possible, then redraw the frame over the previous one overlapping exactly the upper left corner but adjust the right and bottom edges as needed. Render one frame again, check the size again, iterate until happy or you smash your keyboard into smithereens. (what the hell is a ‘smithereen’ anyway. Nigec? Another brit word?)
-For the render output, if you are doing any video post-process, select AVI Codec. This will containerize but not compress your frames. If this is matching your final Blender face textures as-is, then select the Theora preset.
-Set your animation output path directly to the Video sub-directory of your Dagon project.
-Uncheck ‘extensions’ or blender inserts the frame numbers in the filename for some stupid reason. Name your file, and you have to put in the extension, ogv. (unless you are going with AVI, then use that extension.)
-Hit Animate. You are done in Blender.
-In your script, when attaching your video to a spot, use your coordinates you noted earlier. The Y value must be reversed, hence the face resolution:
spot = node:addSpot( Spot(NORTH, {x, 2048-y}, {auto = true, loop = true})
spot:attach(VIDEO, "minimovie.ogg")
…
…
I say, this would be extremely staight-forward if it weren’t for the region hassle. That would be a ridiculously easy python script to write and could scarf up all the other little steps as well into one neat little panel. Hint hint …