I don't know much about Terragen, trying to understand how it works in terms of overall workflow .
what I mean is , I can't understand how jobs are separated between 3d packages and Terragen
It has camera motion import ability , thats just a piece of a whole work to have final image
AFAIK Terragen is not only OS background Image creator )
I think good example will contribute better thank 1000 words:

Lets say we need to create animation of a small town (tell a story like rango shrak or whatever...)
lets split final image into 2 pieces:
1) model of a town (ready for rendering, with trees, grass, animated characters, simply saying: everything.)
2) and Environment (ground terrain outside town + trees + grass + clouds , symply saying: all Terragen Image
In that case terragen (Terra mesh) connects that town (ground mesh) where red boxes is on the image . Right ?
in case if there is gaps like in my example , if there is no gaps and all are surrounded with houses its easy because you will never see the place of connection
(lets take Maya as an application) there is how I see that process:

you're creating whole town (green color) in maya completely (trees in maya grass in maya all in maya)
IBL light+sunlight are also created in maya , in other words as I said you just need to render the image or animation
NOW , you 'are exporting degraded (lowpoly combined) mesh into Terragen, in order to have reference mesh that helps you to see the boundaries of the town )
now you need to create the environment in terragen , and if you have sunlight in maya , you need to make same sunlight in TG to correspond the overall light direction
when environment is ready you need to render all cameras with imported maya cameras so to make it same movement as original town animation , to postprocess them in post
of course with disabled all imported objects .
Are described workflow correct ?
Best Regards !