+1 for the topic!
My main c4d > vvvv use case so far is definitely the first one you mentioned @max_onion, geometry + texture + animation c4d export / vvvv import.
Mainly regarding real-time preview of installations in architectural context.
To sell an idea one usually needs a decent preview of the piece, be it an arrangement of LED walls, some mix of light sources and screens, or even kinetic installations.
For that reason I usually model the architecture in c4d, add some GI lighting, bake everything, then recreate the scene in vvvv, ie. using vvvv along the lines of D3
Then I start working on the real-time, reactive or generative part in vvvv.
Also, down the line, I really like it for developing my patches because I then always have a realtime preview of how the output will feel in space. Even more so if you look at it in VR.
And simulating spatial interactions is more realistic. Parameters like "how far does someone need to walk to cause a certain effect" can be anticipated to some extent.
Now, the less fun part: the current workflow.
So far, I had imported many separate collada files in combination with textures, which sucks with big scenes and nested transforms. N-gons cause trouble, untextured or non-uv-mapped-geometry as well, then big geometry files on vvvv 32bit, and so on.
A neat vvvv import node would be indeed very sweet.
Would love to help!