While messing about with Voodoo and VideoTrace to try and find a way to extract 3D models from a video and reinsert them as overlays to interact with, I discovered I have a rather shy model.
Voodoo actually tracks the scene really well and displays the points perfectly within itself, so I'm not sure if it's its export function that messes up, or Hitfilm's import, but with some videos the results are almost perfect, others: not so much. But the rough tracking worked well enough to test the concept, plus it was nice and short.
Likewise, VideoTrace can make much better models than I made here, that was just the quickest boxy shape I could throw together to approximate the jeep.
So, with that in mind: if anyone can tell me why the model does a bunk when it has to share the scene with a small orange plane, that would be great.
Tagging the usual helpful peeps: @Ady , @SimonKJones, @Triem23 , @Aladdin4d
(Video updated to answer question Simon asked below.)