Without seeing your footage, I can offer at least this: when I have inserted an object with the 3D camera tracker in a really demanding scene, I have typically tried three or four sets of points before getting the best match. And that gets me about 90-95 percent of the way there - I usually tweak the positions manually just a bit to get the look more natural. Like many of these fabulous complex tools, the 3D camera tracker might not fully complete your task, but it saves you oodles of time getting you most of the way there.
OK. Great. I'll play around with it some more.
Here's the original footage, just in case you or anyone else may have more to add after viewing it. Once the crane shot reaches it's stopping point, you can see a rock poking out from the right. I was going to place more rock there.
I'd also like to add, the whole behind my wanting to place more rock in this shot is to cover the moving pixels around that rock (someone else removed the cables that were keeping the actors safe but you can see some pixels - from the cables - moving on the very edge of that rock.) I thought putting an object to cover up some ugly pixels would be easier than trying to paint pixels in a *moving crane* shot.
After having seen the shot, I wonder if you would be well served by moving the covering rock independtly, putting it in the foreground, and thereby adding more depth to the shot - maybe puttng a light on the foreground rock to reinforce the movement. Once again the 3D camera tracker might give you a good start but you would have to tweak it manually. Easy for me to suggest this, though, since I don't have to do all the hard work ;-).