I have a project where I need to superimpose traffic lights at a number of intersections where none are currently presents.
I need to maintain perspective of the superimposed images whilst I pan the intersections.
Can anyone steer me in the right direction
print a still of your video intersection. Take it into a street with a traffic light and set your camera up so the shot is similar. Then photoshop your street lights into you old video by cutting it form one and matting it, then pasting it into the other. I do these things all the time and it takes a bit of an eye and time to get it all right. Just had to take something out of a commercial that we decided we didn't want it in the background after we first decided we did. It took a bit of time to do but the result is that no one knew that there was something there before. Here is a before and after:
If you're purely panning and the lens is nodal or close to nodal then there won't be perspective change. (i.e. 2D solutions can work) That's why you can do pan/tilt on foreground miniatures if your camera is set to nodal. It's also how the better 360 cameras work.
To test set up 2 c stands slightly offset. If you're on nodal the spacing won't change.
Nodal can be center of lens or can be offset forward or back depending on the specific lens.