well, I haven’t tried it. It was theoretical, based on the way things work.
Thinking about it again after your very accurate comment about the render having to stay fixed, is to separate the 3D elements (nulls - solids e.t.c) into a separate comp and use that comp offsetted by your inverse camera shift.
One problem is that the camera shift is calculated in percent of the screen size so maybe create an expresso that converts the percentage to a pixel value, multiply that by -1 (to negate it) and assign that value to a null and export that data to AE and use that to drive the counter motion of your 3D comp.
Don’t forget only to offset things in 2D space and not 3D space. Moving the camera, will change your parallax and you don’t want that. Remember that moving a comp on your screen in 2D space reflects the same thing as the camera offset does in C4D.
That should bring everything in place.
Again, this is all theoretical, so if you make any progress, I’d love to know if it actually works.
I may give it a spin later and try it myself.