Support for rigid camera rig as input?

Hi DA team

I have a use case where I have a 360 camera, and I transform the spherical images into cubemap faces (i.e. multiple “views” with known relative geometry).
In that setting, I know the extrinsic constraints / fixed transform relationships between those cubemap faces

Is there currently a way to encode such rig constraints into DA’s input (beyond just providing independent poses per view)?

If not, would you consider this a reasonable extension or “conditioning” mechanism?
Thanks