The live-action scenes were shot at 23.976fps and telecined: they can be inverse telecined. The effects shots were created at 29.97fps and can’t be inverse telecined: they can be decimated, but that leaves you with jerky movement. One can go ahead and deinterlace them, but the result will be a mess because they’re not really interlaced in the first place.
So the SPFX are 30p?
If this is the case, what if we used motion compensation to generate 4 intermediate frames for every group of five and then drop the originals? This would give us the 4:5 ratio we need without jerky movement.
A ab’ B bc’ C cd’ D de’ E | F fg’ G gh’ H hi’ I ij’ J
The only issue left, I think, is figuring out what to do between each group of 4 because the lack of an intermediate frame there will result in its own jerkiness.