Yes, the z-axis steppers were moving, but weren't keeping the nozzle at a constant distance from the bed. I've always been rather impressed with the performance of the manual mesh bed leveling, so I expected this to work just as well. Maybe it's just a bug? When you get a chance, could you confirm yours is compensating correctly? I've been trying to find a good explanation of how each method differs. I thought UBL, ABL_bilinear, and MBL all could correct for more than just a tilted bed... and I think that's still the case. But UBL and MBL are mesh based, where ABL_bilinear is grid based?? In my start script, I just followed G28 with a G29. That way I know it's active during my print. I ran G29 multiple times to make sure the results were consistent, and they were. I also used M48 on the stock IR sensor and it was very consistent on my 800grit sanded PEI bed.