I have been trying it over the past couple of days and am happy with the results. I did the manual measurement calibration instead of attaching an accelerometer to each axis. The top taller tower in each of the ringing tower pictures doesn't have compensation and the smaller tower on the bottom of each picture has resonance compensation applied (mzv shaper). The cubes were printed using the same gcode with a max acceleration of 2000 mm/s2 at a base print speed of 100mm/s and 50mm/s for inner and outer walls and a 0.4mm layer height with a 0.6mm nozzle. According to fluidd it was reaching 43-50mm/s on the walls and 80-100mm/s on the infill and was doing 250mm/s travel moves. Honestly, the difference isn't as noticeable as I thought it would be but I do think the resonance compensation makes the prints look nicer.
Ringing tower X-axis
Ringing tower Y-axis
CR-20 X-axis with direct drive mod (compensation turned on)
Cr-20 Y-axis (compensation turned on)
CR-20 X-axis with direct drive mod (compensation turned off)
Cr-20 Y-axis (compensation turned off)