I played with this a lot for a few different reasons. So there are a bunch of papers out there about this, all kind of conflicting. Another one says something along the lines of holding torque depends on where you are on the sine wave, if your steps happen to hit a full step position you still get all the holding torque. Same with 1/2, 1/4, ect.. And then you get in to holding torque, isn’t as important as power while in motion. I gave up and just did some crude tests myself.
I started with a large circle and plotted it with the cnc, I changed the step rate and at 1/8 stepping and a 20″ circle you could clearly see the stepping, so that was out. 16th stepping had no visible jaggies. I also tried to measure the power while in motion with a scale at 1/8 vs 1/32 and saw no noticeable difference. So I went with 32nd. I am seriously considering a force gauge to test the new parts and I would love to test this as well.
I do want to switch back to 1/16 but for speed reasons. I don’t like maxing out the arduino processor for 32nd stepping so soon I will switch to 16th just to have some overhead, if that works that way.