Active cooling has been a cheap way to get more power out of underrated motors. You want as much power as the motors can handle. When you can't afford a variety of motors, there's active cooling.
Testing with the narrowest field of view, the slightest motor cogging is revealed.
Got the tiny motors up to .9A with active cooling. The pitch still has .3A without active cooling, to minimize vibration. The tiny motors have the least cogging artifacts, so active cooling might get them all the way. It's never going to be perfect, without bigger motors.
10 bit PWM, 1024 step sin table added a tiny bit of precision. No amount of power or PWM precision could completely eliminate cogging. 8khz, 32khz, shifting the magnets away from the stator didn't matter. If you don't get a bog standard gimbal for $100, you'll end up paying more to find ideal motors.
Slightly better in the running timelapse with software stabilization, which means it could be superb when flying. For a running timelapse, it's probably not worth supporting yaw or pitch at all, if software stabilization is required. A simple roll stabilizer is what you want for running.
Wound up a 3rd Turnigy 2205 1350kv for the yaw. Grinding off the enclosure made it a lot easier & opened up more airflow. It was just about as easy as a DT700. The enclosure served no purpose.
Aluminum cooled it off a bit better.
Given enough current, active cooling, a high resolution DAC, brushless motors have a lot of potential in highly accurate pointing. A 16 bit DAC can get 65535 steps in 90 deg. They don't need an IMU.
There were some test videos using the passively cooled DT750, revealing how much worse the cogging & the roll was without enough power.
Brushless motors would have made a decent laser projector, years ago. Kids were making stepper motors for laser projectors, but they had to machine the bearings. No-one realized a hobby motor or a fan motor could be used for a pointing task.