First thought that pops to mind is soft-media (baking soda etc) blasting. Set up peel/tests and develop some result for relative soda blasting. Monitor supply air pressure, maybe humidity, distance from sample, time to strip to different degrees of transparency until the paint is gone. Maybe quantify transparency using an electronic-based coating thickness meter.
Dave
Generally, I will not give you the answer to your question, but I *will* guide you into discovering how to solve this yourself.