People with elephant memories may remember me posting a message titled "Canon is not going the distance" in this forum 8 years ago. Complaining that Canon did not report the focus distance in the exif data of the Canon 20D and later models (I don't know if this is still the case).
Having switched to the Olympus EM-5 last year and preparing for a reef-mapping trip to the Philippines I decided to look into this again. Guess what, the EM-5, or exiftool's interpretation of it, reports the estimated focus distance, the estimated image diagonal in the focal plane, and the step count from the actual stepper motor that drives the focus-elements in the lens. The estimated focus distance is a good ballpark for intermediate distances, but not for very close or distant focus. Better than nothing but not ideal. The estimated FOV diagonal is interesing and useful. But the best is the stepper motor count. The image below plots the measured (by me) focus diagonal versus 10000/step_count.
I think this is amazingly accurate and easy to interpolate. With a better model of the relationship it may even turn into a straight line. For long distances it is going to remain difficult because the number of steps per distance is getting increasingly small.
As was pointed out 8 years ago, a flat port would add an additional magnification but this just means I have to recalibrate once underwater or figure out the constant magnification factor due to refraction at the water/port interface.
I also tried it with the 12-50mm lens. But both zooming and focusing change the stepper motor counts so it gets a bit trickier. I'll try tomorrow to make calibration curves for different zoom settings and see if I can come up with a two-parameter function or interpolation. The F3.5-6.3 aperture of the kit lens will likely also result in less accuracy then a lens with wider aperture. However, so far I am quite impressed with Olympus. Eat that Canon