These are some results from a simple simulation code of microlensing events.

What is shown is the image of a distant star field, projected onto the image plane that contains the solar system, with the Earth orbiting the Sun, its orbit tilted at 30 degrees with respect to the image plane in this example.

An event is when the observer is within the Einstein radius of the lens with respect to a source, projected into the image (observer) plane. We consider heliocentric and terrestrial observers. The screen flashes briefly when an event occurs.

The first simulation is a 0.5 solar mass lens at 4 kpc, with a distant, very dense starfield at 8 kpc with a surface density of a million stars per square parsec (unrealistic density, just to be able to compress the simulation time to a reasonable 5 years). The proper motion is dominated by the 100 km/s proper velocity of the distant starfield. Over the five-year duration of the simulation only one event is seen. As the projected Einstein ring is rather large, the same event is seen, essentially simultaneously, by a heliocentric and a terrestrial observer.

The second simulation is that of a nearby lens, moving at 20 km/s proper velocity, with a distant starfield, again at 8 kpc, with 100,000 stars per square parsec. As the projected Einstein ring area is small, events are localized. The terrestrial observer sees 8 events over the two-year simulated time period; the heliocentric observer, five. They are not the same events.