Using a Python script, I've tested the effect of different parameters and algorithms on the quality of the ultrasound image. Here is a 100 x 100 test with 50 elements, beam focusing on a 16 x 16 point grid, MVDR beamforming, de-modulated signal, 1 MHz fundamental imaging.

It's not easy to simulate the propagation of ultrasound in tissues, so this picture is a greatly simplified, unrealistic approximation with lots of noise. But a real device with the same parameters should produce something at least as good as it, and I now have reference code for such a device.