It would be amazing if there was a shift function that allowed to set stereo spread for all voices. The idea being that any newly triggered oscillator would randomly get a pan assigned which places it in the stereo field. Spread would control the width. If for performance reasons the Reverb could not be calculated individually for each channel it could use the summed signal. I played with that on the Hydrasynth and it makes the machine sound massive as it feels as if you're being hit by different sound sources around you.