..I recorded a short segment of live music at a festival with the mics taped to either side of the ball facing forwards and with the caps flush in the ball facing directly to the sides. ..using the high boost grids flush in the surface of the ball which were much, much better sounding...
FWIW the best results with the ball were with the caps completely flush to the surface facing out, not taped on tangentially facing forwards.
I recall reading that Schoeps' KFM8/360 Sphere mic has "the required equilization 'built-in' to the electronics". I now assume this is a pretty healthy dose of high frequency boost, similar to a diffuse field curve, but more so since they probably use their diffuse field omni's in there to start with. To make that test recording (with the short grids) more aceptable when I played it back, I added a rising EQ curve that started lower than the high-boost grid curve does, so the high-boost grid may not be exactly right to optimally to correct the ball baffle effect, but it sounded very good with the ambient (non-musical) campsite stuff.
The longer, flat-ended high boost grid also fit perfectly flush to the surface of the ball when installed which was somehow more satisfying to my design eye than leaving that little bit of screen sticking out with the short grids.
My high boost grids are completely flat on the end with no screen visible form the side, unlike the drawing on the DPA chart above. Yours?