I will have to check it out.A guy who researches that quite a bit is saying that the hardness doesn't matter as much at high angles.
I'm not so convinced.
Those guys at BF are using Lorrin Birds formula for Russian APBC shot which he developed by scaling that graph. In fact that is who originally posted that 122mm graph.
The only other high angle data I have is from modern tests (Military Technology magazine) of long rod penetrators during the late 1990's. Blunt rods do indeed have an effective high slope penetration less than the cosine rule. But not that much less, like only 10-15 degrees.
I suppose the whole ricochet modeling vs. armor penetration modeling is what I am getting at. The Germans did not like testing at high obliquity since they could not get reliable results. But that is an effect of ricochet. The smallest factors, such as projectiles not flying in the projectile path (pitch and yaw, etc.) mess with expectations at extreme angles. Shape of projectiles and what hits first, also mess with expectations. The Germans didn't like testing against face hardened armor either. It was not easy to make consistent FH armor like it is with homogeneous armor.
By the way...is that a M79 or M61 round stuck in the armor of the jagdpanzer IV(A)?