Originally posted by RobA_Oz I understand what you're saying there. The thing we don't have is Ricoh's basis for the 5 stop improvement, either what they classify as "acceptably sharp" or how the interaction between combined motions is interpreted as stops of improvement. The latter is possibly important to understand, because we don't know how much of the gain in SR is attributed to the change to 5-axis motion compensation, and how much is refinement of the linear motion control.
Yes; I that document section 6.8 discusses what is they call the determination level for image stabilization. Also you will find an article by Falk Limo here:
Falk Lumo: Pentax shake reduction revisited where he discusses blur to be beyond 4 px or greater than 20 um on a K7. So clearly from either article the answer to the question of how much the sensor moves in an IBIS system; it really isn't much and is evidently well within any design constraints on the K mount or the FE mount. I would therefore offer that there is a high probability that Sator's assertions are in error and the gentleman from Fuji may be misquoted, misunderstood or simply stating the Fuji engineers made an error in analysis.
I wouldn't have any worries as to whether the K1 will have a working IBIS system. It will. Whether it provides 3 or 4 or 5 stops of compensation and across what range will be interesting to see. And from Falk Lumo's article there are some interesting limitations on the SR system, at least as implemented in the K7. I would hypothesize that while the amount of correction the current system in either the K3 or K1 can perform there is still likely a similar performance envelope that beyond which SR won't make a difference. Something likely similar for example to the 1/125 second figure whereby SR is not effective although perhaps within that envelope the newer units as in the K3 and K1 are more effective that what Falk found in the K7. I found Falk Lumo's article is quite interesting.