Again, you seem to have no idea how IBIS works. The point of moving the sensor is to keep the image centered on the sensor in the way it was positioned at the start of activating IS. Extra pixels do not help for that, they are however needed for other reasons, like dealing with geometric distortion.

Also, you don’t seem to realize that cameras like the e-3 and especially the E-M5 can activate IS before shutter release, and having to reset IBIS at the start of exposure would introduce some significant delay. The sensor movement possible in those cameras has enough latitude to prevent that issue in virtually all casrs.

That resetting IBIS takes significant time is totally obvious from why you can disable IBIS during continuous shooting. Doing so gives a quite noticeable imprlovement of the framerate.

Not to mention, the reset is pretty easy to observe both from sound and from a very slight repercussion.

First you have to get over the fact that there arent the pixels to support this specious claim,

Second, you probably need to figure out how far a sensor moves to offset shake in a shutter duration, this is also true to video which has shutter speed control.

Third, you need to figure out that 0.5 mm or less is actually a long way across an image.

I say again, you dont know what you are talking about, but you do have the opportunity to prove me wrong, so lets hear it.

To take it one step further, video stabilization is even more demanding (as the sensor can’t be re-positioned in the middle “just before taking a photo”), which is why an even smaller portion of the sensor is used (again in the case of OM-D in video mode).

We will examine E5 specs and deduce what we can. The imaging part of the sensor itself records 4032 x 3024 = 12,192,768 pixels and the spec reveals it has 13,100,000 pixels available. It is important to note that while not all of these ‘spare’ pixels are used for IS.

With a remainder of 907,232 pixels at 54,691 pixels a sq mm gives us 16.59 sq mm to distribute across the 60.6 mm perimeter of the sensor, that gives 0.2737 mm wide boundary around the sensor.

Given the IS mechanism can be +/- that brings us up to 0.547 mm, or just over half a mm total.

Realistically all sensors have a margin for proper functionality so some, actually much of that margin is shared for those tasks so its actually considerably less than 0.5mm, and by a ways at that. IOW when you claim several mm you are talking out of your ass.

IS and image circle are unrelated factors