[interested_observer]

For any nightscape shot that is wide enough to encompass both nortern and southern starfields either side of the celestial equator, there wil be differential star trailing. There is no way that the O-GPS1 or the built-in gps units can counteract for both movements.

Trailing at the extreme edges is unavoidable with the astro tracer due to how wide rectilinear lenses warps the image. Everything is prolonged at the edges in order to keep lines straight and that means that stars appears to move faster at the edges of the frame when using such lenses.

But since the sensor is a solid piece of equipment it can only compensate for one speed of movements, and thus stars will still seem to move at the extreme borders.

And as funktionsfrei said, all cameras that can use the astro tracer can rotate their sensors.

I do agree. I'm located in Arizona, not sufficiently south to really get any southern starfields. Actually the Milky Way core is nearly always under the horizon, so as the northern arm swings across the sky, the extreme reaches of the arm has a greater angular movement. You can easily see this using Stellarium. The lens I was using at 18mm has a field of view of 77 degrees x 44 degrees. Yes, it's wide - but reasonably mild compared to 15mm and 10mm on a full frame sensor that gets along to 100 degrees wide and beyond.

Yes, I agree that the sensor can rotate, however the longer the exposure - and track, the closer the sensor moves to its limits, which is where the rotational component is needed the most, but will be limited out. Thus, the shorter duration tend to limit the need for rotation, while the longer durations - need the most - hence my reference to 2 axis essentially dominating the application of the third. When you need the third (on the longer tracks), I believe that it is essentially limited out by the physical constraints. But who really knows, since Pentax has not really provided any information on their implementation. That is why I am thinking that the new 5 axis sensor package may perform better on the longer duration tracks.

The longer the track the more trailing along the edges is observed. Limiting the track to 60 seconds, essentially solves the problem. 60 second tracks are still 6x more exposure than what one would acquire with a single frame using the rule of 200 (200/focal length) which essentially is designed to contain the stars within a pixel.

Another approach is to use the portrait orientation, and stitch. This approach maximizes the use of the center of the sensor where you have a good star point (and the frame edges are not sufficiently wide to aggravate the trailing problem). You can also shave or crop the extreme sides off. For me, the problem is most pronounced with landscape shots. Most of the landscapes (I am interested in) fit into about an 18mm field of view. Backing up is an option at times, but if you back up too far, you loose your perspective on the landscape element within the frame. It's all a balancing act....

This approach to star tracking suites me just fine. I really do not want to go with a tracking mount - just too much equipment to lug around.