Rolf,
with the DSI II, I have 8.3u x 8.6u pixels.
With the F.L. 1524mm (f5), I calculate that to be an 11.3 x 14.1 arc min field.
Nyquist formula tells me that is 1.16 arc sec/ pixel H and 1.12 arc sec/ pixel V.
For all intents I average that to about 1.14 arc sec / pixel.
Plate solves online in real time confirm a similar field and resolution.
That's why upsampling works so well. The seeing might be something like
2.00 to 2.50 on any given night so I can safely upsample my raw data
by a factor of anything from 1.5x to even 2.00x (
or 4.0x on rare , exceptional nights) and see a benefit after processing the data.
In the homemade worm thread I calculated the periodic error of the
entire circumference of the worm.
It has cyclic error wrapped all the way around the worm, but the worst
is about +/- 6 arc sec of PE.
See here and
here.
As for exposures, I could go longer, 10 sec sets certainly make it go deeper. 15sec and 30 sec sets really make it pop out.
But there is always a trade off.
In an hour I can have 350 frames of 5 sec each.
On a good night I can nearly stack all of them.
On a bad night maybe I can stack only 150-200 that truly have
round stars and no major effetcs of seeing bloating the stars within that
5 sec exposure.
That's the trade off of going so narrow field. Wide field is a bit more forgiving.
You are correct, it is exactly planetary imaging principles at work.
.
...you da close up Cen A groupy 
It is a great galaxy to image...just ask Rolf...I used to be the chief Cen A -a- phile but Rolf took the title from me in a cheeky looooong exposure passing manoeuvre 
