How are you managing the field rotation?
I'm not. I shot the above image with 12 second subframes. Honestly I think 12s was a bit too long too (sharpness in sharpcap was soft), I might try 8s or 4s. I shot with NINA, and ran sharpcap's folder based stacking in the background.
In the center sharpcap reported green for field rotation of around 1.3~ pixels @ 12s frames, but red on the edges. As you can see over my 8 hour shoot the field rotation made a squarey-circle. Sharpcap used the first frame in black, and the light-polluted purple frame is the stacked result of all the remaining area of the image after field rotation.
I'm trying to figure out stacking in Siril right now - however its not fun given the 2,000+ frames it produced at 52 megabytes each (104 gigabytes of data!
)
I also have a rotator so I could have Nina estimate the original degress of the image - was around 227 degrees going off of memory last night. I could then have a loop of shoot, rotate, shoot, rotate, etc. I happen to have a primalucelab rotator on my dobsonian to test out field de-rotation. Sadly even though this rotator advertises field derotation - they don't have it's capability in the existing PLAY software despite advertising all over it does field derotation. I'm a software engineer myself so I can do the calculations based on the mount's current coordinates and talk to the ASCOM driver of the rotator/etc.
Now that centering is working quickly - the quick and dirty in me is trying a shoot - rotate - shoot first in Nina to see if that de-rotates it "enough."
Field rotation is really weird. On a 1 hour shoot it does hell with the diffraction spikes. 8+ hours in - sigma pixel rejection removed all the diffraction spikes and gave me refractor-like stars with no spikes - heck, I'm surprised it even removed my mirror clip diffraction too. I didn't need to use any apodizing masks to remove diffraction spikes, no anti-egret masks, no lost aperture with mirror clip masks!
Ultimately though field rotation has a huge price to pay - it turned my lovely APS-C style sensor into a 4/3 crop sensor. So if I wanted to do wide sweeping mosiacs with this dob - that's getting rough without solving field de-rotation.
Another issue I've found with my rotator is it slightly changed my flats and tilt. So I'm not sure how the calibrated images would look like with it either. Without the rotator with a secure mechanical threaded connection I could mostly re-use flats for an entire week until dust buildup was too much and I had to either A. clean out the dust (which let me keep the same flats - I still took new flats just to be sure, couldn't really tell a difference
), or B. take new flats.
I rotated my camera roughly 2 degrees last night to get the initial diffraction spikes to line up perfectly square with my initial image. That required new flats as that minor change was just bad enough that the images looked really bad on the edges.
So there is huge risk that you might get some garbage calibrated frames with constantly de-rotating on a cheap dob, especially a fast dob where the secondary doesn't even appear as a perfect circle through the focuser (and if its fast enough - it shouldn't appear as a perfect circle when you're collimating it enough.)
So one might want to make a master flats library for a field derotated image, say 40 frames at each 1 degree rotation, so you're stacking 360 degrees * 40 frames per degree = 14,400 frames going into a master flat. Now imagine having to do that every day due to dust. Yuck. I woudln't do anything less than 10 frames/degree, so 3,600 frames as the bare mininum.
Even worse - that master calibrated file might not work well enough on the edges, and only works on the center. Or even worse - you now have to calibrate with the nearest degree flat, or even worse. Now, imagine trying to figure out what degree flat to use as you could have Nina write out the degrees in the file - that'd roughly be the same number given field de-rotation. You'd have to write your alt-az coordinates in the file name from Nina (is that even possible?) to then have software re-rotate that to pick the correct flat! Even then - what if 1 degree wasn't good enough for a flat? 1/2 degree? 1/10 of a degree? 1 arc-second? 
The last issue is I also have random slippage at times with my Starizona nexus coma corrector & reducer with the rotator. I chose the 2" visual back adapter for my rotator as while the coma reducer has m48 threads on both sides - I've read that some people had to order a collarless version of the coma corrector to get it far enough in the draw tube on some dobs! Thankfully I didn't need a collarless version - but I don't think if I got a m48 adapter on the rotator I will be able to focus. I could absolutely be wrong, but my understanding is the coma corrector needs to grab light inside the draw tube, not outside. Please someone correct me if I'm wrong about this. I might just order the m48 adatper to test it out.
Regardless - I don't think it'll change the need to take at least 360 degrees of flats as even if you have a perfectly collimated newtonian (which isn't possible without robotics), I'd think that rotation alone will always require new flats with a newtonian design because of collimation. That's not bad if you're wanting to frame things for a large mosiac on an equitorial newtonian - new flats for a new chosen manual rotation is reasonable.
So that also means field de-rotation on a dob means shooting images without a coma corrector - who would want that? Maybe that's fine for planetary and lunar, but Autostakkert already has experimental field-derotation that seems to work for me.
So yeah, its a fun experiment. So you either capture cheap/fun images with an APS-C sized sensor that gets cropped to 4/3 because of field rotation, or get a wedge for this (not sure how good this would be on a wedge), or you get a proper imaging telescope and a proper equitorial mount.
It rocks for EAA - its really cool to have this fast dob setup. From the comfort of my own home it was showing the horsehead nebula within minutes - despite the light pollution. I can't wait to take it to a bortle 2~ zone! For less than 5 minutes field rotation isn't really noticeable on an EAA style of setup. The image is rotated 1-2 degrees over that time period.
I'd say if you want picture to picture pixel perfect images on your sensor - you'd need to go with a "proper equitorial mount", which I'm still considering doing. The unconventional dob approach was fun, it was a fun puzzle to accomplish this, and it was a great intro to astrophotography for me.