Dear Jean-Luc, I will read and respond more thoroughly later today but I wanted to mention that the situation where AZ-GTI (and others) in AltAz can excel is in EAA (Electronic Assisted Astronomy). With my Nikon D5100 you can capture deep sky pics of short exposure (10-15 secs) at 1000 ISO and live-stack them and have a wonderful experience of almost-real-time seeing of objects you can't see with your eyes. As you mentioned, doing this requires somewhat precise tracking (sufficiently precise to avoid blurry images during the 10-15 secs of the exposure), although movement / rotation between pics is unimportant since the live stacking can correct it.
Jasem mentioned EAA was in the roadmap of Stellarmate. More later, thank you for your response!

And to the why not use the wedge, I have it, but if I bring it I also have to bring the counterweight, which adds about 4kg to the package and also requires a more serious tripod and kind of overweights the AZ-GTI at least theoretically. In AltAz you can use AZ-GTI with minimum weight and becomes perfect for traveling. Additionally, Alt-Az doesn't require polar alignment so it can be snappier to set up.

@geehalel: I can´t stress enough how thankful I am for your help and efforts !
Just posted within the astroberry section of this forum to Kaczorek, the maker of astroberry. Maybe he has some idea, what the problem is.
What makes me curious is, that the compilation of the official branch of libindi did work, apparently. At least I got a working libindi afterwards. Did I understand you correctly, that it is your assumption that the compilation of libindi does not work on the astroberry ? I have the feeling that I missed something, anybody more knowledgeable with linux would automatically do.

"I don't know when does field rotation appear in altaz mode, but plate solving may also fail in altaz mode for that reason, even with a perfect smooth tracking. Or am I wrong ?"

Field rotation would not disturb pictures with an exposure so short as I was describing above (10-15 seconds)

I've played a little yesterday evening with the skywatcherAltAzMount driver. I've used the AZEQ5, 150/750 scope with a DMK41 and a 50/206 guider with a ASI120mm. Nothing aligned. Capture/solve/slew to target works great with the ASI120@206 in 4/5 iterations, but strangely with DMK41@750 it solves but does not slew at all. I suspect a problem with V4l2 devices as the laptop also has a webcam. Sync is not working, there were 3 sync points that I tried to delete, but then tracking became weird. Looking at the code, Sync is only allowed when the mount is parked. I made some images of Dumbbell @10, 30, 60 and 120 seconds to observe star drift/rotation. A silly question, I wonder if the field rotation axis is always centered with the sensor center, or is actually the scope axis. By the way I also noticed regular jumps as depicted by @Federico. I suspect the Aux encoders which are not disabled on AZEQ, and/or the tracking algorithm which updates its data every minute. Lastly I tested guiding, when calibration succeeds (I suppose RA/Az axis does not move enough at ~50° altitude using standard guiding rates), those jumps disturbed the algorithm too much (added to field rotation as the guide star was selected on an edge of the image). Guiding in AltAz mode is quite silly, however the guide tool shows the drifting graph (see the jump @45s) and almost every points are in the upper right corner (field rotation presumably, I did not make the calculations). I may use that to test Alt/Az tracking, selecting a centered star as guide star.
Thus I've learned a little more, and your posts also help me to understand what were your concerns and what you would need. As I understand it, alignment is not primordial as long as Capture/Solve/Slew is working, guiding is of no use as your setup does not have a guider, and you only really need a smooth tracking. What I have in mind as a first step is to create a new driver out of the existing indi repos, hence astroberry users may compile it without disturbing their installation. I will derive it from indi-eqmod and/or indi-skywatcherAltAzMount and then concentrate on the tracking feature. I'll keep you informed when that will be initiated and when I will need you to perform tests.

@geehalel thank you SO MUCH for your work. Indeed, having a driver available to track correctly to be able to plate solve as a quick sync technique would be amazing. Let me know whenever there is something to test. All my very best and again, thank you. Federico

@geehalel: you are amazing ! Many thanks, I remain excited to test !

@geehalel: regarding your question of the field rotation axis: I guess as long as the axis of the mount and the guide scope are parallel, the center of the rotation axis should be the center of the sensor, i.e. the distance between mount axis and guide scope axis is far too small to cause any parallax as the points projected on the sensor are at infinity. Things should become more complicated, once both axes are not (perfectly) aligned: in the extreme case of the guide scope axis being perpendicular to the mount axis, there is no more rotation, only translation. In between there should be a mix between rotation and translation, i.e. the center of rotation is moved away from the center of the sensor. My understanding is, that the deviation introduced by such misalignment is what is coined "cone error". Maybe you find some hints in the logic regarding cone error correction to model this; there should also be some literature where this is being described / worked out.
Does this makes sense ?

on second thought: as long as both axes are not perfectly aligned, the center of rotation moves rapidly away from the center of the sensor: let´s assume we have a 1.5 degree fov with our guide scope. If the alignment error between mount and guide scope axis is more than 0.75 degrees, the center of rotation is already off the sensor. I think you can calculate the center of rotation by calculating the fov of your guide scope / sensor combination and substract the deviation angle of the scope axis and mount axis in degrees. How many pixels this equates to, you can derive from the fov calculation. The direction of the deviation should be the opposite of the axis deviation in relation to the sensor orientation.
This is without guarantee. Just my understanding.

on a third thought: I probably should have rather asked you, what you are aiming at. With my hints above, I was in "equatorial mode". In Alt-Az there is no rotational axis of the rig itself. If you want to derotate the exposures made on an alt-az mount, I think the way to go is by star pattern alignment, i.e. agnostic of any mechanical rotation calculation.

Dear @geehalel, @knro et all,
Here's a picture I did yesterday using Stellarmate with my AZ-GTI working in ALT-AZ, Nikon D5100, and SkyWatcher 127 MAK, just one exposure of 30 seconds:

I used Synscan driver - just launched Synscan Pro app, did a one-star alignment with a star nearby the Orion nebulae, then goto Orion nebulae and alignment using Stellarmate astrometry.
Look how well defined the stars are, given that this is a somewhat long exposure and that it was in ALT-AZ, and that I only have done a one-star alignment.
I think if I repeat and align over 3 or 4 nearby stars before goto Orion, I could probably get even more dot-like stars and definition at even longer exposure times.
Of course, this opens the door wide-open to shorter exposure astrophotography, I did 10-second exposures, and then the stars were perfect dots.
This speaks volumes about how awesome the AZ-GTI is (especially given its price). This quality might not be enough for professional astrophotography (I think it might, if perfected a little) but it definitely is enough for EAA!
If only the AZ-GTI driver could do the same the Synscan Pro app do, then we would not need that app as an intermediary and we could start even faster: just turn on AZ-GTI, do alignment-sync using astrometry, boom, ready to start working.
Let me know your thoughts,
All best,
Federico

Only wanted to reiterate my continued interest in this topic...
@geehalel: I guess I owe you an apology regarding my newbee comments concerning the field rotation: I was actually not aware that you are Jean-Luc, the creator of the EQmod driver. If I would have known, I would have spared you these trvialities...
Anyway, I hope you are making progress regarding the Alt-Az driver ?
Cheers
Dirk