Hi every one !
This topic was very active the time I was not there ! Here are some answers about, to my mind, some critical points:
What is often done is to use a pair of shutter blinds (see image below). The first blind (the green one) 'drops' and uncovers the film or sensor; it is closely followed by a second blind (the blue one) which re-obscures the sensor. As the first blind goes 'down' it is uniformly followed by the second one so as the 'gap' travels down the sensor, all parts receive an equal duration of illumination. The gap between the trailing edge of the first shutter and the leading edge of the second one determines the exposure time.
Thanks to Dave who has, for me, explained the same things that are detailed in the link I've post earlier this day, we can all assume that there is no way to have a perfect global shutter with a mechanical system. Indeed, there is always the time for the blind to drop, which cannot be at an infinity speed. Nevertheless, this drop speed is still very high (most of the time faster than the shutter speed) and the second blind is closely following the first one only at very high shutter speed ! (As it's explain in my link).
What we actually called rolling effect is due to slow sensor readout which is more common with electronic shutter. This rolling effect appears when the sensor readout is slower in comparison to the camera/drone moving speed, so that there is a visible movement between the reading of the first pixel line and the last one.
In a word, only electronic global shutter has not any rolling effect, but the rolling effect due to the drop speed of the blind in mechanical systems is almost nonexistent.
-----
The X5S camera has only an electronic shutter, hence you will have to use software based rolling shutter compensation to get rid of it.
That's a thing SAV... I has dreaming on working with such device. Are you sure there is only electronic shutter ? Then it may smash my dreams... I'm currently not confident on rolling shutter adjustment...
-----
Agisoft handles P4 Adv/Pro (and X4s) data very well. When optimizing data using this sensor keep the "Fit rolling shutter, Fit k4, Fit p3, and Fit p4" options unchecked.
Yes it is ! But still you should be active on the choice on the optimized parameters. Indeed sometimes more distortion polynomial factors may fit better the real distortion of your camera, sometimes your model is not strong enough to accurately converge to stable values. Looking at the correlation matrix of the adjustment (available in the processing report) may show you the strength of the adjustment. I personally always go to the maximum number of parameters (but without rolling adjustment as I work with mechanical shutter) and then, if the adjustment is not strong enough and the adjustment move apart (typically 1 or -1 value in the correlation matrix), I reset the smallest parameters to 0 one by one (never further than K3-4 and P3-4) until the adjustment gets more strenght
-----
Congrats!
Hi SAV !
Thanks for the support ! I'm now graduated 
Thank you !
(critical point indeed)
-----
Sorry for spamming the thread but, what do you think about the X4S camera upscaling images ie not a really 20MP camera and how could that affect photogrammetry projects?
That's indeed spaming cadm8, but a good point to highlight !
As SAV suggested after your message, it might be a true 20MP effective pixels, but the sensor himself could have more pixels. I guess we will never have the true truth about that
Nevertheless, we should take it into account, especially about the "frame looking" of the pictures even with a quite small focal length in comparison to the sensor size (The link you've posted show better the "fisheye looking" of the original sensor). We definitely cannot process those pictures with a fisheye distortion model, but, as you can see on attached pictures, the image residual are always bigger with a P4P than with well-known GoPro or even with a full frame sensor. (All the attached pictures are coming from huge surveys so that their strength gave us the possibility to accurately adjust the values of the internal parameters
Happy reading !
Regards
