Firstly.. Yes, I fully agree with the suggestion to add a resume option to mosaic/framing mode.
It would require a way to save in the fit file (or another file) the size, orientation and absolute position of your selected mosaic frame, and where it left off imaging within the frame. This last bit is because when you're doing larger frames they often cannot complete in the available time, let alone make multiple passes. (thus the incomplete corners) Any option to resume should start where it left off and finish working outward along the same plan not back at the center of the frame, as you may then just not finish the frame in the same spot night after night. This may create some challenges if it starts and finishes in a different sky orientation, but this also presents an opportunity if it fails to stack in a corner at the end of one session, you may be able to fix it the next night starting back at the original rotation angle.
As for making it faster.
I doubt that the capture time % in mosaic mode will ever get significantly better and here's why.
- It has to track while exposing.
- It needs time to move between exposures
- It has to move in ways that minimize gear lash, because every time it changes actual drive direction and then subsequently tries to track through gear lash, the scope itself doesn't move at all for a few seconds, and any exposure taken during that time causes star trails.
- It appears to shoot on a fixed schedule
Let's say you're setup to shoot 10s exposures. You'll see at the bottom of the screen it continuously counting down from 10 to zero with no pauses or variations. That means those cannot all be exposures, because it must move some of the time.
With that schedule, if you always alternated between exposing and moving to a new spot you'd get a 50% capture rate, because it has to track while exposing for 10 seconds, then move to its next position hopefully inside the next 10s window, but it won't start the next exposure until the timer resets, so at best..
10s exposure window, 10s move window, 10s exposure window, 10s move dinwo... etc.. 50%.
But sometimes it must execute a move that reverses direction, and that will cause gear lash, and even if it starts tracking immediately (before the 10s timer expires) it usually takes enough time to recover from the lash, that it'll ruin the next exposure. They don't show an error for this, but you'll see that it just doesn't add to the total exposure time for another 10-20s.
I think they have a pretty good guestimate of how long it takes to recover from lash and they just quietly drop the next frame and shoot again. Also, because we know that when it can't stack, it also stops moving this means it has to successfully stack before it can move, which can add more delay.
One way they appear to try to speed it up a bit is to do two exposures for every position they move to.. so it's really
track and expose 10s continue to track and expose 10s.. move 10s.. track and expose 10s track and expose 10s.. and so forth..
But even if that went perfectly giving you a 66.6% capture rate it means it takes that much longer to cover the whole frame because it's trading move time for tracking/expose time. Add in the larger/longer moves and recovery from gear lash, and we end up back at or below the 50% capture rate.
Also, some folks have a mental image that if you zoom 2x then you're just creating a 2x2 grid of exposures. If true you could get 1 full pass (with perfect alignment to the frame) in 40 seconds of exposure time.
Obviously it does not do that as the edges of each exposure are pretty bad (didn't really recognize how bad until watching the process on mosaics), so instead they slice up the frame and create a ton of overlap so that the center of the exposure passes over all parts of the frame, which also means that it has to move nearly half the exposure box outside the frame to get clean coverage on the edges and corners (can see it doing this when you switch to the atlas while it's working). I have a hypothesis about how fast they move (especially when moving left), but the long and short of it is.. when you slice a large 2x frame up into thin slices, and then do a shoot shoot move shoot shoot move sequence in a spiral pattern working from the center out to and beyond the edges.. it is really slow. If they attempt to use less overlap the quality will suffer, the noisy edges will create more visible streaks, and the stack alignment issues we see now in the corners gets worse, potentially causing it to fail more often (as it does sometimes now).
I've got 3 relatively long exposure mosaics, and I'm finding the sweet spot is between 1.3-1.5x frame, and I rotate the frame to get optimal coverage of the target. I'm shooting for 75-90 minutes of exposure time, which costs 3-4 hours of real time.
If it isn't dropping too many frames that is good for about 3 full passes over the whole mosaic, which is what's needed to clean up those noisy corners which will otherwise litter large areas of the mosaic. In my opinion, you'll get better results with smaller frame, and multiple passes, than any really large frame.. until and unless they implement a resume feature, where we can come back the next night and the next, and make it better and better.
