Mosaic function errors again: Page 2

Mosaic function errors again

Tech@ZWO

fedorauser
I checked the updates of your log, the imasge that been discard are because the stars are too less on the images. Do you notice if there is any obstruction, such as clouds during enhancement?

Daemon42

Tech@ZWO For the session described above (which was not the last of the night) it was clear skies, no clouds, trees or other obstructions during the entire time. It ended at 12:56am local MST, but last sub file for that session was at 12:09am.

I ran another session after that and it did end when the target moved below the roofline of a building, so that one can be ignored. Last sub file at 2:36am MST, session ended at 2:42am.

This same issue has played out time and time again (3 nights in a row).. At 1.4-1.5x framing, it gets about 22 minutes of exposure over an hour or so of real time, when it's coming around to image that upper left corner of the mosaic and starts failing to stack there. What I notice each time is that corner shows visually high sensor noise, and if I stop it there the live preview shows that there's not a lot of stars there. Since an hour has passed there's about 15 degrees of sky rotation, my hypothesis is that it is having a hard time aligning the few stars against the noisy corner from earlier pass. Because it appears to spiral outward to build the mosaic, there's more and more sky rotation every time it gets back to that upper left corner. For the earlier session last night, it eventually figure out how to stack that corner after about a half hour of failing, and that's where I got another 7-8 minutes of exposures, and then failed again yet again in the upper left corner on the outermost pass of the spiral, as described above.

If I recall the entire time it was failing it claimed there were star trails.

Daemon42

Finally got one to run for 75 minutes of exposure time..
Careful to refocus.. Started a little later, so maybe less noise in the first pass, so it didn't fail when it came back around to that upper left corner..
Here it is with a few edits in post..

jv

I had a similar problem on 3 Dec 2024 while imaging NGC 7000. The nebula was about 60 degrees above the western horizon and the sky was clear. The S50 was focused, levelled to a reading if 0.1, and the compass was calibrated. I was using a frame 2X default size and rotated. The S50 imaged for 18 minutes, 40 seconds, then started rejecting each subframe because of too few stars. I let it run for a few hours to see if the issue would resolve, but it never did. Interestingly, the S50 continued to track the nebula across the sky.

I have filed a bug report and a suggestion that an error counter be added to the software so that, after some consecutive number of these failures, the S50 just moves on instead of getting stuck in one part of the sky.

support

jv Since mosaic mode overlay is based on the previous sheet as a reference to overlay stitching in the neighborhood, if this subframe position is not captured before moving to the next position may offset the target you want to capture and not get the imaging you want. We also this feedback to address this issue. Thank you for your suggestion.

Daemon42

support If skipping ahead creates too little overlap, then try backing up to make the previous area clearer, or temporarily move inward a layer in the spiral from the point it failed. Almost every time it has failed for me it's up in that upper left corner near the end of one complete shell of the spiral. Attached are two images showing again what I think is happening.. This is where it got stuck (3 nights in a row).. having imaged almost the entire frame, but then failing in the upper left corner..

And below is same spot zoomed in and the sky atlas showing where it was imaging. Notice how incredibly noisy it is in the corner from an earlier pass. Obviously the noise is reduced by repeated stacked exposures, but I think the issue is that since nearly and hour of real time has passed the sky has rotated a fair bit, there's not a lot of stars there (due to the nature of the target) and it's trying to image halfway outside of the saved frame. This combines to make it struggle to align the latest exposure against the noisy corner. But if it moved inward (in this case, straight down) it could clean up the noise with more of the frame to align on before continuing the spiral out pattern.

BTW, why must every post be approved in this forum now? It makes discussions here really drag on for days and days.. since we have to wait another day for someone to approve every message.

starnet59

What about adding the option of saving RAW 10-second FIT files in mosaic mode (or 20 or 30 seconds to save disk space) like the one that already exists for single frame exposures? For those of us interested in doing our own mosaic stacking this very minor change would be a huge improvement. The list of benefits would include removing satellite trails, multi-night mosaic sessions, mosaics larger than the 2x current maximum, not losing multiple hours of exposure because of clouds, or the target running into buildings or trees.

I mean very simple changes could make our lives using the S50 so much easier and could be implemented in days, not months.

starnet59

Daemon42 Approving every post is heavy handed, counter-productive and self-defeating. I only check this site once a week now and dialogues are virtually impossible.

yes, spam is a problem and must be managed, but the solution is not approving every message but approving users. People that have posted many times and never spammed do not have to have each new post approved. Monitor only new posters and if someone spams, give a warning and then ban.

bkushner

I have this issue and have reported it several times. The failed stacking happens on a perfect sky no reason to fail it just does.

support

bkushner Seestar stacking failures are related to a variety of factors such as wind, ground vibration and other environmental factors.
If you are shooting close to the zenith, the rotation will be more severe, which can also lead to stacking failures. Too few star points can also cause stacking failure.
It is also recommended that you use the default exposure time of 10 seconds; longer exposures can easily cause stars to be stretched out.

Daemon42

support No wind, no ground vibrations, no clouds, no obstructions, and horsehead nebula (the one I was struggling with) is nowhere near the zenith. Besides, with an alt/az scope, the stars rotate the same number of degrees in frame over same span of time whether they're near the zenith or ecliptic. 15 degrees per hour. Fewer star points could be an issue, especially when it's imaging half outside the frame, and there's not much overlap in the corner, but that means the algo needs to change so that it adjusts (moves inward, as I suggested earlier) to create more overlap. It should not just keep trying the same spot and failing forever. A stop and resume feature would also help because could recalibrate horizon, etc, and try again later or next day when there's similar alignment to when it started.

LA3QMA

starnet59 Or give a few more people in different time zones the permissions needed to i.e block/remove messages. Not sure how many users of this forum that actually use the "Flag" function but im using it relatively often and i hope that the admins of the forum gets these and act based on if its spam etc.

Support@Seestar

Daemon42
Mosaic mode for superimposition requires that you shoot with reference to the last image you shot, and not shooting the end of the last one will cause the shot to deviate from the desired target. Thank you for your understanding.

Daemon42

Support@Seestar sigh I'm well aware of how it works, and clearly you are not understanding my suggestion how how to fix the problem. I'm not suggesting that it just skip past the area where it gets stuck to some place that there is no overlap. What I suggested (more than once now) is that it move inward a little, to create more overlap when it gets stuck.
Because the mosaic mode appears to spiral from the center slowing working its way outward, as it gets to the outer edges of the frame, and particularly (in my experience) the upper left corner several factors contribute to it failing to stack.

The corners have more visible sensor noise (I've posted several examples. Please look at them).
We're at the outside edge of the frame, so it's taken quite a while to get around the last circuit of the spiral which means more real time has passed (at least 2x the exposure time), and the sky has rotated quite a bit
Because it's often centering its exposure part way outside the saved frame, the only overlap possible is over the half that is inside the frame, which is the noisy rotated bit from 1 and 2, and this hampers alignment.

If instead of it just getting stuck in one spot forever it simply shifted inward toward the center of the frame from wherever it is stuck, then it will create more overlap inside the frame. If it can stack those exposures it will reduce the noise in that corner, then it can proceed forward, around the spiral, and back out to to recover the area where it was stuck. Please read what I'm actually saying, before dismissing it out of hand.

Support@Seestar

Daemon42
I'm sorry for the unpleasantness. Firstly I understand what you meant, I didn't mean to reject you, maybe my reply was a little bit simplistic, which may have led to the meaning not being quite right. I know you want to solve the problem of mosaic edge noise or too few star points and other factors that cause the resolution to fail, so that the mosaic stacking fails to stagnate and can not continue to stack the shot. Thank you very much for your suggestion, I have given feedback to the product department and R&D department, but it will take a while to optimize, thanks for your advice and patience.

bartholomule

Support@Seestar I've seen this happen too, where it goes about 1.3 hours and then makes no more progress. The amount of time before getting stuck depends on the mosaic parameters.

If something goes wrong, it should probably recenter (with a plate solve), and then work its way back out towards the part it needs, with an increased overlap size until a path of aligned stars is matched for multiple exposures -- then it could just resume as normal. Fixing the error handling would be a good first step towards multi-session mosaics (resumed over multiple nights) as well.

Support@Seestar

bartholomule
We will continue to optimize our algorithms to make these issues less frequent, thank you for your feedback!

« Previous Page