806 replies to this topic

[Domdron]

Speaking of alignment, I shot the Moon and M24 last week, and noted that when I set ISO somewhat higher than 1600 (I tried up to about 3000 IIRC), the alignment would stop working. I used a simple tracking mount, but had polar-aligned it only very roughly, so there would be some drift. Might it be because then noise gets more significant, and it tries to align on the noise instead of stars/objects? 

 

Anyway here are the pics, both first edited in AstroShader and then in the photos app. 

 

Moon, iso_1682_0.001s_x1000_01-Jun-2025_19.22.57.786

 

M24 Small Sagittarius Star Cloud, iso_1212_1.0s_x296_02-Jun-2025_02.35.59.059

 

 

[GeezerGazer]

Sorry I've been a bot quiet on here I been away and had a busy few months with life.

 

Mmm this is odd.

 

I haven't changed anything with the alignment in v1.0.29.

 

My guess would be a unique combination of target brightness and movement between frames. The fact that changing the alignment improves this makes me this this is the most likely.

 

There is some upper limit to the amount of movement the alignment function can tolerate above which it completely fails. I've got some big plans for the image processing pipeline that will also involve alignment algorithm changes, so may resolve this issue.

Yes, it was odd.  I'm pretty sure it was not a tracking error from the mount because when I had to close AS, I finished the photo session with NightCap, which has no alignment function, and all images showed that the mount was performing fine.  I also do not think it was from the subject matter being too dim within the preview, because I switched to several very bright objects which did not align in AS.  I do think the problem arose from app corruption in my phone.  

 

I did delete and re-download AS, and it performed very well four nights later, until I closed the app on the phone during the session to check an image in Photos, which froze the preview screen when I re-opened AS on my iPhone 15 ProMax.  To resolve this problem, I closed the app, powered off the phone and then re-booted.  The preview screen was no longer frozen, so I took several more images.  The preview screen did freeze again about 2 hrs later.  This time I opened the AS library... when I closed the library, the preview screen was not frozen, and I took several more images without issue.  

 

This problem was reminiscent of the frozen preview screen problem we had in 2024.  I wondered if this problem was specific to my iPhone 15!  I haven't seen anyone else having either of these problems.  I have had a few preview screen freeze-ups over the past 6 months; all resolved with a re-boot.  

 

June 21-25, I'll be on a 5 day, dark site observing trip, where cell service and wi-fi are not available.  If I have any problems I'll take notes.  

[Ripvanhalen]

This is a great idea!

 

I'm not 100% sure how to implement it but will have a think. There is not set number/threshold of registration points for alignment. Another challenge is that the alignment process is quite heavy so running it during the preview is likely not a viable option.

 

I suppose there are 2 elements to the alignment failures:
1) Regular trailing - which could be alignment failures or overly long exposure times
2) Wild errors, like GeezerGazer's above

I may not be able to tell you with certainty before you take the image whether it will be aligned properly. But I should be able to say after the first 2-3 images whether it will be one of the 2 options above.

What would you think about some early alerting function during the capture process to state the alignment has failed?

Thank you for the consideration. I thought of it as I was using my phone to photograph a photograph and noticed how the phone put up little boxes around the faces. It reminded me of an auto track feature of a gimballed camera I used at work, that could track moving targets--it would put a little "breathing" box around its currently selected object. Pressing the button on this work camera caused the gimbals to move to maintain the object at the same location in the frame. It occurred to me that the iPhone API might have features that could support similar functionality. (But I have no clue).

In any case, the early alerting function (post shutter activation) you mentioned (hat tip: domdron!) seems like the much more easily implemented feature, and would provide a lot of utility.

Again thanks! It's a wonderful app, and I have gotten some great shots of whole constellations with it w/o the scope. It's only with my telescope where I am getting hung up by trailing.

Best regards,

-Chris

 

PS: I think it's just "1) Regular trailing - which could be alignment failures or overly long exposure times", b/c my exposure times were well below a second.

Another thought from the random idea factory: the user can give the app an alignment point suggestion by touching the screen, much as the iPhone camera app takes an autofocus+autoexposure command from the user. So if I were to touch a star on the screen, the phone could respond by displaying a little circle around it. Ideally, it would work in the same fashion on a moon crater, or on a planet, but I know software improves incrementally.

But the core of my concept, I suppose, is that I know what kinds of image features are most easily tracked in live video, and I'm wondering if there isn't some little HMI back and forth with the user that would help the app do live stacking with moving targets.

Edited by Ripvanhalen, 14 June 2025 - 11:40 AM.

[Domdron]

Another thought from the random idea factory: the user can give the app an alignment point suggestion by touching the screen, much as the iPhone camera app takes an autofocus+autoexposure command from the user. So if I were to touch a star on the screen, the phone could respond by displaying a little circle around it. Ideally, it would work in the same fashion on a moon crater, or on a planet, but I know software improves incrementally.

I don't think that the alignment works on a single star/feature, but a function which matches global similarity of the images. Or at least that's what I seem to remember from a previous post from astroshdr here, maybe I'm wrong?

In any case, a single feature alignment would be problematic if that feature suddenly dims e.g. because of a small cloud, a nearby satellite appearing or sth like that.

[Domdron]

I might add that realtime alignment seems to be a difficult problem in general. E.g. I'm using Ekos/KStars polar alignment routine, and in the adjustment phase you select a star to use as a reference while adjusting alt/az, while the camera repeatedly takes pictures and circles the star. But very frequently, it circles the wrong star*. So identifying a single star reliably even in a quite controlled scenario seems to be difficult and unreliable, assuming that Ekos doesn't lack sophistication (which I don't think).

 

* One can still continue with the polar alignment by just visually tracking the star and moving it along the lines into the target crosshairs, but the displayed error and direction are wrong, and that's annoying. 

Edited by Domdron, 17 June 2025 - 02:53 AM.

[GeezerGazer]

I don't think that the alignment works on a single star/feature, but a function which matches global similarity of the images. Or at least that's what I seem to remember from a previous post from astroshdr here, maybe I'm wrong?

In any case, a single feature alignment would be problematic if that feature suddenly dims e.g. because of a small cloud, a nearby satellite appearing or sth like that.

Agreed.  Using "Set reposition interval" with the "Highlight images" function shows how the entire image is used for manual re-alignment.  Single star (automatic) alignment would require a completely different alignment system and would not account for field rotation from alt/az tracking mounts.  

 

Overall, I have not had any repeat alignment problems since I deleted and then re-downloaded the app.  So I suspect my alignment problems were related to app corruption within my phone... that it was not a general app problem, but was specific to my use.  

Ray

[Ripvanhalen]

Agreed.  Using "Set reposition interval" with the "Highlight images" function shows how the entire image is used for manual re-alignment.  Single star (automatic) alignment would require a completely different alignment system and would not account for field rotation from alt/az tracking mounts.  

 

Overall, I have not had any repeat alignment problems since I deleted and then re-downloaded the app.  So I suspect my alignment problems were related to app corruption within my phone... that it was not a general app problem, but was specific to my use.  

Ray

Liked. I think a sufficient way of implementing the @domdron idea of a live status indicator would be two numbers: translation and rotation or other inter-image measurement of the movement needed to align the latest image with the stack (or with the previous image). Maybe I'm all wet here, but...as I imagine two rectangles being overlaid on each other to maximize total correlation, I believe that ultimately at least two reference points in the new image must be shifted some number of pixels, possibly with some angular movement (what I learned in high school as "translation of axes"), so that they line up with corresponding parts of the existing image.

 

For a perfectly aligned EQ mount, both figures should be zero. For a non motorized mount, both figures would be nonzero. For an az/El tracking mount perfectly aligned, the image rotates about the tracking point (I.e. rotation rate is non zero) and its translation is probably implemenation dependent.

In any case, with my non motorized mount, if I see zero and zero, I have trailing and I need to try again.

 

Edit: this might be doable as a live thumbnail graphic representation of the drifting reference points or the drifting corners of the image (unless you have a perfect tracking EQ mount, in which case there is no drift of any kind--all the reference points stay put.)

Edited by Ripvanhalen, 22 June 2025 - 04:59 PM.