2022-05-11: APP 2.0.0-beta 1 has been released !!!
Download links per platform:
Siril for Stacking, APP for Image Processing
Yes, its me again, having problems with Analyze Stars.
Last night I was blessed with a rare occurrence: excellent seeing conditions. I captured 2.4 hours of M31 in LRGB using bin2 for RGB. When I went to process it in APP more than half of the color frames were rejected by Analyze Stars. As you may recall my camera is slightly under-sampling at bin1 and of course is worse in bin2 but under normal seeing conditions I have no problems. My experience with Analyze Stars in the past is that it is quite conservative as to what qualifies as a star. The log file showed that 100 stars were detected for a given frame but upon further analysis it determined that there were only 4 stars that qualified!
A friend advised me to try Siril. I took a crash course in it today and was rewarded for my effort. Please see the attached photo. (There are some artifacts on the brighter stars that you might see. That was due to stack alignment in Siril. I'll go back and redo it.)
It's been a while since I watched APP video tutorials but I recall seeing more controls on the Analyze Stars tab. Perhaps you gave us greater control in the past but decided to yank it out?
Thanks. I am hopeful that you can give us more flexibility in the future.
There may have been more controls, but some have been taken out because APP is able to do what those options did, automatically in future versions. It's unlikely these would have helped in this case, did you try switching the automatic target off?
Vincent, I tried that but it just delayed failure to Registration, even using Triangles. Thankfully now I have an extra arrow in my quiver named Siril.
If you want it would be nice to have some of the subs so we can have a look at the reason why it fails? It might help to improve the star detection process. Mabula did improve on this very recently so maybe this is in the next version.
You are using the beta?
Vincent, I am running 1.074.1. Perhaps I should download the beta first to give it a try. If need be, what is the best way to send you image files? Do you need just a single light frame? Do you need the calibration masters? It is a small sensor so the files are not that large.
Yes, it might help or just wait for the next release, the beta has a few problems of its own (hence the beta indicator). Not sure if you can keep the old version separately actually, I'm going to try that.
If you want to upload just a few subs and some of the calibration subs you use, that would be great. You can log in to our APP server (it has no certificate so your browser might complain, but it is safe) and login with username/password: appuser
Just create a folder with your name on it and upload it there. Thanks!
Vincent, I've uploaded the following:
11 lights (5 fail analyze stars)
master bias, dark, flat
3 bias frames
3 dark frames
3 flat frames
Thanks very much.
Thank you! I hope Mabula has time to have a look at it, he's very busy now implementing the last things for the next release.
edit: he's going to have a look
Thank you @oopfan for sharing your data, I am investigating now and it's quite clear that the star detection/analysis fails because of your undersampled data. I will try to have this fixed in the next release.
With regard to the settings in 3) Analyse Stars, by further improving the module we have got rid of several settings making things a lot easier for almost all APP users. I think your undersampling case can be included as well without introducing new settings in 3). As you can understand, the key for us is to see all the data that is fed to APP to make it work on all the data. Your data is special, since it is really undersampled a lot... some stars are like a hot pixel as you know.
I downloaded 1.075 this morning and processed my difficult M31 dataset. For those of you joining the conversation my camera's pixel size relative to my scope's focal length ranks it as "slightly under-sampling" at bin1, so things are worse using bin2. Nevertheless I image using bin2 for RGB. Up until recently I've not had a problem. With this dataset however I received failures in Analyze Stars that pushed me to seek alternate software. Siril was able to process it but I wasn't very happy with the alignment, then again I am not expert at Siril.
This morning I downloaded 1.075, deleted all previous masters, basically starting from scratch. Everything ran perfectly. The star count was pleasingly high for the bin2 images, and I was even able to select Quadrilaterals instead of Triangles for Registration but after using the RGB Combine tool I knew something was not right. I zoomed in on the stars. They were square which said to me that I was looking at a bin2 resolution. Sure enough as I moved the cursor over the image the header was very nearly the bin2 size of 696x520.
I found a way to force APP to upscale it to bin1 size by inspecting the "File" pane (if that is what it's called.) I saw that APP had chosen a bin2 frame as the reference. I sorted the entire file list by quality and discovered that that frame did not have the highest quality ranking. I chose to override the reference with the highest ranking frame which happened to be a bin1 frame. I reran registration and integration. Now my LRGB image is perfectly sized.
Mabula, can you recreate this with the data I previously uploaded to the APP server? If APP doesn't choose a bin2 frame as the reference perhaps you can override it to see if you can replicate it.
Thanks very much Mabula.
Excellent news! About this specific case I think @mabula-admin can have a look, at least nice to know there's a workflow to get it right.
Here is a side-by-side comparison that shows quite clearly the superior alignment of APP vs Siril. Plus the color balance was better keeping everything within APP. Notice the green cast to the Siril image vs the blue cast in APP which is correct.
Yes, the alignment is one of the key strengths of APP, together with the calibration process. Mabula put a lot of effort in getting that right.