19 June 2021: Our upload server https://upload.astropixelprocessor.com/ has been migrated successfully to our new office with higher upload and download speeds (nearly 10MByte/sec up/down ) ! We now have 1 general upload user called: upload with password: upload. The users upload1 - upload5 have been disabled.
31 May 2021: APP 1.083-beta2 has been released ! APP 1.083 stable will follow soon afterwards. It includes a completely new Star Reducer Tool, New File Saver Module, Improved Comet registration and much more, check the release notes here!
Frame registration never finishes.
I have loaded 400 light frames, darks, flats etc. to run an integration. But the registration never finishes. APP runs through Blue, Green, Ha, Lum and Red, but when the progress bar reaches 100% it returns to 0 and APP starts over with the blue frames. One registration takes about 2 minutes.
I started it, went away for an hour and a half and when I came back APP was still doing registration, over and over again.
Running 1.83 beta 2, but the same happened yesterday with beta 1. All settings are default.
Files are from ASI1600.
Any ideas, anyone.
11:33:52 - 4) REGISTER: setting a new reference frame: from instrument ZWO ASI1600MM-Cool
11:33:52 - 4) REGISTER: setting a new reference frame: from pixelcount 16 megapixels
11:33:52 - 4) REGISTER: setting a new reference frame: from channel Luminance
11:33:52 - 4) REGISTER: setting a new reference frame: based on star analysis results: NGC 2841_Lum_HFR-2.79_G75_O13_30.00s_0086.fits
11:33:52 - 4) REGISTER: starting normal registration mode...
11:33:52 - 4) REGISTER: normal registration: creating reference pattern recognition descriptors...
11:33:53 - 4) REGISTER: created pattern recognition descriptors of the reference frame
This is the output after finishing registration of the last red frame. Thereafter APP starts over with the blue files. And it just keeps going.
That is interesting, so it keeps looping through the data at the register step? Never seen that before. I'll probably ask for some of the frames you have to see, but could you try to see what happens when you use 100 in total?
I just ran a different set of files (different target), 159, and that worked OK.
When this has finished integrating, I will try again with just 100 files as you suggested.
Thanks! Not sure why it would do that, but you can try to do 100, then 200, then 300 and finally 400 again. May take some time of course, but it could point to the root cause somewhere. Which would be interesting for us as well.
There are some oddities here that I don't understand:
I load 402 lights, but registration counts 401. It always count one below it seems.
If I uncheck lights, they are still included in the registration process as long as they are loaded. I have to physically remove them from APP to exclude them. Why?
I have started loading color by color, all blue OK, all green OK, all Ha NOT OK. Then registration failed.
Then loaded 10 Ha OK. 168 lights in total. I will try more tomorrow. It is very time consuming since star analysis has to be redone every time files are added.
I'm not sure if this is numbers game alt all. I have done much larger stacks than 400 previously. I also have removed all calibration files except darks for these tests so that will not interfere.
I have been back and forth with files for hours and and APP has showed a few irritating solutions.
1. Every time I add more light frames and press analyse start, APP asks me where I want to store my project. I'm sure I have selected that folder 20 times during these tests. Why can APP not remember this path?
2. If I add more lights and press analyse files, the existing analysis is deleted. So if I and one more light fram to an existing 300 everything has to be re analyzed? Why is this?
I propose to change the way this works if possible.
So I have run this integration color by color and now I'm running everything. I removed some bad blue files so now I'm down to 386. Today I am not able to recreate this error. I even let the PC running over night not to change anything.
For a while I was pretty sure something was off with one or more Ha files, but now they are running fine too. So I don't get it. Some glitch somewhere that is now gone.
Not sure if I should say problem solved, but at least problem is gone, for now. 🙂
@heno Good to read that you are making progress on this. Some answers to a few of your questions:
APP registers all files against the reference file. It doesn't have to register the reference file against itself, which is why you'll always see that N-1 files get registered with N the total number of files.
How do you uncheck the lights? In the file list at the bottom? If that still includes them in the stack then that sounds like a bug to me. I'll check with Mabula.
About the popups that ask you where to save the files: this has been reported and Mabula is looking into modifying that behavior.
Adding more files may change the reference frame, which would require all files to be registered against the new reference frame. But if the registered frame doesn't change then it makes sense to keep the registration. Mabula is looking into optimizing this for a future version and I'll mention this to him as well.
I used the panel in the bottom to uncheck the files. At first attempt I unchecked 100+ files to get below 300, but the console output still stated that 401 files were used. It seems like a bug to me. But som many strange things happened yesterday.
I can see that if the reference tile changes everything has to be re analyzed. But is it not during star analysis the reference file is selected? So how would APP know when to re analyse or not? Maybe it could be used selectable to keep current reference file/re analyse? Just asking. I'm on thin ice here. 😉
Good to see it working Heno, seems like there might have been a system issue and maybe restarting helped out? Still weird indeed. These things can be challenging to find out.
Concerning your other questions; we are going to make some changes to when the user is asked for directories and such. You're right that registration always starts, Mabula has mentioned that he may change this as well as that would indeed make more sense. Thanks for that tip!
APP does redo everything when you change the reference, this is because it has to register now in relation to that other file. You can select a reference frame manually though, after the analysis step you can set that in tab 4 (register tab). You click the "set reference" button and then select the frame in the list below. Not sure if that works when you redo analysis all the time, that is something people don't usually do. 🙂
just let me see if I understand the star analysis correctly.
Am I correct assuming that first APP analyses all the light frames and then, based on the result, it picks the best frame for reference?
Are any calibration files used during this step or does that happen later?
If this assumption is correct I see no reason why APP should ever redo the star analysis for already analyzed lights. APP already have that result and all that remains is picking the best reference frame when the additional light frames have been analyzed.
Well I guess that may be the case (I would have to know about exactly how that algorithm works), but I don't think it does that when simply redoing the process. What you can do is skip the steps you already did, for that you'd have to save the frames in between. So let's say you register all frames, then you can save the registered frames, load them in when opening APP next time and select "disable star analysis" on tab 3. Because they are saved with the registration in them, that is not needed again. That will registered to a particular sub then though, so not really a solution for you in this case.
I'll ask Mabula if it would be possible to save the star analysed frames as well, that would make it possible to load those frames in and select different references I think.
edit: Or, maybe that does work actually? I've never tried this so bare with me. Try saving the registered frames, then next time disable star analysis and pick a different reference frame manually. It does need to register again then anyway, but I think this is because of how the algorithm works (given distortion etc). I've asked Mabula about this so will come back to it. Good question!
Apologies Heno for my delay in joining this topic. Let me clarify everything now:
First of all, if you deselect frames, APP will still try to register them. If you don't want to include them, you should remove from the list for now. I will change this behaviour a.s.a.p. that if you deselect them, they will not be considered for registration and (also not for star analysis).
Secondly, if registration comes into a loop where it never finishes, it means that whatever registration frame is chosen, APP can not register all light frames, so there must be a problem with some of the frames then. First it tries to register on the reference frame as chosen after star analysis or the manual one with the select reference frame button in 4). If that fails to register all frames, then APP will choose another frame and tries again. Now it is clear we need to change that behaviour as well. Probably, APP should just pop-up a message stating it could not register certain frames and ask you if you want APP to try again with a new reference frame or just stop the process to study the frames that could not be registered.
APP 1.084 will and should address all the other inconveniences. We will make it possible to load/save settings like the Work Directory and load/save complete projects. Furthermore, I will include in that version, that if you have registered 100 frames, and then add 50 more. APP will only do star analysis and registration on those new 50 frames, preserving the analytical results already present 😉 ( Will make this work from 3)star analysis to 5) Normalize ).
Please let us know if this answers all the problems that you faced.
Hi @mabula-admin, @vincent-mod, @wvreeven
Sorry for my late reply. I have been at my cabin where internet is not the best.
Mabula, thanks for coming back to me. The answer to your bottom question is yes, I think it does.
I will be most excited to see the new version with all the changes incorporated.
Thank you for listening in and your responsiveness to user problems and suggestions.