I tried searching for this on the forums, but I couldn’t find anything quite like my issue. Notice the stars says “0” for the last few images. 0017 was the last one sub from the previous night. So it seems to be happening randomly last night, and tonight (0018 and later) they are all coming up zero.
Here is the log from this session tonight. Any suggestions would be great! Thanks.
sg_logfile_20160928220222.txt (192.4 KB)
Here is frame #0018. Let me know if you have any problems or need more frames.
https://dl.dropboxusercontent.com/u/54413223/Sh2-129_LPS_1200sec_0018_-10.5C.fit
Thank you for looking into this. I know you guys are busy!
I wonder if it has something to do with my exposures? The file I uploaded for you is a 20 minute exposure. Tonight I have been shooting 10min and 1min and I have not gotten this “zero stars” issue at all.
It looks like our star finding algorithm hates this part of the image:
I can’t really say what is creating that artifact…
Hmm… I have no idea what that spike could possibly be. I don’t even know how I would figure it out. I tried adjusting the sliders so it surrounded the one spike, but there was nothing obvious in the display. Just looked like a blank screen.
The same spike actually exists in all my subs. Here is #0013 which was able to successfully tell me the number of stars from the image history window. The same spike can be seen in the histogram.
https://dl.dropboxusercontent.com/u/54413223/Sh2-129_LPS_1200sec_0013_-10.2C.fit
Any idea why one of them works and the other one does not?
Not immediately… I would need to review our auto stretch code… it’s been a while since I looked at that. It might possibly be an overscan region (on one or more of the edges). A spike like that means that you have some very dark pixels in the image somewhere and they appear to be below the mean background ADU. I though we checked for this, but I could be wrong.
Ahhh. Well that’s probably it then, because I just tried moving the sliders again and that is the only thing left when I move them to encompass the “anomaly” in the histogram.
So in my mind, the only mystery left is why some subs work and some don’t. Ordinarily, I wouldn’t mind so much. Image History doesn’t matter that much to me, but I think it also has an effect on the “Grade Images” tool, which I really like using to help me find bad images. When I run “Grade Images” on all my subs, it always fails to grade anything for those subs that it doesn’t like. Must be using the same code for star detection.
Check this out… Looks exactly like the Image History from above…
Well, OK… not EXACTLY like the Image History, because the HFR and star counts are different from the history vs. the Image Grader. But the lack of data for those subs is the same for both.
It will also adversely affect auto focus if no stars are found… that said, this issue may not present itself when the integration time is only 10 seconds or so.
I have finally had a chance to take a look here. I will introduce a bad fix into the next beta. With that artifact in place (the one on the left which I believe is due to very dark overscan regions) this fix might or might not address the image history issues (more data required to validate that). What it should address is (and I’m not sure if you ever experienced this issue) is the problem where auto focus frames with this overscan artifact would return no stars.
Thank you SO MUCH for looking into this!! From the sounds of your comments, I was afraid you had decided this wasn’t worth developing into a software fix.
No, I have never had issues with autofocus. When I do autofocus, I always get plenty of good stars. That being said, its possible that I’m avoiding the problem by using the option to “crop autofocus frames” by X percent. I used it for a long time when I wasn’t using a field flattener, because I wanted to avoid using elongated stars for computing the HFR. My settings are currently to crop by 10%, but I’m not sure if that is enough to crop out the overscan regions. Can’t take a sub right now to check, but I will net chance I get if you think it will help?
Right now AF cropping will not help the issue because it is an artificial crop… meaning we still perform a full image histogram transform and image analysis, we just ignore found star data in the cropped areas. In the future, this will be smarter.
