Blinking and Rejecting Light Frames

Matthew Humphrey

In the FastTrack and Fundamentals videos, Adam comments on several occasions that he would not throw out a Sub-Fram just because it has a satellite trail, or other imperfections.

I'm curious as to how bad a frame would need to be in order to remove it from the set prior to processing.  Or, can PixInsight be trusted to exclude data that is bad enough.  As per the videos (that I've watched so far), I am using .500000 for the Minimum Weight value in the Image Integration setup.  

I've attached a couple of examples of sub-frames.  One which is adequate, and the other which (in the past) I would reject and remove from the set.  You can see the stars trail in the second image.  Using this example, would it be recommended to remove this file before processing?  or leave it in and let PI do it's thing?

BlockHead

Hi Matt,

There are two parts to the answer.

1. Does the image in question represent a small number of the total number of frames? Do you have many better quality images? If so, you can safely leave the frame in. Rejection and weighting will properly add it for what it is worth (or not use it entirely based on the weighting threshold).

2. How much useful data does the image have? A satellite trail or airplane does not affect the useful data of the object in the field. You received all of the galaxy's (or whatevers) photons of light in addition to the extra garbage. Poor tracking and focus is different. You are changing the amount of useful signal. If #1 above holds...it is OK to include and let rejection and weighting do their thing. But, it could be that you know a priori the frame will not pass the test. Here is an example. I take a 15 minute exposure. And for 10 seconds out of 900 seconds the telescope dances a waltz. Right? Now every single bright star in the image looks crazy! But I would not throw this away. There are 890 seconds of PERFECTLY guided data there. So rejection will take care of the 10 seconds of garbage. If, however, you have an image where the exposure time was 60 seconds (or less) and the poor guiding means NOTHING was tracked well during the exposure... well then the frame has little useful data. It probably will not pass the weighting threshold. You can include it ..no problem. Or you can remove it because you already know the answer it is one less image to work with. 

The point is make is that you can trust the math for rejection and weighting to do their thing for anomalies. If the average quality of the frames looks like "crap"...then the integrated result will too. No magic there. This is about the convenience of weeding out the poor stuff in a sea of lots of better quality stuff.

-the Blockhead 

Matthew Humphrey

Hi - Thanks for the excellent explanation!

I think this played out exactly as you describe.   I had 150 files, and after going through all of them, there were 19 that I would typically throw away because of star trails/poor guiding.  However, I let the process run with all 150 included.  When I looked at the final image, in the FITS header, "ImageIntegration.NumberofImages" is 131.

I take this to mean that PixInsight and I were on the same page, and it tossed out the 19 that were of poor quality. 

Thanks for clarifications.