Comet Operand doensn'y match comet position

John Kmetz

Have been trying to follow the comet subtraction techniques as outlined on 1.27.23. After trying this instruction closely 2 times now, my comet subtracted images still have the comet and a wondering black hole which moves further away as the blink session goes on. Either I don't have the right operand in place, or something else is out of order.The comet stays in place, the black hole drifts away. They don't match , the hole should follow the comet?

Any idea how to fix?

Thanks,

John

BlockHead

Yes... you need to pay particular attention to reference frames.

Let me ask you these questions:

1. What was the StarAlignment reference frame that all of the images are aligned to?

2. What frame did you choose to align your comet images to as a reference?

The answers to #1 and #2 need to be the same. It *has* to work for at least this frame (subtracting the comet). If then the subtraction drifts off... it means the calculated path is not correct for some reason. You can generate the path mask and see the behavior and compare with your images. BUT... you can literally take this calculation out of the equation if you force ever position to be a fixed position. 

I suspect this issue is the reference frame thing and not a calculation of the comet path.

-the Blockhead

John Kmetz

Q1. Yes. I used the same reference frame which is my Blue 30.

Q2. The same Blue 30 for the comet alignment.

But I found one mistake. I had Star Aligned checked below the operand input field. So this caused the output images I had in Blink to be Comet Aligned, and the full original comet stayed in one place while the Blink stars were moving. But when I selected Comet Align below operand, as your Jan. 27 Comet subtraction video (~6:30), the Blink images are now star aligned and the comet moves. (This seems like reverse logic to me but it works as is the video).

But now my problem has changed. The hole now follows the comet at the same distance as I blink through with the stars stationary. So my Blue Comet image I have used as the operand is in the wrong place. I need to restack the Blue comet only images again somehow into the correct position.

Sorry I have missed something here. Will try again.

John

BlockHead

Yeah, good. I am pretty sure I mention this in my video. 

The "Comet Aligned" is describing the operand. It make sense if you think about from the operand point of view. 

Concerning the offset..it really does mean there is a discrepancy in chosen reference.

-the Blockhead 

John Kmetz

Thanks Adam. I see how the operand works now. Yes, my error was in the reference frame selection, which was readily available in the log. I need to redo all steps with the correct one in place.

I have now renamed this file with a _ref ending to make sure the same one is always selected.

John

BlockHead

Great.

If you feel I did not make this clear enough in my video- let me know and I will try to add more language.

-the Blockhead

John Kmetz

Thanks Adam. The log shows the reference file name and location. Remembering this name and location hours or days later was the challenge I think, as the multiple steps take a bit of time on my slower PC. But now with the _ref ending I can find this much easier. One thing to note is the reference file can also be used with ImageIntegration, to further assure alignment. So now I have the operand in the right location everywhere.

Sorry to report I have another issue. Even with the operand subtracted, the comet core is still showing through on the blinked comet subtracted images. First I tried the linear Comet Blue integrated image, then a stretched image, and then an overstretched image with the core highly brightened. But none one these blacks out the core except in perhaps a little in one frame. I also tried a Game mask covering the core as the operand, but this just leaves a big black spot where the core was. So even though a have it place right now, the subtracted images are not right.

Any suggestions welcome.

Thanks,

John

BlockHead

The subtraction will never be perfect. You are fighting laws of nature and variability here.

Just be certain you are subtracting like colors for the operand. You can also try turning off the linear fit and see if that helps.

Remember, you will be integrating these images...so there will be some additional rejection as well.

-the BLockhead

John Kmetz

Thanks Adam. Should be able to finish from here I think.

John