Specify reference frame?

Questions, support and discussions
szymon
Posts: 24
Joined: 29 Aug 2019, 18:17

Re: Specify reference frame?

Post by szymon » 01 Sep 2019, 13:58

I turned on the extended solver log in case it's useful:

Code: Select all

12:57:23  Working on star alignment solutions. Blink frequency will increase after completion.
12:57:23  Start finding stars
12:57:24  1133 stars found of the requested 500. Background value is 58. Detection level used 502 above background. Star level is 502 above background. Noise level is 20
12:57:24  Selecting the 500 brightest stars only.
12:57:24  Finding stars completed in 984 ms
12:57:27  Start finding stars
12:57:29  673 stars found of the requested 500. Background value is 500. Detection level used 54 above background. Star level is 54 above background. Noise level is 41
12:57:29  Selecting the 500 brightest stars only.
12:57:29  Finding stars completed in 1782 ms
12:57:29  79 of 88 tetrahedrons selected matching within 0.005 tolerance.  Solution x:=1.000141*x+ -0.000010*y+ -0.369074,  y:=-0.000136*x+ 1.000218*y+ -0.000651
12:57:31  Start finding stars
12:57:32  930 stars found of the requested 500. Background value is 289. Detection level used 685 above background. Star level is 685 above background. Noise level is 27
12:57:32  Selecting the 500 brightest stars only.
12:57:32  Finding stars completed in 1016 ms
12:57:32  88 of 100 tetrahedrons selected matching within 0.005 tolerance.  Solution x:=0.999949*x+ 0.000028*y+ 0.091694,  y:=0.000014*x+ 1.000108*y+ -0.243695

szymon
Posts: 24
Joined: 29 Aug 2019, 18:17

Re: Specify reference frame?

Post by szymon » 01 Sep 2019, 14:00

Here are links to the other two frames (Ha and Oiii) in case they are useful:

https://www.dropbox.com/s/46rk06glo831t ... d.fit?dl=1

https://www.dropbox.com/s/svea25jyjw2eb ... d.fit?dl=1

These don't have the artefacts when the alignment is done on all three -- only the Sii image posted earlier.

-simon

han.k
Posts: 235
Joined: 23 Nov 2017, 15:04
Location: The Netherlands
Contact:

Re: Specify reference frame?

Post by han.k » 01 Sep 2019, 16:50

The lines are a kind interference patterns. If you stack images which are shifted and rotated at some pixel places are not covered. You get it when you stack a few images. It is like spreading an 999 pixel image of on a 1000 pixel image. It doesn't fit.

The value at the lines is zero.

Normally the program will detect if pixels are not covered and will fill in the positions using value of neighbor pixels. So it is rare that it does not happen for your stack of 6 images. I haven't seen this effect for 6 months or longer.

If it happens again, pleas ask for help and I will investigate in detail.

szymon
Posts: 24
Joined: 29 Aug 2019, 18:17

Re: Specify reference frame?

Post by szymon » 01 Sep 2019, 16:55

Actually the six frames together stack just fine, without the lines -- it's when we stack the three frames (for alignment) that we get the empty space. Are you suggesting that if my Sii stack has more data then it will go through the alignment just fine, without ending up with empty space?

szymon
Posts: 24
Joined: 29 Aug 2019, 18:17

Re: Specify reference frame?

Post by szymon » 01 Sep 2019, 17:34

Ok, I think that this was due to one of my stacks being 'upside-down'. My Ha stack had a meridian flip in the middle, and while stacking was able to work just fine, it produced a stack that was 180 degrees to the Sii and Oiii stacks. This appears to be what caused the artifacts when aligning the three stacks. When I flipped the Ha stack 180 degrees in gimp and did 'export stacks' again, I got perfect stacks without alignment artifacts.

han.k
Posts: 235
Joined: 23 Nov 2017, 15:04
Location: The Netherlands
Contact:

Re: Specify reference frame?

Post by han.k » 01 Sep 2019, 17:44

Meridian flip should do no harm. If it happens again, please report.

Regards, Han

szymon
Posts: 24
Joined: 29 Aug 2019, 18:17

Re: Specify reference frame?

Post by szymon » 04 Sep 2019, 12:44

Hi Han,

So, it seems that this happens every time when exporting the aligned files, and it is reproducible. Here is an example for you to download and try out:

https://www.dropbox.com/s/winq6dsft5nie ... t.rar?dl=1

There are four stacks:

2019-08-27_03-59-10_CLS_19.70_300.00s_1@ 20xFD 20xF 4xD 12xCLS_stacked.fit
2019-08-27_03-51-44_Sii_19.60_300.00s_6@ 20xFD 20xF 4xD 6xSii_stacked.fit
2019-08-27_03-16-37_Ha_19.70_300.00s_12@ 20xFD 20xF 4xD 18xHa_stacked.fit
2019-08-27_01-58-58_Oiii_20.50_300.00s_6@ 20xFD 20xF 4xD 6xOiii_stacked.fit

I have also included the 'aligned' versions of the stacks. Here is how to reproduce the problem:

- Load the four images up into the blink window of the latest ASTAP for windows
- Run the blink once so that it aligns the images
- Look at the images as they blink through -- they are absolutely fine, no black lines
- Then click on "Export Aligned" - at least one of the exported images will have black lines

Every time I try this, with various sets of data, at least one of the files comes out with the black lines. In the example data, it's the CLS filter, which you can see by looking at the files.

Here is a screenshot of the Luminance stack (CLS filter) which has been aligned, but not exported:
cls_aligned.jpg
cls_aligned.jpg (1.1 MiB) Viewed 683 times
Here is a screenshot of the same stack after clicking the export aligned button:
cls_aligned_exported.jpg
cls_aligned_exported.jpg (1.09 MiB) Viewed 683 times
I hope that helps :-)

-simon

han.k
Posts: 235
Joined: 23 Nov 2017, 15:04
Location: The Netherlands
Contact:

Re: Specify reference frame?

Post by han.k » 04 Sep 2019, 13:24

Hello Simon,

I will have a look

Han

han.k
Posts: 235
Joined: 23 Nov 2017, 15:04
Location: The Netherlands
Contact:

Re: Specify reference frame?

Post by han.k » 04 Sep 2019, 14:18

Now it is clear to me. By aligning (shift and rotate) the pixels you get pixels positions which are missed. Like trying to spreading 999 pixels in a line over 1000 positions. At one place there will be a hole or a black spot.

In ASTAP if images are stacked, it keeps a record of how many pixels land at a pixel position. The summation is later divided by the number of pixels added. So if you stack 10 images. Some pixels will get the summation of ten values, other 9 and a few 8 due to outliers. That is not a problem as long you keep a record of the number of pixels summed/stacked.

That doesn't happen with the new export. The images are not stacked yet.

There are three options:

1) Calculate on sub-pixel level.
2) Interpolate missing values and fill the black positions.
3) The program you use detects the black spots as outlier and fix the stack.

I assume 3) is not the case, so option 1) & 2) looks like the best options. For 1) I have to spread each pixel in ratio over about 4 pixel positions max. Option 2) is also possible. I assume with option 1) you loose resolution. A single pixel star could will be spread over 4 pixels if there is a 0.5 pixel shift in x and y.....

szymon
Posts: 24
Joined: 29 Aug 2019, 18:17

Re: Specify reference frame?

Post by szymon » 04 Sep 2019, 15:15

Option 2 seems the most sensible to me, but of course I leave it up to you if you can implement such magic :-)

Post Reply