SHG software spectral line curve-fit fails often

[stefanoaz]

Hi All:

I'm been using the SolEx for about a week. I'm getting good spectral lines and have gotten my focus technique down to where the mount slew jitter is the limiting factor (new mount and Evostar 72ED coming soon). Since my two good scopes are both SCT and not (presently) usable with the SolEx, my temporary "test" telescope is an Orion f/3.4 guide scope with 2x barlow and added backfocus, with ND16 Hoya filter and Astronomik IR-block filter. Attached are recent scans in H𝛂, H𝛃 and Ca-H.

I''m finding that INTI, D. Smith's SHG and JSolex all have occasions where they fail to detect the spectral line in my ROI. I make sure to exclude other significant lines from the ROI, I've tried different exposure levels (short of saturating the histogram), but I get failed line detection on some SER files from all three programs.

INTI gives me the highest percentage of line detection fails, SHG fails the least and JSolex is somewhere in the middle.

Does anyone have suggestions for debugging this? I imagine others here have run into this issue - what have you found that helps makes this function more robust?

Steve

Sun_111249_20240521_colorized_0_00.jpg (471.36 KiB) Viewed 445 times

Sun_112126_20240521_mix_0_00.jpg (419.94 KiB) Viewed 445 times

Sun_114053_20240521_mix_0_00.jpg (473.65 KiB) Viewed 445 times

[stefanoaz]

One big correction to my original thread. D. Smith's SHG actually did not fail on a single one of the scans in question - so he seems to have solved a software issue that some SER files experience with the original INTI curve detect/fit, and the routine in JSolex.

One thing about my SER images for that session which produced some line detect fails, is that the image has a fair amount of tilt - 10-15 degrees due to not aligning the plane of the SolEx with the RA rotation plane.

Steve

[stefanoaz]

Here is an example of an unprocessed image set produced by D. Smith's SHG, which succeeded in his, but fails in INTI and JSolex:

Sun_110717_spectral_line_data.png (485.02 KiB) Viewed 395 times

Sun_110717_shift=10_ellipse_fit.png (479.01 KiB) Viewed 395 times

[Montana]

These are a fantastic set

Alexandra

[stefanoaz]

I found that the exposure/gain of my SER videos was a bit low - this is particularly a problem in 16-bit mode. Also - I'm using my color planetary camera for this - the ASI178MC, in bin-2-monochrome mode. There is loss from the Bayer matrix filters that I would not have with a monochrome camera. Today, with brighter video I got consistent reading from all three programs I use.

I try not to acquire things until there's some evidence they will enhance results - astrophotography can be a money pit. In this case a monochrome camera is justified: I'm undersampling with respect to the optics of the 72ED, and the extra gain would allow higher slewing speed.

Here's the latest H-alpha, H-beta and Helium-D3 (looks like I saturated the image at the center, but other than that the offset technique worked. I used Cédric's Spectral Editor to find the pixel offset for the emission line, but I found that I got a better difference image from INTI than I could get from an image-math script with jSolEx.

H-beta

Sun_083619_20240525_mix_0_00.jpg (398.97 KiB) Viewed 303 times

H-alpha. The prom at 8-oclock is the sunspot region rotating into view, that gave us the auroral display on May 10.

Sun_090603_20240525_autostretch_0_00.jpg (410.07 KiB) Viewed 303 times

Helium-d3 emission line. Looks like I overexposed the frames near the center.

_Sun_082445_free.jpg (437.2 KiB) Viewed 303 times

Ca-K

Sun_084604_20240525_mix_0_00.jpg (389.25 KiB) Viewed 303 times

[Montana]

Superb images I do prefer the monochrome versions (but that is my own personal preference )

Alexandra