mschuster's Scriptbox

[mads0100]

No kidding?  What's the name of it? 

Do you happen to have an example 'approval' script?  I just don't quite get what it's telling me for creating the expression and I want to see one before I make my own.

[Sean]

Mike,

Another vote of thanks for SubframeSelector. I just did a little experiment with 20 luminance frames from last night. I went through them visually using Blink, and carefully selected the frames that I thought I should reject. This is a very subjective process, and I've never really liked doing it, plus it's quite slow for an entire night of imaging LRGB frames.

I then loaded the 20 frames into SubframeSelector, and sorted on FWHM, SNR, and Eccentricity, noting the frames outside the plus or minus 1 sigma levels. SubframeSelector detected exactly the same subset of bad frames, and best of all I used it to automatically copy the good and bad frames to sub folders for processing. This gives my a lot of confidence that I can now adopt a more quantitative approach to subframe selecting, plus it's much more efficient.

Thanks!

Sean

[mads0100]

Hey Sean,

    What was the expression you used for eccentricity?

[EorEquis]

Mike,

Add in another thank you for the SFS.  It's made a huge difference to my end results.

Do you happen to have an example 'approval' script?  I just don't quite get what it's telling me for creating the expression and I want to see one before I make my own.

For what it's worth, I posted my own, along with the thinking behind it, here, in a thread where Mike was kind enough to explain some terms down to my level of ignorance.  *lol*

It's working pretty well for me, and has shown good results for one other user there, so perhaps it can serve as a starting point for you.

[Sean]

What was the expression you used for eccentricity?

Currently, I'm just looking at the plots to see the trend, and clicking out the images that are outliers. I'm still trying to characterize my system and seeing, so don't have any hard and fast numbers yet.

Mike suggests in the documentation that Eccentricity less than around .42 is not detectable to most people.

Sean

[mads0100]

Thanks for the tip. I didn't link the documentation.

[jeffweiss9]

Mike-
After using SubframeSelector.0.6 successfully for some time and and relying on i, I hit a crash in a series of unbinned luminosity subs I took the other night.  Here is the console log:
 
SubframeSelector.0.6:

Measure: Processing subframe: 1 C:/ML8300/m31/October 31 2013/TestBld7314.00000327.Mosaic  A4.fit
Reading FITS: 16-bit integers, 1 channel(s), 3326x2504 pixels: done

StarAlignment: Processing view: TestBld7314_00000327_Mosaic__A4_target01
TestBld7314_00000327_Mosaic__A4_target01:
Structure map: done
Detecting stars: done
3739 stars found.
Blending grayscale bitmap: done
2.954 s
*** Error [000]: C:/Program Files/PixInsight/src/scripts/SubframeSelector.0.6/src/SubframeSelectorMeasure.0.6.js, line 227: Error: Image.sample(): coordinate(s) out of range.

This is the first time I had any kind of problem but I get the same failure for the whole series of luminosity images I took Thursday, while all the RGB binned images ran fine, as normal.   If you could take a look, I can send you the fits file that generated this error.  Comparing the fits header with that from another luminosity image from a previous session that SubSelector liked fine, I don't see any differences whatsoever.  These 'problem subs' are processed successfully by PI v1.8.

Thanks very much.
-Jeff

P.S.  PSFEstimator.34 also has the same error, while FWHMEccentricity0.7 does not crash, if that is any hint.

[pfile]

jeff - really 0.6? 0.95 is the latest released version right now i think.

rob

[mschuster]

Hi Jeff,

Sorry for the problem. I am imaging at a mobile site in Arizona now, and I will investigate as soon as I get home in a couple of days.

Regards,
Mike

[mschuster]

Jeff, yes Rob is right, there is a new version available. It should download in an automatic update for 1.8, but if not you can download it from a link above. This might be a quick fix. I will investigate in any case when I get home.

Thanks,
Mike

[jeffweiss9]

Mike (& Rob)-

Thanks. Yeh, I picked an old one from the script pull-down.  v0.95 doesn't have this problem and I just deleted 0.65.  So no issue.

...except that, to me, aspect ratio (ratio of minor to major axes) is much more physical than eccentricity (or at least it's what I'm used to).   Any chance you could add that to your latest FWHMEccentricity.js.  Then I could stop using PSFEstimator in the field for individual images (although I still would later read the SubframeSelector  .csv into Excel for my plotting and, if necessary, convert to aspect ratio as sqrt(1-ecc^2)).

They are really great, useful scripts.
-Jeff

[mschuster]

Hi Jeff,

Home from AZ now, I am glad you got it working. Aspect ratio is been on my to do list for a while, I deleted it from the newer scripts to keep things simple as possible, that may have been a mistake, and I have been way too lazy to reimplement it.

Jeff, Sean and EorEquis, thank you for your comments!

Regards,
Mike

[fulatoro]

Mike,

Do you have a write up on how to use a B-mask to test squareness? I have been assuming that I can simply use the eccentricity data to help fix my tilt issues. Am I being too optimistic?

Moussa

Jim, thank you very much for your comments!

Riccardo, I am not motivated to work on this. IMO a single frame just does not give enough information to determine a focal surface and aberrations without strong, simplifying assumptions. For example, FocusMax needs to make multiple exposures to find the location of the focal surface at just one point on the frame. In my tilt example above, there is no way to tell which corner is inside/outside of focus given one exposure. I like using the Bahtinov mask for squareness measurements because it provides more information.

Regards,
Mike

[mschuster]

Moussa,

IMO an eccentricity map is not a good way to correct tilt. I like the B-mask for this purpose. I suggest targeting a field of brighter stars. Use the mask to accurately focus a frame center star, and then check the diffraction patterns of the stars in the corners. Inside or outside focal surface errors will be visible in the corner diffraction patterns. You may need to slew a bright star to each corner in turn. Use a green filter, this improves the legibility of the diffraction patterns on refractors with a bit of chromatic aberration like the FSQ. With an accurate digital focuser you can repeatedly refocus the corners and measure the amount of tilt. On my FSQ setup I use a detilting device for correction. See my write up here.

Mike

[vicent_peris]

Hi Mike,

One question. I'm using now the SubframeSelector and it detects between 100 - 250 stars/image. It always says "Warning: insufficient number of fitted stars". I think 100 stars should give good statistics...?

An one suggestion. Some cameras don't have the saturation point at the top of the digital counts. Could you add a parameter to specify where is the saturation point?


Thanks,
Vicent.