Explanation of Shape metric, FWHM,...

FWHM is Full Width at Half Maximum. So yes it tells you the "sharpness" of a star basically. Min is the minimum diameter it found in your data and Max the maximum. Your system will be causing this to be either bigger or smaller compared to other systems. So when you see a sub of your system and you zoom in and see very nicely focused and great guided stars, those values are about the best you can get with your system.

Right-clinking in the list pops up a list in which you can choose to show a graph, in there you can plot of the FWHM and other parameters. The subs with the smallest FWHM will have the best stars, but not necessarily the best signal etc.

Dear Vincent,

Thanks a lot for your reply. So do I understand correctly that there is no fixed scale for the values?! But I can assume that a smaller min FWHM is better? 

What about the shape value? Does a higher value mean rounder stars? Are there any bounds?

Kind regards,

Max

Not really a fixed scale no, a smaller value for the min will indicate that the smallest stars are even smaller indeed. 🙂

The shape is an indication of the roundness, and I have to say I don't really know which way is better. I'd think closer to 1? as the ratio of the width and height becomes more similar with roundness. But I don't actually know how APP is calculating this to be honest.

Thanks a lot for your reply. It would be nice to know what the shape value means though. I had images with stars without any visible trailing and a shape value of 0.5.

Kind regards,

AstroSchorsch

Hi AstroSchorsch,

So I had a deeper look in the forum, I found this explanation from Mabula; "Roundness is determined by the difference between min and max FWHM, if min and max are equal, you have perfectly round stars." which is similar to what I thought it would be.

I'd like to follow this up - I have to admit to being REALLY confused.

I have seen references to "shape" (aspect) as being that 1 is better and that lower is better. 

What should we be looking for in APP when it comes to "good" vs "okay" vs. "terrible" data with regards to shape?

I think best is to see it by eye as well. 1 is really good, the point at which you really wouldn't accept it is a bit of personal preference I think. I do have to say I don't really know the values you'd get with real bad data as I never looked at that on purpose. You could try that, take some subs with guidance issues (small and larger ones) and see.

Thanks Vincent, I’m getting 0.3, I guess that’s bad... 

Hi,

I never really got a grip of the shape metric. I just ignore it and check visually. The registration error is more telling in the end. However, I'm generally still unhappy with the provided metrics.

What would be nice:

- Properly document shape metric. How is it calculated? I would love to just have the formula to understand it. Is it long vs. short axis of an ellipse?

It would be great to have something like this implemented:

https://www.visiondummy.com/2014/04/draw-error-ellipse-representing-covariance-matrix/

caclulcate that per star, for different areas of the image,... this could quantify guiding issues, coma, sensor tilt,...

As I am a programmer myself, I'd even enjoy adding such functionality.

- I would love to have FWHM in arcseconds as well, I think all information needed should be in the fits headers? This would be a measure I could use to compare image quality over multiple setups.

Kind regards,

Max

@xthestreams Well, how does it look visually, do you have a screenshot zoomed in on the stars?

Thanks Vincent - attached. Excuse my terrible processing, the sad state of my flats (I have just gone back to mono and need to update my calibration files) and what might be a collimation issue (a GSO RCT is a great scope for the price, but wow, what a pain to get right!)

NGC 4945 might have slipped out of focus during the night (I am also trying to move to Voyager - not as much fun as kstars/INDI, but it works better with my Paramount). 

The reason I am looking at the star shape number is to try and work out if the collimation changes I am making are worse or better than before. I *know* the correct answer is start test, etc, but I would also like to be able to know if the numbers I am seeing in the Shape column mean that things are objectively better or worse, even if they are not 100% reliable. 

Right now, I really don't see that the numbers serve any purpose if no one can help me understand a "good", "bad" or "great" result in teh Shape column, which makes putting any number there mostly meaningless.

Not seeing the attachment. But I think you can say something with these values. Close to 1 is as good as you could get, 0.3 is quite a bit off and would indicate that could be improved. Now what the reason is for the star-shapes, that can be more than 1 thing and isn't something you can conclude from it. You'd have to be 100% sure that collimation is the only thing that is going on, then you maybe can use it. But even better would be to have a method, at the scope, that gives you the answer.

amazing - thank you

tried attaching again