AutoStakkert!2 quality graph

[[whrudey]]

Does anyone know what criteria AutoStakkert!2 uses when generating a Quality graph. I'm not sure that it is necessarily a good predictor of an eventual image in ImPPG.

[Davidm]

I'm not sure myself, but assumed it would be a relative scale for the individual capture and not an absolute scale that would allow a measure of absolute quality.

[eroel]

Bill:
Though AS!2 is a superb software, I have never been able to interpret the graph.
Regards,
Eric.

[pedro]

same here. I usually stack 100 to 250 images

[[whrudey]]

Seems that I'm not alone on this issue. I'd assumed that a quality graph like the following would render a poor image since little of the green line is above the 50% line. My assumption was that this was a reflection of the number of frames having better than 50% "quality" by whatever criterion.

AS2QG poor.jpg (12.25 KiB) Viewed 3750 times

In contrast, the following example should yield a good image for the above reasoning. Yet, in my experience, there seems little actual correlation between the two.

AS2QGgood.jpg (12.64 KiB) Viewed 3750 times

I'd also assumed that variables such as sharpness and grey scale were also involved in generating the graph. Any other ideas ?

As an alternative example, Avistack2 generates a quality curve and adjustable cut off level as shown here, which strikes me as more meaningfull:

AVSexample.jpg (39.46 KiB) Viewed 3750 times

Still confused !

Bill

[GreatAttractor]

I'm not sure myself, but assumed it would be a relative scale for the individual capture and not an absolute scale that would allow a measure of absolute quality.

I'm sure that's the case.

I'm not sure that it is necessarily a good predictor of an eventual image in ImPPG.

Bill, that's why I strived to provide convenient and responsive tuning of processing parameters in ImPPG, so the user can do this as often as needed. Any sort of "automatic processing depending on quality" would be a difficult thing to implement.

In contrast, the following example should yield a good image for the above reasoning. Yet, in my experience, there seems little actual correlation between the two.

Not a surprise, since the quality is relative. All the graph is telling us is that the second capture has more frames closer to the best (in this capture) quality.

Though there is a noticeable difference (as I remember from perusing AviStack's graph) in a really good seeing – then most of the frames are close to 100% and there is just a small "tail" below 50%. In average/bad seeing the distribution is more uniform.

I'd also assumed that variables such as sharpness and grey scale were also involved in generating the graph. Any other ideas ?

I suspect it's some sort of sum of what's returned by an "edge detection" (or variation thereof) filter. Anyway, that's how I've done it myself in my current project and it works very well

[[whrudey]]

We may have progressed a bit but, I' still wondering what the specific criteria are. May have to contact Emil for an exact answer - next step in this sage. Meanwhile, after scouring the Web, I found this:

AS2qualiaty graph.jpg (56.47 KiB) Viewed 3682 times

Cheers,

Bill

[eroel]

Bill:
I got this today while stacking a 600 frame avi, but I can not understand the graph.
Best regards, hope that Emil can help us.
Regards,
Eric.

[[whrudey]]

Have E-mailed Emil on this issue as follows and, am sure that he will provide a definitive answer:

____________________________________________________________________________________
As one of the many thousands of astro imagers who use your marvelous software application, AutoStakkert!2, I've often wondered what criteria are used in generating the Quality Graph. Is it based on sharpness, tone range or, whatever ? I posed this question recently on SolarChat with this link: http://solarchat.natca.net/viewtopic.php?f=4&t=18906 but, thus far no one has offered a satisfactory answer.

As you are the author, may I impose upon you to provide details ?

Regards,

Bill
____________________________________________________________________________________

Many thanks in advance,

Bill

[Marcello]

Very interesting post, hope Emil will give us answer

Hi GreatAttractor, your new project looks very interesting!

[AutoStakkert]

The quality graph is indeed relative, and I would not trust it that much, because one outlier, good or bad, will skew the graph and make the 50% 'rule' not applicable anymore. To be honest, I don't really pay much attention to it.

I often just manually skip through the frames and judge the seeing that way manually. If I'm not sure how many frames I want to stack, I'll try different stack sizes. Keep in mind that the difference between the best 40 and best 50 percent stacks is very very small, and depending on how you process your images, you might get more out of a larger stack - one containing more frames - than one containing fewer but sharper frames. Some people even use combinations of stack sizes (e.g., the best 10% sharpened a bit, and overlay this with a 50% stack), but I pretty much never bother with this.

The quality graph is based on measured contrast (differences of intensity from groups of pixels to nearby groups of pixels); pretty basic stuff.

Don't forget that frames that fall below the % point you set to stack, can still be used during stacking, because each AP typically uses it's own quality local quality estimator.

(p.s. I thought I had an account here, but it does not recognize the username/email, although according to the confirmation email I'm sure I registered with the credentials I tried).

Emil

[Valery]

Thanks a lot, Emil. All is clear now.


Valery

[[whrudey]]

Many thanks Emil - your input is invaluable as well as very helpful.

Bill.

[eroel]

Emil:
Thank you very much for the explanation and nice to have you here in this forum.
Best regards,
Eric.

[Marcello]

Hi Emil,
thank you so much for the explanation (and for sharing your excellent software, of course)

Marcello

[grimble_cornet]

Very useful information - thank you.

[pedro]

Thanks Emil.