Siril plot feature

[chubster1302]

Has anyone come across any good explanations of how to evaluate the results from the plot tab and its various outputs? I "sort" of get the basics but im not fully understanding it. For instance what exactly is this telling me. Should i be binning all the images on the left hand side of this plot ? 

[scotty38]

I don't use Siril but I assume this is down to poorer seeing in the first half of the images due to high clouds, sky brightness or whatever. What "I" would do is stack them all and then stack just the second half and see if you can see any difference. I am never quite sure what a good or bad number should be so I take the approach I just mentioned, rightly or wrongly 🙂

Edited July 12, 2023 by scotty38

[chubster1302]

It was two separate imaging sessions which is why "I think"  the result looks the way it does, which sort of makes sense but yeah, its a lot of "guessing" for a newbie to understand. Im not even sure exactly what the green line indicates. What would be an acceptable FWHM for instance 🤔

[ONIKKINEN]

The plot here shows your measured star fwhm in pixels in the vertical direction and the frame number in the horizontal direction. Brutal honesty next: Your first night sucks compared to the second, no way to tippytoe around that. Either seeing was terrible or you were out of focus, or both.

Your second night is actually pretty stable. You can see that the general trend of the night was worsening with the occasional anomaly (the sharp spikes, investigate these to learn the reason. Could be wind or guide hiccups). The general worsening of the data can be explained by either focus drift or the target getting lower in the sky which will considerably start to hurt under 35-40 degreed.

You can learn a lot from these plots and take that information to the next outing. For instance it looks like you could use more frequent refocusing, and/or shoot the target while its higher in the sky.

The relative difference between the first and the second night is too high in my opinion and the first night might need to be scrapped. First i would try to stack with the wFWHM weighting option ticked and see what happens. That will give the bad subs a very low weight  but at least the data is used even if only in spirit. You can see the wFWHM plot if you change the Y axis selection.

As for what is a good value depends on a lot of things like your scope, camera and typical seeing conditions. Giving a number value for a good target is difficult and youll only learn your average sharpness by repetition. Convert the values to arcseconds (you need to know your pixel scale) to have a more useful metric than pixels. But for stacking purposes you want to have some limits to the relative difference between subs and i think even a 2x difference is a bit too high. Weigthed stacking helps but if i were you i would reject everything above 5px in this graph and stack the rest with weighting set to either wFWHM or number of stars.

On the flipside if you keep imaging the same target for more nights then it more or less evens out to an average of all the good and bad subs.

Personally i think a fwhm of around 3 pixels or less is a decent target. If considerably more, consider binning x2 after stacking.

[chubster1302]

2 hours ago, ONIKKINEN said:

The plot here shows your measured star fwhm in pixels in the vertical direction and the frame number in the horizontal direction. Brutal honesty next: Your first night sucks compared to the second, no way to tippytoe around that. Either seeing was terrible or you were out of focus, or both.

Inen

Many thanks Onikkinen, really appreciate the reply. I realise from the graph that basically anything above the green line might be considered no good and I had come to the conclusion that the first night was carp. So thanks for confirming. 

Where does the green line come from, how is it generated? 

[ONIKKINEN]

15 minutes ago, chubster1302 said:

Many thanks Onikkinen, really appreciate the reply. I realise from the graph that basically anything above the green line might be considered no good and I had come to the conclusion that the first night was carp. So thanks for confirming. 

Where does the green line come from, how is it generated? 

Not sure about the green line tbh, have not paid attention to it usually.

Siril documentation has this to say: Green curve: sorted values by order of decreasing quality

So some kind of quick representation of the loaded dataset.

[wimvb]

The green line is just the number of subs (x-axis) up to that fwhm (y-axis). There were a few subs with very low fwhm. The first increase (first half of the diagram) are actually the subs that you collected during the second night. The right half of the green curve represent the subs that you collected during the first night, where fwhm was generally higher. The graph shows that about 20 subs had a fwhm larger than 5 pixels. These are the peaks in the purple line. When you stack, you can use a weighting factor that takes fwhm into account. Subs with a low fwhm will weigh in more than subs with a large fwhm. In PixInsight I sometimes use the formula

weight = (max - fwhm) / (max - min)

where max is the highest fwhm, and min is the lowest fwhm. When fwhm = min, that sub's weight = 1, and when fwhm = max, that subs weight = 0.

As @ONIKKINEN wrote, the first night was bad. This can have several causes, of which you can control a few. What you can control is focus and to a point guiding. Either use a focus motor for autofocusing, or a Bahtinov mask. Focus can shift during the night (either due to temperature drift or the focuser slipping). If you look at the background median (should be a similar diagram in Siril), this can tell you if there were high clouds, low clouds passing by, or if the target was lower in the sky (more sky glow). Plotting the altitude that your scope was pointing at, can also give an indication. The capture software that I use (Kstars/Ekos) plots the telescope's altitude as well as the image's median value during an imaging session. As the targets altitude decreases after the meridian, the background median value starts to increase.

What you can't control (but can monitor) is atmospheric conditions. If seeing is bad, the guide curve will be noisier and the guiding error (rms) will be larger. This will give a higher fwhm value. On the other hand, high thin clouds (bad transparency) can give excellent guiding with low rms, but fat stars because their light is scattered in the atmosphere. Passing clouds raise the median background value and give fewer stars to guide on. The guide star signal will also be lower, so PHD will report lower SNR.