Livestack alignment settings

[clouzot]

Sorry Robin, this may look like the never ending story, but we see beginners hitting the same issues again and again: when Sharpcap refuses to stack, it's really hard to find the correct settings, and there's no way to guide them. Nothing that cannot be solved with the correct gain/exposure combo, of course.

However, it's not always something one can achieve, especially when working with high F/D ratios. My most recent example is a session I dedicated to small planetary nebulas. In this case, you want the highest focal the OTA can support, and the shortest possible exposures (to "freeze" the atmospheric turbulence to some extent) while keeping the highest possible dynamic range (because those objects are highly dynamic in nature), all of which yield a narrow field / low exposure / mid gain combination, in other words a recipe for disaster when it comes to star detection.

Of course, my exposures were nowhere short like what the lucky imaging guys do, more like a couple seconds. But still, Sharpcap had issues detecting stars. Being aware of the available settings, I played with both the Digital Gain, min/max HFD, noise reduction, black level, all to no avail, until I remembered to manually clear the stack. Then, magically, Sharpcap detected a few stars, stacked some frames, then stopped again. It's not the first time I've seen that happening, and other Sharpcap Pro users routinely clear the stack after each and every parameter modification, for it won't always detect stars if it had a bad start. I ended up with a working parameter combo, but I still don't know how I did it.

Two enhancements would be possible in my humble opinion:
1- ensure the first frame can be processed through star detection again when parameters changed. I suspect it only tries once, then stops when it can't find stars, and never resumes. I can understand why it doesn't resume in general cases (you don't want the stack star detection to be ruined afterwards).

2- one has to set those alignment parameters more or less blindly, save for the "highlight stars" feature, whose only use is to tell you star detection DOES works. What would be interesting is a way to display the result of the processed frame when it DOES NOT work as intended... Perhaps some crude image preview, post Alignment tab settings. In this way, it would probably be far easier to set all these parameters.

Thanks!

[admin]

Hi,

I suspect that your comment 1 is correct – I hadn't really thought about it that way before, but the first frame goes into the stack without needing alignment and if that is a bad frame in some way then you may never be able to get the stacking to work. However, SharpCap does re-detect stars in the existing stack if you change one of the star detection parameters like the digital gain or black level,in theory that could let you get going even if the initials detection parameters were wrong.

I do have a goal to make more of the star detection parameters automatic in the future – I think that in particular the black level and digital gain could be calculated automatically without too much trouble, leaving less parameters for people to have to adjust. I guess your request in part two is to be able to see what the frame would look like with the black level and digital gain parameters applied to give you a better way of choosing those correctly.

Cheers, Robin

[clouzot]

Hi Robin,

I guess your request in part two is to be able to see what the frame would look like with the black level and digital gain parameters applied to give you a better way of choosing those correctly.

Exactly. Being able to see what the star detector "sees" would be a huge improvement. There are multiple examples where one can perfectly see stars in a (streched) subexposure, while the star detection returns 0 detection. If you are able to automatically detect what the black level and gain should be, that would help a lot. I guess even the min/max FWHM could be detected thence.

[BlackWikkett]

I do have a goal to make more of the star detection parameters automatic in the future – I think that in particular the black level and digital gain could be calculated automatically without too much trouble, leaving less parameters for people to have to adjust.

Hi all,

I mostly use Live Stacking and have always been able to adjust settings to find parameters that will allow SC to find enough stars to enable alignment and stacking. This is one aspect of Live Stacking that I've found works quite well and would ask not to have user adjustable settings removed for some "automatic mode." If this type of feature is implemented could this be done as a toggle and the effects of the toggling auto on be reflected in the manual settings? If there's one thing I've learned about folks in the astronomy hobby they want more control not less.

Thanks, Chris

[clouzot]

Hi all,

I mostly use Live Stacking and have always been able to adjust settings to find parameters that will allow SC to find enough stars to enable alignment and stacking. This is one aspect of Live Stacking that I've found works quite well and would ask not to have user adjustable settings removed for some "automatic mode." If this type of feature is implemented could this be done as a toggle and the effects of the toggling auto on be reflected in the manual settings? If there's one thing I've learned about folks in the astronomy hobby they want more control not less.

Hence the request to be able to "see" what the various settings do. If not, you're essentially controlling something blindly.

It's good to hear that you don't have difficulties finding the correct parameters, but that's also because you're an experienced astronomer and I can assure you it's not the case for lots of beginners. We (small EAA forum) are overwhelmed with messages like "I started the livestack and saw no gradual enhancement, what is it all about", or, from the more attentive pupils, "It doesn't want to stack because it says it can't align". A comprehensive investigation follows in order to find what is really happening, until we reach the Alignment tab, for which the only advice we can give is "try, one parameter at a time, and report here". All parameters have a sensible meaning, but you can't really tell what's happening under the hood.

[orionic]

I am excited to see this conversation has begun, as I have wished for the same things Clouzot has proposed. Alignment works great for me most of the time, but when it occasionally stops working, the reasons are unclear and it can be very difficult sometimes to get it working. Also, quite often although I tell it to use 15 or 20 stars, and I can visually see dozens on the display, SharpCap only finds around 7-8 at most.

Here are some specific suggestions I would like to add:

1. Provide an option to "view pixel values used for alignment". What this would do is apply the settings for noise reduction, black level threshold, and digital gain, and then display at each pixel on the display, a brightness that indicates the value the alignment processor is using to detect stars. Presumably, each pixel would then be black if it is below black level threshold, and white, if it is above (I'm assuming the processor doesn't care about gray vs. white).

2. Also provide two sub-options for #1, a choice between "show first frame" and "show current frame". So we could toggle easily between the live view, the alignment pixel values view for the first frame, and the alignment pixel values view for the current frame to deduce how to get more stars detected.

3. Even before implementing any new features, it would be helpful to know in what sequence the different parameters are applied. I'm guessing black level threshold comes last, but I don't know whether noise reduction, or digital gain, is applied first. (Also, I'm not sure if the values used for the Gaussian blur in "noise reduction" follow the same scale used by Photoshop, i.e. is it a standard formula?)

(4. Much less important but just a thought. For min and max star width, it's pretty easy to imagine how those would work so I don't think there's a huge need to give the user feedback for these two parameters. However I suppose it would be helpful to have an option to display a red number on each star, indicating what width SharpCap has decided that the star has. Might be too processor-intensive though.)

Thank you for reading and considering our suggestions!

[admin]

Hi,

I can certainly answer the questions regarding the ordering of the image processing actions:

1) Gaussian Blur - the strength parameter is used as the sigma for the gaussian kernel in each direction. Not sure what photoshop etc treat the value as.
2) Digital gain - a simple multiplication with clipping at the white level
3) Threshold applied - all pixels below threshold% considered black
4) Feature detection
5) Filter by area - number of pixels in a star must fall between min star size squared and max star size squared.

There is also a final stage which checks the star images and discards any that look like they contain two or more stars rather than a single star.

I've taken some notes on the ideas that have been developing in this thread for future improvements to star detection.

Cheers, Robin

[donstim]

Hi,

I think there are some good ideas here, and I share Chris' concern about retaining the ability to tweak individual parameters if any automated settings do not provide the expected or desired results.

One item I would like to bring up is that alignment for stacking can be adversely affected by poor mount tracking, either in the first frame or subsequent frames, where the stars are "smeared" or become star trails. Especially for the subsequent frames case, if there is no option to view each frame before the stacking alignment is attempted (or that option is not used), it may perplex some people when SharpCap says not enough alignment stars when they can see plenty of stars in that first frame and SharpCap had accepted it.

Clouds or other obstructions suddenly entering into the frame can also affect stacking unbeknownst to the user staring at a screen and wondering why all of a sudden there are too few stars to align on.

Anything that provides more visibility to the user as to why alignment for stacking cannot be found would be helpful.

Thanks,
Don

[GSBass]

Thought I would post on this thread instead of trying to ask a new question, I attempted live stacking for first time last night but it was a fail, I just found a star field that gave me half a dozen stars that would fit on my 462 with a very long FL scope, I adjusted parameters so I could clearly see the stars on screen but sharpcap could not see any of them. As a newbie it just seems to me if I could see them the program should, kind of made me wish I could just manually select three of them so I could continue learning the correct exposure for overall image. I’m dead in the water til the next clear night to try to figure what I did wrong..... I was doing all the basics, resetting live stacking, making sure my alignment show stars tab was selected, even overexposing to ensure the program could see them. If everything really needs to be within perfect parameters to get it aligning then it’s going to be very frustrating to learn everything else, Would it even be possible to force the alignment by drawing/telling the program which stars you want to use in your preview window

[admin]

Hi,

the best thing to do in a similar situation is to save a couple of frames using the snapshot button (having turned off live stacking). If you then share those images here on the forums someone can test them out and advise on the best alignment settings or what other changes to make. Without seeing the images and the stars in them it's difficult to provide good advice.

thanks,

Robin