Eliminate subs with satellite trails?

[Gfischer]

Not sure pulled pic off internet. Looks like my version and I’m running latest

[admin]

Hi,

'Unstack' has been in SharpCap since version 4.1.12338, which was released in July 2024. If you are not seeing it then you must have an older version than that. Check out the downloads page for up-to-date versions : https://www.sharpcap.co.uk/sharpcap/downloads

cheers,

Robin

[Jean-Francois]

Hello,

Strange .. you have 17 images stacked for a total of 0.00 s ... ??
Are you taking sub-millisecond exposures ?

Jean-Francois

[abillionnewstars]

I have a super fast camera and computer

I was indeed on an older version of SC.

What are the odds of SC identifying Pluto while live stacking?

[carlomuccini]

Hi

Just last night I was using version 4.2 beta livestacking with 15-second exposures saved every 15 minutes.
Sigma clipping was activated starting at 5, so after 15 x 5 = 65 seconds or so, and at minute 10, well after the 5 acquisition frames, a pretty large satellite passed right in the middle; unfortunately, I didn't have time to press the Unstack button.
At that point, what should I do? The exposure had been going on for a while...
I continued the acquisition and hope that by adding the various 15-minute frames, the trail will fade...
It's been like this all the time and it's getting significantly worse as time goes by.

60frames_SV220_900s_2026-07-14T22_25_04_WithDisplayStretch.jpg (333.1 KiB) Viewed 731 times

Carlo

[admin]

Hi,

try looking at the sigma clipping settings in the 'Stacking' tab - adjusting those can make the sigma clipping more sensitive and help it reject trails. The default settings are pretty conservative - in other words they only reject quite significantly different data.

I'd suggest turning up the Sigma Threshold from 3 to 5 and turning down the Sigma Low Limit to very close to zero (or to zero). Try stacking some saved data via the folder monitor camera with those settings and watch the number of pixels being rejected in each frame (shown in the sigma clipping status below the sliders). In a 'non satellite' frame, you want the rejected pixel number to be relatively small - certainly less than 1% of the total pixel count. If you are seeing like hundreds of thousands or millions of pixels rejected each frame then you will have to turn up the sigma low limit a bit (normally the pixel is accepted if the value deviation from the mean for that pixel is less that <sigma threshold> * <calculated standard deviation at that pixel>, but the sigma low limit puts a hard lower limit on how small the acceptance window can be).

I will have a look at changing the defaults for those values - it's a long time since I added those features and cameras now are very different to how they were back then - it feels like the defaults that worked well a decade ago aren't right any more...

cheers,

Robin

[carlomuccini]

I've never changed the Sigma Clipping values ​​(I don't think so).
What should I set them to?

Now are
Sigma Threshold = 3
SigmaLowLimit = 5.07

SigmaClipping.jpg (30.99 KiB) Viewed 687 times

Carlo

[admin]

Hi,

I would set the low limit down towards minimum and the threshold to maximum as a starting point - then re-stack some old data using the folder monitor camera and keep an eye on the 'Status' to check that the number of rejected pixels is not rising too high (about 0.5% of the total frame at most).

Your post made me spend a day or so re-visiting how sigma clipping worked, since it is an old feature that has been untouched for some time. I think I have a better way of specifying the 'low limit' in terms of ADU rather than percent, and I have increased the threshold range to a max of 10. But, those are small fry compared to another change...

Because SharpCap does sigma clipping live on each new frame as the camera captures it, you can't use the traditional approach of working out the mean/standard deviation of all data then throwing out values that are too many sigma away from the mean for that pixel. Each frame has to be dealt with as it arrives, without storing them in bulk, so SharpCap has an 'establishment' phase of the first 10 or so frames where it works out some rough statistics without applying any sigma clipping, then enters an 'active' phase where the statistics are slowly refined and also sigma clipping is applied to new frames.

The problem with that approach is that if you get trouble during the establishment phase then it ends up in the data and is not cleaned up. If the establishment phase is clean then you are good to go, but nothing can be done about satellite trails in the establishment phase except to watch carefully and use the 'unstack' button if needed.

I've now worked out a fix for that problem... During the establishment phase SharpCap now keeps track of the minimum and maximum values receive for each pixel in the image. Once the first 10 (or however many) frames of the establishment phase are over, SharpCap goes back and checks for each pixel in the image to see if those maximum or minimum value pixels should be eliminated due to the sigma threshold set, and removes them if necessary.

That means that from the next update, if you get a satellite trail in the first 10 frames, there's no need to unstack - just let the stack continue and at the end of the establishment phase, the trail should vanish if the sigma clipping settings are set appropriately. Of course, if there are two frames tha both have unusual pixel values *at the same pixel* then only one will be cleaned up, but that will be much less of a problem - if two satellite trails cross then that might happen at a small handful of pixels, but not in a long line...

cheers,

Robin

[carlomuccini]

Hi

Last night I spent the evening testing SigmaClipping with version 4.1.
I set the values ​​as you suggested, and I must say it worked really well.
Threshold at maximum and Low limit at minimum.

I've attached an explanatory screenshot.
It's a live stack, saved every 15m, composed of 15-second frames.
At frame 20, a streak of a very fast satellite appeared, passing diagonally under the brightest star.
I didn't touch anything and let SigmaClipping do its thing.
At frame 25, the trail was barely visible, and by frame 29, it had already disappeared.
A nice result !

Carlo

SigmaClipping.jpg (264.66 KiB) Viewed 632 times

[Jean-Francois]

Hello Carlo,

You have some misunderstanding of the sigma-clipping process ...
It should eliminate the trace completely as soon as the actual frame is processed.

If you see a fading out of the trace ... it is simply the stacking effect : more signal is adding together and the trace contribution is going down.
If the trace(s) appear during the initial 5 images, they will stay. But if you see it, you can "Unstack" the frame with the trace(s) (but only the last frame and only before a new frame is processed).

Do you saved all the raw frames ? If yes, you can redo the process with the "camera folder" and so test different settings.
Play with the settings until the frame with the trace(s) does not appear on the stack image.
But it seems that is "no" ... you select "Save None"

Do you see that Robin will update the sigma-clipping in the beta 4.2 ?

Best regards,
Jean-Francois