[Nikkau]

It's bad faith, a normal stacking would have made them completely disappear, it's one of main reasons to do stacking, remove things which are not on all images, and it's works flawlessly.

You have to actively tweak your settings to create this kind of photo.

[vilhelm_s]

I guess the most basic way to stack (just add the images together) would leave them in, but

> Almost every modern astronomical post-processing program has a rejection process (sometimes referred to as sigma-reject) to remove unwanted signals, though the exact sequence will depend on which program you use.

https://skyandtelescope.org/astronomy-blogs/imaging-foundati...

[woko]

From your link:

> The way this process works is that, while averaging all of the pixels in a series of, say, 10 images, the program mathematically calculates which pixels fall far away from the mean value because they're much brighter (or much fainter) compared to the same pixels in other frames. The algorithm then discards those out-of-range pixel values so they don’t affect the final image.

Wouldn't this process remove part of the comet trails as well as the satellite trails?

I mean, I get how it works if all you care about is relatively static like distant stars, but would it work for this specific use case?

[spacemark]

The most common algorithm to manage airplanes, satellites, hot pixels, and other undesired photons in astrophotos is a process called Kappa-Sigma Clipping. It essentially rejects pixel values from subframes in your image stack that fall outside a user-inputted deviation from the mean.

In other words, the process works wonderfully to get rid of the starlink-emitted photons, but you lose that subframe's signal, lowering your signal to noise ratio. Not the end of the world. But inconvenient and sometimes costly to professional astronomers.

[CydeWeys]

Yeah, but what % of subframes (small portions of large images) are ruined by noise caused by moving objects? Way less than 1% I'd imagine. It's just not a big problem. And certainly not worth outlawing new satellite launches over.

[Sharlin]

No, because the comet does not perceptibly move in the sky during the acquisition. All comet pixels are present in every image.

[_Microft]

The comet does not change but its position in the sky does btw. Longer exposure times turn points into streaks if the object is not tracked to compensate for this.

[Sharlin]

Yes, I assumed tracking as it's basically a mandatory requirement when you do telephoto astrophotography, and definitely used in the OP photo. The alternative is to shoot wider angle and align the images during stacking, but either way you have to get your subject's pixels aligned or the result is just blur.

[CydeWeys]

This is true of every single object in the night sky, and is why an equatorial tracking mount is table stakes for good astrophotography.

[sixothree]

Take a look at this

https://photoshoptrainingchannel.com/remove-tourists-stack-m...

[ipsum2]

> You have to actively tweak your settings to create this kind of photo.

No, the most basic stacking is just to add the images together, resulting in trails. I don't see this as bad faith at all.

[trickstra]

If it created trails, then it would also multiply the brightness of the stars and the comet by the number of pictures (17 in this case). Each satellite appears only on one photo, while the stars and comets appear on all 17 in the exact same spot. The only way to stack them to get a normal looking stars and comet is to make an average. And since the satellites only appear on 1 out of 17 they would effectively disappear.

Unless you actually want to have the satellites there, then the stacking would just cut the region with the satellite from each photo and simply glue them together. That's how you can get this image.

I actually just tried it in Hugin. Normal stacking from 2 pictures by default made the objects that appeared only on one of the pictures semi-transparent. If I did this with more pictures, they would be so transparent that I wouldn't see them at all. But I could manually select a mask to cut out the portions that I don't want. If I select and exclude the object (satellite streak) it would disappear completely. But I could also purposefully include the streak and in that case all the satellite streaks would be included in full brightness in the result. That's most likely how this photo was made.