Stacking, Flats, and Tree Limbs - Forums

Topic starter

Mar 23, 2021

Hi,

Question for the community on whether flats can successfully remove tree obstructions from light frames.

I recently planned an imaging session and believed that my entire session would stay above the tree in my yard, but I was wrong. About two thirds of the way through the session I started to see subtle, dark, finger-like shadows on my light frames that was caused by the tree in my yard partially obstructing my line-of-sight to the target. Since there aren't leaves on the trees yet, the shadows were relatively narrow and marched across my frames as I tracked the target. There was no wind so the trees limbs were not in motion.

The next morning I captured 20 flat frames. Does anyone know if DeepSkyStacker can use the flat frames to subtract out the shadows (fooling the program into thinking they are just dust and dirt on the sensor)? I would try the experiment myself, except I figured the data was no good and deleted the affected light frames.

Well Written Engaging

Björn Arnold

Mar 23, 2021 · Edited: Mar 23, 2021

Hi,

I think it’s not working:
Flats remove static dirt in the optical train. Your tree is moving (w.r.t. sky, telescope etc.). (Besides that usually only the dirt in the optical train is considered and not the stuff outside)

However, depending on size of the tree and the exposure times of the lights, I could imagine that the sigma clipping of the stacking application could catch it. Think of the tree as satellite tracks but just opposite intensity. You might need to play (likely reduce) with the sigma. Just a guess. Didn’t have that case myself.

CS!

Björn

Helpful Insightful Respectful

Norman Hey

Mar 31, 2021

Ha! Didn't think of trying as Bjorn suggests–the concept seems sound. I have large maples that cut off my southwestern sky completely when in leaf, but in the winter, like golfers say, "trees are mostly air"… Next clear night, I will give it a shot and report. The key will be having enough unobstructed frames to be the base to which the obstructed frames are compared…

Engaging

Jérémie

Mar 31, 2021 · Edited: Mar 31, 2021

Norman Hey:
Ha! Didn't think of trying as Bjorn suggests--the concept seems sound. I have large maples that cut off my southwestern sky completely when in leaf, but in the winter, like golfers say, "trees are mostly air"... Next clear night, I will give it a shot and report. The key will be having enough unobstructed frames to be the base to which the obstructed frames are compared...

Rejection algorithm could work only if the tree is present on a part of the image only for a very a few frames, - which is unlikely as the motion of earth rotation is low compared to satellites for example - otherwise, its « distribution » among the frames will be closer to the average +/- a few sigma... and it won’t be rejected. You would need to to set rejection aggressively... which will likely reintroduce noise in the final master.
I have had the problem with buildings and antennas, imaging from the city : rejection algo do not work well with static objects present in many frames...
At some point, I was even thinking of aligning my frames, and for the ones that were obstructed, use the data of an unobstructed one - or from an integrated one based on all of the unobstructed ones, to have a better SNR - just for the obstructed part (easy to create a mask by hand or playing through levels in Photoshop) and use Pixmath in Pixinsight to composite these frames. And reintegrate the bunch of frames (unobstructed + « frankensteined » frames).
Of course it would introduce bias and kill the nice effect of increasing the SNR on the part that were « Frankensteined », but would still be better than a few minutes of antennas, trees etc... in the data.
Didn’t have time yet to try though :-)

Helpful Insightful Respectful Engaging

andrea tasselli

Frequent contributor

Mar 31, 2021

Happened to me few times in the past. I had to chuck the affected frames.

Nick Ambrose

Mar 31, 2021

If you use Pixinsight, this might work for you. You'll get lower signal in those areas you mask off so it will be a tradeoff

https://www.youtube.com/watch?v=ZZVJKo47Ctk

AstroCapture325

Topic starter

Mar 31, 2021

Hey everyone,

Thanks for the additional insight/comments.

Norman Hey

Apr 1, 2021

Nick Ambrose:
If you use Pixinsight, this might work for you. You'll get lower signal in those areas you mask off so it will be a tradeoff

https://www.youtube.com/watch?v=ZZVJKo47Ct

I had that in mind as a fall back try--neat trick that looks like it might have a few applications.