Milky Way over the Olympics using Pixinsight for calib + stacking

[lord_of_light]

I had a chance to visit Olympic National Park last month and took some Milky Way shots from near the Hurricane Ridge Visitor Center. When I got up there at midnight there was some type of astronomy event going on with 100+ cars and lots of people at the visitor center, so I found a quieter place nearby and took some shots.

The equipment was Sony A7iii with a Sony 20mm f/1.8 lens shot at f/2 with a Star Adventurer tracker. This lens has more coma in the corners than I would like which has me interested in the new Sigma 20mm and 24mm lens, but otherwise I'm reasonably happy with this as a travel setup. The shot consisted of 20 x 1 min exposures for both the sky and foreground. My process is I set up the tracker, compose the shot, shoot the foreground with the tracker off and then turn it on to shoot the sky. I'm reasonably happy with the shot, though it's certainly not the best I've seen. The thing about it that interests me the most was the preprocessing and stacking which I did with Pixinsight.

In the past I've stacked my tracked sky exposures with Sequator and then used Photoshop to stack my foreground and merge the sky foreground together. I started that way this time, but I wasn't fully satisfied with the result. I had several problems. First off 20 mins of total exposure results in a large blurry gap between sky and foreground that needs to be filled in in some way. Normally I've use a short untracked sky shot (typically 10-30 sec exposure) for this, but balancing the color and exposure was difficult in this case and the short exposure adds extra noise to the image. In addition the Sony camera I'm using has lots of hot pixels which didn't get removed by Sequator and while I can do it in Photoshop, the process doesn't work as well as Cosmetic Correction in Pixinsight.

I didn't find alot of information online about using Pixinsight for wide angle MW shots but I did read that its stacking algorithm could handle the distortion from a wide angle lens which is what made me want to give it a try. I took a set of dark frames (no flats) and went through the calibration process with Cosmetic Correction enabled. The result was elimination of almost all the hot pixels w/o blurring the image. The foreground images were already well aligned so I was able to simply stack those using Image Integration to produce the final foreground shot. Next the calibrated sky images were Star aligned with Distortion Correction enabled and then Integrated to produce a sky stack. This worked pretty well, but there was still the large blurry gap between foreground and background so I used the GAME script to mask out the foreground from each of the sky shots and then integrate the masked images together. This allows virtually all the sky portions from each shot to be used w/o averaging in any of the masked out foreground portions from other shots. The result is a much smaller gap between sky and background with lower noise in the region. Finally, the combination of foreground and sky, plus contrast enhancement, etc... was done in Photoshop.

I think the Pixinsight method resulted in a better image than what I got from my first attempt using Sequator. That said it took alot more effort. I'm looking for a way to simplify the process. Also I had wanted to more processing in Pixinsight (i.e. use DBE to remove some of the gradients) but I found that it made the final combination of sky and background more difficult so I set that aside for now.

Milky Way over the Olympics

 

[Alen K]

In the past I've stacked my tracked sky exposures with Sequator and then used Photoshop to stack my foreground and merge the sky foreground together. I started that way this time, but I wasn't fully satisfied with the result. I had several problems. First off 20 mins of total exposure results in a large blurry gap between sky and foreground that needs to be filled in in some way. Normally I've use a short untracked sky shot (typically 10-30 sec exposure) for this, but balancing the color and exposure was difficult in this case and the short exposure adds extra noise to the image. In addition the Sony camera I'm using has lots of hot pixels which didn't get removed by Sequator and while I can do it in Photoshop, the process doesn't work as well as Cosmetic Correction in Pixinsight.

Nice image!

Re Sequator, did you use the “Select Best Pixels” “composition” (stacking) mode? That implements three-iteration kappa-sigma clipping. It has never failed to remove hot pixels for me. It also, of course, removes trails from satellites and airplanes but sometimes I have needed to increase the clipping threshold to do that.

Since you knew Sequator might have difficulty stacking images from a 20mm lens, I assume you selected the “Complex” option in “Reduce distor. effects.” The author claims it can handle the geometric distortion from a 20mm equivalent lens “acceptably.”

PS. There is also the “Reduce dynamic noises” option that tries to detect anomalous pixel values like hot pixels and, presumably, replaces them with localized average pixel values. This would normally be of value for single images or for very short stacks.

 

[nighthiker]

Hi!

My knowledge in PI is limited, so I still use Sequator for stacking (using the setting mentioned by Alen) and only do the DBE and stretching in PI. After DBE my sky image has a grey background (probably due to the terrestrial foreground) , but that can be easily corrected. The stretched image is often looking too clean, so I blend some of the non-DBE image into it to make it look more realistic.

To my knowledge one of the strong points of Sequator is the option to reduce distortion effects for UWA where many other programs (e.g. DSS) fail. I haven't tested that in PI, but the example is looking promising.

 

[lord_of_light]

In the past I've stacked my tracked sky exposures with Sequator and then used Photoshop to stack my foreground and merge the sky foreground together. I started that way this time, but I wasn't fully satisfied with the result. I had several problems. First off 20 mins of total exposure results in a large blurry gap between sky and foreground that needs to be filled in in some way. Normally I've use a short untracked sky shot (typically 10-30 sec exposure) for this, but balancing the color and exposure was difficult in this case and the short exposure adds extra noise to the image. In addition the Sony camera I'm using has lots of hot pixels which didn't get removed by Sequator and while I can do it in Photoshop, the process doesn't work as well as Cosmetic Correction in Pixinsight.

Nice image!

Re Sequator, did you use the “Select Best Pixels” “composition” (stacking) mode? That implements three-iteration kappa-sigma clipping. It has never failed to remove hot pixels for me. It also, of course, removes trails from satellites and airplanes but sometimes I have needed to increase the clipping threshold to do that.

Since you knew Sequator might have difficulty stacking images from a 20mm lens, I assume you selected the “Complex” option in “Reduce distor. effects.” The author claims it can handle the geometric distortion from a 20mm equivalent lens “acceptably.”

thank you for the comments. I've been using Sequator for awhile now and always use both the "Select Best Pixels" and "Complex" settings, and you are right those settings do a decent job removing hot pixels and handling the lens distortion. I should have been more clear that the problem with hot pixels was with the foregrounds shots which I don't normally process through Sequator. Instead I've been using Photoshop to stack the foreground and it's there that the hot pixels are so noticeable. By switching the foreground stacking over to Pixinsight I was able to use the Cosmetic Correction process to remove them.

 

[lord_of_light]

In the past I've stacked my tracked sky exposures with Sequator and then used Photoshop to stack my foreground and merge the sky foreground together. I started that way this time, but I wasn't fully satisfied with the result. I had several problems. First off 20 mins of total exposure results in a large blurry gap between sky and foreground that needs to be filled in in some way. Normally I've use a short untracked sky shot (typically 10-30 sec exposure) for this, but balancing the color and exposure was difficult in this case and the short exposure adds extra noise to the image. In addition the Sony camera I'm using has lots of hot pixels which didn't get removed by Sequator and while I can do it in Photoshop, the process doesn't work as well as Cosmetic Correction in Pixinsight.

Nice image!

Re Sequator, did you use the “Select Best Pixels” “composition” (stacking) mode? That implements three-iteration kappa-sigma clipping. It has never failed to remove hot pixels for me. It also, of course, removes trails from satellites and airplanes but sometimes I have needed to increase the clipping threshold to do that.

Since you knew Sequator might have difficulty stacking images from a 20mm lens, I assume you selected the “Complex” option in “Reduce distor. effects.” The author claims it can handle the geometric distortion from a 20mm equivalent lens “acceptably.”

PS. There is also the “Reduce dynamic noises” option that tries to detect anomalous pixel values like hot pixels and, presumably, replaces them with localized average pixel values. This would normally be of value for single images or for very short stacks.

My main reasons for trying Pixinsight was to: (i) remove the hot pixels from the foreground shots -- this works well using Cosmetic Correction before stacking the shots; (ii) see if Pixinsights star alignment process could handle the lens distortion from an UWA lens -- it seems to handle this at least as well as Sequator; (iii) find a better way to process the sky data at the boundary with the foreground to lessen the blurry gap that has to be filled in when I stitch sky and foreground together. With 20 minutes of data there is considerable sky rotation, so that gap was pretty big. However using the GAME script in Pixinsight and enabling the "Clip Low Range" option in Image Integraion, I was able to use all my sky data to fill in most of this gap, making the stitching process easier (and more pleasing imo). The downside of this was that using the GAME script was a bit tedious. Next time, I think it might be able to accomplish the same thing, with less work, if I just paint out the foreground from the sky shots, before stacking them.

Note: in Pixinsight, the "Clip Low Range" option in the stacking process will automatically ignore pixels that are set to black, so it provides a way to selectively discard parts of each image (i.e. by painting the foreground portion black in each image) so that they don't get integrated into the final stack.

 

[NatureFiles]

Beautiful shot and epic location.

Would you recommend the Star Adventurer for someone new to astrophotography?

 

[lord_of_light]

Beautiful shot and epic location.

Would you recommend the Star Adventurer for someone new to astrophotography?

certainly. I'm happy with it as a relatively low cost travel friendly star tracker. I have tested it with lens up to 300mm f/2.8 and while it works for that, I have since moved to a larger Ioptron mount for anything big and heavy. However for wide angle milky way shots, it works just fine. There is a new SA mount that is similar in many ways, but offers declination tracking and GOTO, and while I haven't used/tested that one, I might have been tempted if I was in the market.

 

[NatureFiles]

Beautiful shot and epic location.

Would you recommend the Star Adventurer for someone new to astrophotography?

certainly. I'm happy with it as a relatively low cost travel friendly star tracker. I have tested it with lens up to 300mm f/2.8 and while it works for that, I have since moved to a larger Ioptron mount for anything big and heavy. However for wide angle milky way shots, it works just fine. There is a new SA mount that is similar in many ways, but offers declination tracking and GOTO, and while I haven't used/tested that one, I might have been tempted if I was in the market.

Thanks for the feedback. I'll be checking those out.

 

[matthewu]

Thanks for this great writeup on stacking!

Would love to understand better how you think about planning a shot like this:
- Why 20 x 1min vs more shorter or fewer longer exposures?
- I haven't seen other examples of stacking the foreground. My normal goto for foreground is single long exposure + Lightroom or Topaz denoise. Curious if I should try stacking instead.

In terms of PixInsight processing:
- Integrate in Pixinsight sounds, from the name, like it might simply average the photos. Was wondering if there was more to it? eg statistical stuff like removing outliers.
- Does PixInsight have ability to take external mask instead of GAME step you mentioned? It feels like masking options in Photoshop are so good that I'm wondering if GAME adds anything?

 

[zurubi]

I had a chance to visit Olympic National Park last month and took some Milky Way shots from near the Hurricane Ridge Visitor Center. When I got up there at midnight there was some type of astronomy event going on with 100+ cars and lots of people at the visitor center, so I found a quieter place nearby and took some shots.

The equipment was Sony A7iii with a Sony 20mm f/1.8 lens shot at f/2 with a Star Adventurer tracker. This lens has more coma in the corners than I would like which has me interested in the new Sigma 20mm and 24mm lens, but otherwise I'm reasonably happy with this as a travel setup. The shot consisted of 20 x 1 min exposures for both the sky and foreground. My process is I set up the tracker, compose the shot, shoot the foreground with the tracker off and then turn it on to shoot the sky. I'm reasonably happy with the shot, though it's certainly not the best I've seen. The thing about it that interests me the most was the preprocessing and stacking which I did with Pixinsight.

In the past I've stacked my tracked sky exposures with Sequator and then used Photoshop to stack my foreground and merge the sky foreground together. I started that way this time, but I wasn't fully satisfied with the result. I had several problems. First off 20 mins of total exposure results in a large blurry gap between sky and foreground that needs to be filled in in some way. Normally I've use a short untracked sky shot (typically 10-30 sec exposure) for this, but balancing the color and exposure was difficult in this case and the short exposure adds extra noise to the image. In addition the Sony camera I'm using has lots of hot pixels which didn't get removed by Sequator and while I can do it in Photoshop, the process doesn't work as well as Cosmetic Correction in Pixinsight.

I didn't find alot of information online about using Pixinsight for wide angle MW shots but I did read that its stacking algorithm could handle the distortion from a wide angle lens which is what made me want to give it a try. I took a set of dark frames (no flats) and went through the calibration process with Cosmetic Correction enabled. The result was elimination of almost all the hot pixels w/o blurring the image. The foreground images were already well aligned so I was able to simply stack those using Image Integration to produce the final foreground shot. Next the calibrated sky images were Star aligned with Distortion Correction enabled and then Integrated to produce a sky stack. This worked pretty well, but there was still the large blurry gap between foreground and background so I used the GAME script to mask out the foreground from each of the sky shots and then integrate the masked images together. This allows virtually all the sky portions from each shot to be used w/o averaging in any of the masked out foreground portions from other shots. The result is a much smaller gap between sky and background with lower noise in the region. Finally, the combination of foreground and sky, plus contrast enhancement, etc... was done in Photoshop.

I think the Pixinsight method resulted in a better image than what I got from my first attempt using Sequator. That said it took alot more effort. I'm looking for a way to simplify the process. Also I had wanted to more processing in Pixinsight (i.e. use DBE to remove some of the gradients) but I found that it made the final combination of sky and background more difficult so I set that aside for now.

Please clarify something for me:

you said 20x1min for both foreground and background. So does this mean that after each tracked 1 min sky shot you turn the tracker off and shoot the foreground for 1 min? (and repeat)

 

[Alen K]

Please clarify something for me:

you said 20x1min for both foreground and background. So does this mean that after each tracked 1 min sky shot you turn the tracker off and shoot the foreground for 1 min? (and repeat)

That would be inefficient. The usual procedure is to take all untracked foreground exposures before or after the tracked sky exposures. I usually take them at the end of a complete session. If I am doing an all-nighter, that is then in the morning when the sky is just starting to lighten too much to take sky images. But I am not a fan of "blue hour" foregrounds, which would be taken a little later than that or in the evening.

 

[zurubi]

Please clarify something for me:

you said 20x1min for both foreground and background. So does this mean that after each tracked 1 min sky shot you turn the tracker off and shoot the foreground for 1 min? (and repeat)

That would be inefficient. The usual procedure is to take all untracked foreground exposures before or after the tracked sky exposures. I usually take them at the end of a complete session. If I am doing an all-nighter, that is then in the morning when the sky is just starting to lighten too much to take sky images. But I am not a fan of "blue hour" foregrounds, which would be taken a little later than that or in the evening.

I agree with you but I don't know what the OP meant, so let's see what he/she answers.....

I general, I find it painful to blend in PS a foreground with tracked MWs so I'm always looking for new ways to PP that reduce this pain.