Deep Sky Stacker - is this a stacking problem

[Steve143]

I'm still very new to astrophotography but hope someone can help with the image below. What is causing the light band across the top and right hand side of the stacked image? The image was taken with a Canon 70D using a 50mm lens at f2.8 and ISO 1600, untracked, and with a full set of calibration frames. I've noticed this banding on a couple of stacked images now, but not all of them. What am I doing wrong? Any help or suggestions would be most welcomed.

Thanks!  

[MercianDabbler]

It's worth stretching the contrast of some of your individual exposures to figure out where the issue is coming from - lights, darks, flats etc. Most 'regular' image manipulation tools will do it but best to use one that will work on raw files. Darktable for example will let you do non destructive editing on your individual raws... or perhaps use the Canon raw editing software.

If you use Darktable then you will probably need to clone whichever stretching module that you use (levels, curves etc), maybe using 2 or 3 copies to achieve enough stretch. Having done one file you can then copy/paste the edits (aka 'history stack') to other images to apply the exact same changes.

[alacant]

I'm guessing that the banding is visible only on staked images, whilst individual frames seem fine.

It looks like a combination of polar alignment and/or drift error between frames has caused the target to move out of view. The intersection of -in this case 100- frames therefore coincide less and less as the session proceeds. The bands are the non-coinciding parts of each frame.

HTH

 

Edited August 12, 2021 by alacant

[The Lazy Astronomer]

My initial thought is that it is a stacking artifact caused by the fact that your camera is not tracking the sky.

The sidereal rate (the speed at which stars appear to move through the sky) is 15 arcseconds per second. Your camera has 4.1um pixels, which means your imaging scale is 16.9 arcseconds per pixel ("/px) (image scale = pixel size in um / focal length in mm * 206.265), so over the course of your 10 minute total integration time, all the stars will have moved by ~530px in total, which is about 10% of your sensor width or about 15% of your sensor height

Eyeballing it, l would say that the lighter area looks to be covering about 10% of the width of the frame. The reason you also see a lighter area along the top is likely because the camera wasn't angled so as to be aligned with the Earth's equitorial plane (so the stars moved diagonally across the frame).

[MercianDabbler]

18 minutes ago, alacant said:

It looks like a combination of polar alignment and drift error between frames has caused the target to move out of view. The intersection of frames therefore coincide less and less as the session proceeds. The bands are the non-coinciding parts of each frame.

Could be... in which case try 'intersection mode' in Stacking Settings/Result - then it will only stack the area that is covered by all of your frames.

Checking the individual frames would also confirm whether this is happening... you can do that in DSS by clicking on your light frames one by one in the files list.

[Steve143]

15 minutes ago, alacant said:

I'm guessing that the banding is visible only on staked images, whilst individual frames seem fine.

It looks like a combination of polar alignment and/or drift error between frames has caused the target to move out of view. The intersection of -in this case 100- frames therefore coincide less and less as the session proceeds. The bands are the non-coinciding parts of each frame.

HTH

 

Thanks. The image was taken with just the camera and a tripod - no tracking mount. The total exposure time was only 10 minutes but I could try and stack fewer images to see if that removes the banding by removing some of the non-coinciding frames.

[Steve143]

17 minutes ago, The Lazy Astronomer said:

My initial thought is that it is a stacking artifact caused by the fact that your camera is not tracking the sky.

The sidereal rate (the speed at which stars appear to move through the sky) is 15 arcseconds per second. Your camera has 4.1um pixels, which means your imaging scale is 16.9 arcseconds per pixel ("/px) (image scale = pixel size in um / focal length in mm * 206.265), so over the course of your 10 minute total integration time, all the stars will have moved by ~530px in total, which is about 10% of your sensor width or about 15% of your sensor height

Eyeballing it, l would say that the lighter area looks to be covering about 10% of the width of the frame. The reason you also see a lighter area along the top is likely because the camera wasn't angled so as to be aligned with the Earth's equitorial plane (so the stars moved diagonally across the frame).

Thanks. For this image I just mounted the camera on a tripod, pointed it towards Deneb and took the frames. As a total beginner I'm just doing this to learn how to use the camera, take calibration frames and stack the images. To learn post-processing I'm using GIMP but it's a steep learning curve as I've never done this before but there are some great videos on Youtube and had read some useful articles on SGL. Once I feel a bit more confident I think I need to invest in a tracking mount to improve my images. 

[Steve143]

16 minutes ago, MercianDabbler said:

Could be... in which case try 'intersection mode' in Stacking Settings/Result - then it will only stack the area that is covered by all of your frames.

Thanks. I will try that! 

[Chris-h]

Is Dark Optimization checked? If it is, turn it off. It can also cause weird things like this.

[Steve143]

23 hours ago, MercianDabbler said:

Could be... in which case try 'intersection mode' in Stacking Settings/Result - then it will only stack the area that is covered by all of your frames.

Checking the individual frames would also confirm whether this is happening... you can do that in DSS by clicking on your light frames one by one in the files list.

I restacked my frames, which looked fine, using intersection mode and it worked perfectly! The banding was gone and I think the rest of image looks a lot better too! Thank you! 

I'm learning all the time! I just need to get the hang of processing the stacked image. I'm currently using GIMP to practice but thinking about getting Affinity Photo.  

[MercianDabbler]

2 hours ago, Steve143 said:

I restacked my frames, which looked fine, using intersection mode and it worked perfectly! The banding was gone and I think the rest of image looks a lot better too! Thank you! 

I'm learning all the time! I just need to get the hang of processing the stacked image. I'm currently using GIMP to practice but thinking about getting Affinity Photo.  

Glad that you got to the bottom of the problem. Looks like @alacant was right.

I'm using all freebie tools myself. I very rarely pay money for software... kinda ironic since I get paid to work on the stuff. There are plenty of folks who say good things about the paid tools that specialise in AP postprocessing though.

Better tools should do better at cleaning up AP specific problems like light pollution gradients but no tool can put back data that is not there so learning how to get good data and spending the time to get plenty of it is the most important thing IMHO. I can't claim to be anything other than a learner myself.

[happy-kat]

When I use a static mount where I have around 100 images I divide it into three batches. I pick a good frame from the middle of the 100 images and use it in each stack but marked as the reference frame only in the two other stacks. Once processed I stitch them using ICE (Microsoft seem to have withdrawn it now). 

Edited August 13, 2021 by happy-kat