Help with stacking

[StuartT]

Last night I tried my first serious attempt at creating an image of a DSO (as it was very good seeing).

I selected M51 and using a Canon EOS750D at the prime focus of my SCT I shot;

46 lights, RAW, 30 sec at ISO6400

16 darks, RAW, 30 sec at ISO6400

16 bias, RAW, RAW, 1/4000 sec at ISO6400

Then I combined them in Deep Sky Stacker. It was set to a threshold which found between 100 and 250 stars and the stacking was with Kappa-Sigma clipping.

This is the result after stretching. A little underwhelming. Any tips?

(I didn't shoot flats as I can't figure out how to do that without an iPad!)

Edited April 4, 2021 by StuartT

[malc-c]

Hi Stuart,

I'm no expert, but 46 x 30sec is only 23 minutes worth of data which isn't a lot IMO.  Granted you have a high ISO, which can result in a more grainy image, but even when I was imaging at 800 ISO I would still gather at least an hour on bright galaxies such as M31 and M81.   M51 is quite a faint target by comparison.   Also that's with a 200P so like your SCT an 8" but at half the focal length as the 200P is f5.

Do you have the LX on a wedge or is it in an ALT/AZ format, and are you guiding ?

[vlaiv]

You've cropped that image considerably, right?

With DSLR camera and very long focal length scope - you are way oversampling at 0.38"/px.

In order to overcome limitations of that, you'll need to employ couple of techniques. First - use super pixel mode debayering. Second - you'll need to bin your data, maybe even 3x3.

Ideally, you need long exposures and lot of them. When oversampling like that - worst thing is to go for 30s exposures as read noise becomes very dominant (signal is too faint per exposure due to very high "magnification").

Could you post stacked linear data? I think quite a bit more can be "squeezed" out of it by doing the above tricks.

[StuartT]

4 hours ago, malc-c said:

Hi Stuart,

I'm no expert, but 46 x 30sec is only 23 minutes worth of data which isn't a lot IMO.  Granted you have a high ISO, which can result in a more grainy image, but even when I was imaging at 800 ISO I would still gather at least an hour on bright galaxies such as M31 and M81.   M51 is quite a faint target by comparison.   Also that's with a 200P so like your SCT an 8" but at half the focal length as the 200P is f5.

Do you have the LX on a wedge or is it in an ALT/AZ format, and are you guiding ?

Thanks. I wasn't sure how many frames to shoot, so this is helpful. I realise a f/10 isn't great for deep sky, but it's what I have 😉

Afraid I don't have the LX90 on a wedge, just the regular old altaz fork. No guiding (I'm not clever enough). That's why I stuck to relatively short exposures.

2 hours ago, vlaiv said:

You've cropped that image considerably, right?

With DSLR camera and very long focal length scope - you are way oversampling at 0.38"/px.

In order to overcome limitations of that, you'll need to employ couple of techniques. First - use super pixel mode debayering. Second - you'll need to bin your data, maybe even 3x3.

Ideally, you need long exposures and lot of them. When oversampling like that - worst thing is to go for 30s exposures as read noise becomes very dominant (signal is too faint per exposure due to very high "magnification").

Could you post stacked linear data? I think quite a bit more can be "squeezed" out of it by doing the above tricks.

Umm.. I'll have to look all that stuff up! 🤣 

Yes, the image is heavily cropped.

By oversampling, do you mean I am shooting at too high a resolution? I guess with my current setup, I can't really go any longer that 30 sec exposures.

What do you want me to post? The light frames? Or the text files that DSS generates?

[vlaiv]

21 minutes ago, StuartT said:

By oversampling, do you mean I am shooting at too high a resolution? I guess with my current setup, I can't really go any longer that 30 sec exposures.

Yes - to high resolution in arc seconds per pixel - or too zoomed in if you will. It is a bit like using too high magnification for scope and conditions. You get zoomed in fuzzy image that gets darker as you add magnification.

23 minutes ago, StuartT said:

What do you want me to post? The light frames? Or the text files that DSS generates?

After you finish stacking, you can export 32bit fits file.

I think that if you post that file (it might be a bit large) - people can help with processing steps to show you how you can get the most out of your data.

[StuartT]

Ok it gave me two options to save the 32 bit FITS file (integer and rational). So I exported both

integer.FTS

rational.FTS

Thanks again

[malc-c]

Like I mentioned I'm no expert, and a lot of what Vlaiv mentioned went over my head, but from what I do know is that I think you are faced with a couple of fundamental issues.  There is a relationship between focal length and exposures.  It's probably no linear but at a guess for a given ISO and same aperture you need to expose twice as long for a scope of twice the focal ratio.  But the problem with long exposures is that you need to guide (unless you have such a high end mount that has fantastic tracking and minimal PEC.  Now an ALT AZ mount has two issues where imaging is concerned which compound the problem.  The first is that guiding becomes difficult as the drift tends to be on two axis rather than one when comparing an ALT/AZ  mount to an EQ, which is related to the second issue of field rotation.  If left unguided a long exposure will result in field rotation, which is approx 45 degrees in a three hour period.  This can then lead to stacking issues because each sub is rotated a few degrees.

So I guess to improve things you either need to consider a focal reducer or fit some form or wedge so the scope can be used in an EQ format and then possibly consider guiding so that more data can be gathered.

[vlaiv]

I'm not sure I'll be able to do much more than original processing with this data.

I managed to get this as luminance:

Tried to apply synthetic flats, but I don't think I quite managed to do it - three is zone below the galaxy that is till brighter than the rest of the background.

Background is now nice and even - but I don't think I managed significantly more signal in the target itself.

[StuartT]

12 hours ago, malc-c said:

Like I mentioned I'm no expert, and a lot of what Vlaiv mentioned went over my head, but from what I do know is that I think you are faced with a couple of fundamental issues.  There is a relationship between focal length and exposures.  It's probably no linear but at a guess for a given ISO and same aperture you need to expose twice as long for a scope of twice the focal ratio.  But the problem with long exposures is that you need to guide (unless you have such a high end mount that has fantastic tracking and minimal PEC.  Now an ALT AZ mount has two issues where imaging is concerned which compound the problem.  The first is that guiding becomes difficult as the drift tends to be on two axis rather than one when comparing an ALT/AZ  mount to an EQ, which is related to the second issue of field rotation.  If left unguided a long exposure will result in field rotation, which is approx 45 degrees in a three hour period.  This can then lead to stacking issues because each sub is rotated a few degrees.

So I guess to improve things you either need to consider a focal reducer or fit some form or wedge so the scope can be used in an EQ format and then possibly consider guiding so that more data can be gathered.

Hi. You're sure more expert than me! Yes, I am aware of the dual problems of long focal length and AltAz mount. I guess I just wanted to see what might be possible, even within those limits. Hence why I restricted myself to 30s exposures. So far I have been mostly looking at the sun and the moon (where those issues aren't really a problem). But if I decide I really want DSOs I shall buy some new toys.

Question though - since Deep Sky Stacker is able to deal with translation (during registration) why would field rotation be a problem for it? Can't it also rotate images as necessary?

10 hours ago, vlaiv said:

I'm not sure I'll be able to do much more than original processing with this data.

I managed to get this as luminance:

Tried to apply synthetic flats, but I don't think I quite managed to do it - three is zone below the galaxy that is till brighter than the rest of the background.

Background is now nice and even - but I don't think I managed significantly more signal in the target itself.

ooh! Definitely an improvement! Thanks so much for taking the time to do this 😊

Edited April 5, 2021 by StuartT

[The Admiral]

Stuart, if you are doing Alt-Az imaging might I suggest that you have a look at the (rather long) thread dedicated to such imaging (if you haven't already)?

There you'll get a feel for what to expect and what can be achieved. Whatever scope or focal length you use, you will be limited to short exposures in order to avoid field rotation, so anything you can do to maximize the number of photons captured and signal to noise ratio the better. A lower f-ratio could help; you could think about a reducer for your SCT which will bring it down to f/6.3. The other thing that you'll need to do is to capture a lot of subs for stacking, and I mean, a lot. When I imaged M51 with an f/7 102m refractor, I used 200 x 30s subs , and even then the results were not exactly overwhelming (https://stargazerslounge.com/gallery/image/30726-m51/). Also, as Vlaiv pointed out, with the focal length you are using, light from an object is spreading over too many pixels so that the noise from each pixel tends to overwhelm the signal. By 'binning' data it means treating several pixels together as a block so as to improve the signal to noise ratio. With your DSLR you can only do this is software.

Ian

 

Edited April 5, 2021 by The Admiral

[StuartT]

18 minutes ago, The Admiral said:

Stuart, if you are doing Alt-Az imaging might I suggest that you have a look at the (rather long) thread dedicated to such imaging (if you haven't already)?

 

Thanks for this. I'll check that thread out

[malc-c]

11 hours ago, StuartT said:

 

Question though - since Deep Sky Stacker is able to deal with translation (during registration) why would field rotation be a problem for it? Can't it also rotate images as necessary?

 

Basically this was what I was trying to cover.  From DSS website

Star detection
For each picture DeepSkyStacker will attempt to automatically detect the stars.  In simple terms, DeepSkyStacker considers that a star is a round object whose luminance decreases regularly is every direction, and whose radius is no more than 50 pixels.   Note that DeepSkyStacker will reject elongated star images which might occur if your mount isn't tracking correctly.   

I've underlingned the  point I feel might be an issue.  If the exposures are long, without tracking there could be enough elongation of the stars for DSS to reject the image and discard the image.  

In the past I have managed to stack images where the camera had been removed and replaced resulting in some subs being 90 degrees to others (IE like stacking portrait and landscape images of the same subject) - It worked, but that was because all images were taken with a guided EQ mount and the star shapes were nice round and the triangulation algorithms DSS uses still worked.  I had to set an area in DSS that was inside the overlap for it to work, but it still means that that I didn't have to throw away a previous sessions data.  

[The Admiral]

DSS will rotate images to align them, but with Alt-Az imaging, stacking aside, the rotation that takes place during each exposure becomes sufficient to be visible on every image if the exposure duration gets too long. It's unavoidable I'm afraid. Think of it as a mild form of star trail images. It means that detail will be smeared and definition will be lost, even at the centre. Field rotation is at a minimum towards the West and East, and becomes large towards the zenith. Here is a good description https://calgary.rasc.ca/field_rotation.htm

Ian

[StuartT]

Right. I understand now. I was thinking of field rotation between exposures, but missing the important effect of field rotation within each exposure. I can see why that would be a problem.

Thanks!