70% of frames rejected during stacking - donut stars?

[MercianDabbler]

I was quite pleased with myself managing 145 light frames of M104 and surroundings on Friday night... until I came to stack them.

With default settings, DSS picked just 35 for stacking.

I've played with the reducing the star threshold to 5% (41 frames), median filter (40 frames) and boosting the brightness by 50% (43 frames).

Having compared the stars in some of the higher scoring frames with the lower ones, I suspect this sort of thing is the problem...

I don't know if this is the actual cause but the 'donut' like stars do seem to be common in the low scored frames whereas the higher scores have discs with the middle brighter than the edges.

I also don't know why I'm getting donut shaped stars. For certain my focus is a bit hit and miss - there is a 'band of uncertainty' where the star size on the camera screen doesn't change much. On one side it turns into a greenish blob and on the other side I get a spiky blue halo. On this occasion I was aiming for the transition between the 'dead band' and the blue halo. Seeing was probably not great - the stars I was using for focusing were pretty shimmery and LP was also not great. I certainly wasn't clobbering the mount - the camera was triggered via Wifi. I don't think there was much wind either.

Vixen SP102, Canon EOS 70D at prime focus, Vixen RA drive on the SP mount and nothing at all on the decl axis. ISO 1600 and 30s frames. ...  none of which is radically different to what I've done previously without hitting this issue... although ISO3200 and 20s is probably what I've used more often.

I suspect that most of my frames are 'toast' but it would be nice to at least be able to treat this as a learning experience... but I'm not sure what to learn.

[MercianDabbler]

After pausing for thought I've been looking a bit more at the issue. It seems that the rejected frames were mostly the later ones. After spotting this I did a graph of the scores to see if that helped me to understand what was going on...

For comparison I did the same thing for a shorter (69 frames) M104 session 4 days earlier. This session was largely toasted by a different issue but that won't affect the scores.

I did wonder if it was caused by the mount going 'over the hill'. Peak altitude for M104 was at about 23:15 which is around frame 74 on the problem session and around packing up time on the earlier one. It may be a factor but it's clear that the frames were degrading pretty much from the get-go on Friday night. The frames for both sessions started at around 22:20.

So I'm still stumped. My gut feel says this is mechanical rather than optical but I don't know. I have checked the donuts outside of DSS and they are real donuts. Some of the bad frames seem to have wobble mainly in the RA direction which is probably par for the course but a large proportion of them show the donut issue.

[carastro]

I recall having donut stars due to the focus not being spot on.
 

Carole 

[SteveNickolls]

Get a Bhatinov mask for your telescope and nail focus using a high ISO.  If you are careful you may be able to carry over good focus to the next session.

Good luck.

Cheers 

Stdve

[MercianDabbler]

26 minutes ago, carastro said:

I recall having donut stars due to the focus not being spot on.
 

Carole 

Thanks. I've been searching online and found this suggested... but it was for a reflector.

I'm no expert but I'd expect an out of focus lens based optical system to produce a spike in the middle rather than a hole. Certainly that's how it looks when focusing using the live view and 10x magnification.

[MercianDabbler]

16 minutes ago, SteveNickolls said:

Get a Bhatinov mask for your telescope and nail focus using a high ISO.  If you are careful you may be able to carry over good focus to the next session.

I've got one but I never really got on with it - the lines on mine seem vague even with a very bright star and I find myself being just as uncertain as I am using the magnified live view on the camera.

Would you expect one to work well on a 4 inch achromat? I have no idea what to expect, I just know how disappointing my experience with the mask has been.

[happy-kat]

On a small 2" refractor I use a home made of stiff card a Lord Y mask with a camera this creates a strong spike. The canon 1100d I use I control by an Android mobile and get a bigger screen to focus with.

[MercianDabbler]

48 minutes ago, happy-kat said:

On a small 2" refractor I use a home made of stiff card a Lord Y mask with a camera this creates a strong spike. The canon 1100d I use I control by an Android mobile and get a bigger screen to focus with.

That's a new one on me. I shall make one and see how I get on with it.

[happy-kat]

As I use DSLR Controller app on the mobile I save the image locally and do a bigger pinch zoom to see the defraction spike to double check it

[MercianDabbler]

21 minutes ago, happy-kat said:

As I use DSLR Controller app on the mobile I save the image locally and do a bigger pinch zoom to see the defraction spike to double check it

I do similar on my tablet. The Lord mask seems easy enough to make so I will make one to compare.

Stars seem to have been towed away for now though.

[SteveNickolls]

13 hours ago, MercianDabbler said:

Would you expect one to work well on a 4 inch achromat?

Well yes. Whenever I use my StarTravel 102 that's the tool I would use to help focus. You do need a bright star, not a faint star or double star etc and boost up your DSLR's ISO. Use 'Live View' on the touchscreen to help with the focus. good luck with the attempt.

Cheers,
Steve

[MercianDabbler]

55 minutes ago, SteveNickolls said:

Well yes. Whenever I use my StarTravel 102 that's the tool I would use to help focus. You do need a bright star, not a faint star or double star etc and boost up your DSLR's ISO. Use 'Live View' on the touchscreen to help with the focus. good luck with the attempt.

I suspect I used Rigel or Sirius last time I tried... plus 12800 ISO and live view magnification. I can't remember exactly what the pattern looked like but I do remember that it was nowhere near as crisp as the pictures online and that I struggled to have any real certainty about when the lines were all crossing at the same point.

I plan to have another try with the Bahtinov and also make a 'Y' mask to see if either works better than the other.

I'm still not sure that refractors produce donuts when out of focus but my search results online were not exactly conclusive... reflectors definitely do and there are also references to software created donuts but inspection of my raws tends to exclude the latter. I intend to do some testing to find out... once the stars become available again.

Focus would at least be a better answer than some sort of drive induced mechanical oscillation.

If it is focus then I suspect (based on the quality graph) that mine started OK but went bad during the session due to temperature drop. Both nights were OK for setting up without a coat but Friday felt like it got colder than the previous session.

Refocussing in mid session is something I might need to do but not something I relish - I will definitely need to be able to reframe the shot much more quickly than I have achieved so far if I'm not going to burn up all of the available session time.

[The Lazy Astronomer]

1 hour ago, MercianDabbler said:

Refocussing in mid session is something I might need to do but not something I relish - I will definitely need to be able to reframe the shot much more quickly than I have achieved so far if I'm not going to burn up all of the available session time.

Not sure how steady your night time temperatures are, or how well your focuser holds once set, but l find that if I leave my equipment (100mm refracter) outside for an hour or more before starting imaging, I don't need to refocus throughout unless I switch filters (a typical session for me is 4 - 5 hours).

I use an autofocuser and NINA's HFR measurement tool, so I have a quantitative measurement of focus on each sub.

[carastro]

Just to clarify, l was using a refractor when l got donut stars.   But the donut’s didn’t appear until after stacking snd processing, so was unable to see this at the time of focussing.   Also what are you stacking with?  
 

In DSS l found the cosmetic tab would make holes in the stars.  Untick that and it stopped doing so.    I think a number of people have fallen foul of that. 
 

Definitely use a mask to focus, much more accurate than the eye. 
 

Carole

Edited May 2, 2022 by carastro

[MercianDabbler]

1 hour ago, The Lazy Astronomer said:

Not sure how steady your night time temperatures are, or how well your focuser holds once set, but l find that if I leave my equipment (100mm refracter) outside for an hour or more before starting imaging, I don't need to refocus throughout unless I switch filters (a typical session for me is 4 - 5 hours).

I use an autofocuser and NINA's HFR measurement tool, so I have a quantitative measurement of focus on each sub.

The focuser is the bog standard SP102 one - nothing at all fancy. I have been after an add-on mechanical fine focusser but it's never been in stock when I've looked and I've sort of abandoned my hunt at the moment.

[MercianDabbler]

55 minutes ago, carastro said:

Just to clarify, l was using a refractor when l got donut stars.   But the donut’s didn’t appear until after stacking snd processing, so was unable to see this at the time of focussing.   Also what are you stacking with?  
 

In DSS l found the cosmetic tab would make holes in the stars.  Untick that and it stopped doing so.    I think a number of people have fallen foul of that.

Yes, DSS here too but the holes are visible in the raw images when viewed with Darktable (even after undoing the default DT modules just to be on the safe side) so they are not being added by DSS.

[alacant]

19 minutes ago, MercianDabbler said:

they are not being added by DSS

Hi

I think you need to concentrate more of the halo into the red and green. Try the focusing method described here. Works well with old camera lenses, low end EDs and achromatic doublets. 

We'd recommend Siril for calibration and stacking.

Cheers and HTH.

Edited May 2, 2022 by alacant

[MercianDabbler]

Last night was clear. I had other stuff going on but managed to get the SP102 out for an hour of testing before bedtime. I did a rough polar alignment and then used Polaris for testing.

First conclusion - yes I can produce exactly the same donut appearance by having the wrong focus... at least on the fainter stars. Thanks to the people who said that focus was the problem.

Second conclusion - the amount by which I needed to have the focus 'off' was quite large.

I tried to find a way to measure the focuser travel but  I could not get my calipers in between the camera and the telescope tube. In the end the best I could manage was a steel rule and even then I had to hold it some distance away and do my best to eyeball the measurement. Good focus seemed to be with around 13.5mm of the shiny tube showing. The donuts were produced at 12mm.

Having looked up the thermal expansion of aluninium (up to 24 micrometers per meter Kelvin) and assuming a full meter for the tube and a very pessimistic scenario of zero precooling and a 15 degree temperature drop I got 0.36mm which is only about 25% the movement measured in my focus test. Maybe the air temp inside the tube also plays a part in the focal length? I don't know how to do any calculations around this.

So I'm happy that focus is the explanation now but the quality numbers seem to show that focus was OK at the beginning and then went downhill over pehaps an hour so exactly what went wrong remains a bit of a puzzle.

While I was set up I also had another try with the Bahtinov mask. Viewed in live view at 12800 ISO and 10x magnification it was again difficult to read - the spikes seem quite short and broad and have a prominent bright rainbow area close to the star to add further confusion to the picture. Then I thought... why not take a photo to show people on here? So I did... and my 5 second exposure looks totally different from the way things look in live view - rather more the way that I've seen other people's results online. So if taking a frame or two is the answer that that's a useful learning point. The live view 'look' of Polaris is more akin to how the fainter stars look in the picture below

BTW, I think the mask shows focus pretty near to ideal... and the focus was achieved by my usual method of minimising the apparent size of the star so I think that's saying that my usual focusing technique is not terrible.

I also quickly cobbled together something resembing a Lord 'Y' mask for comparison. I did no research on dimensions so guessed 6mm wide bars and the same angles used by the Bahtinov and cut three plastic strips, glued them together at the centre and attached them to the lens cell with low tack masking tape. This produced visible spikes in live view but they are too small to be usable. Again the result from taking a 5 second frame with the camera was very different (and much better).

I'll probably stick with the Bahtinov for the SP102 and pursue the Lord mask for some of my other optics for which I don't have a Bahtinov.

[carastro]

Ah yes, I must admit I use the camera when using the Bahtinov mask.  Didn't realise you were trying to do it without.  Yes if you are using a colour camera you will get a rainbow on the Bahtinov spikes.  

But that seems to have nailed your problem.  Great news.

Carole  

Edited May 5, 2022 by carastro