Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

"Dithering" by rotating camera?

  • Please log in to reply
20 replies to this topic

#1 DubbelDerp

DubbelDerp

    Mariner 2

  • *****
  • topic starter
  • Posts: 292
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 13 November 2019 - 01:57 PM

Folks:

 

I don't have the ability to automatically dither. I'm doing the best I can with an unguided skyguider pro, which is pretty limiting. A few weeks ago when I was working on my image of M31, I collected about 2 hours of lights (limited by clouds) through my duo-band filter, Canon T3i, and AT60ED scope. I was integrating the Ha data using the "Ha-OIII extract Ha" debayer algorithm in APP, and getting some really bad walking noise throughout the image.

 

The next night was clear, so I got 9 additional hours of lights, but because of my manual set-up I had a slightly different rotation of the camera for the second set of data. This data set also showed significant walking noise, although much reduced compared to the 2-hour data set. So just for grins, I registered and integrated the two data sets together, and the result was a wonderfully smooth background compared to the individual integrated data sets.

 

This got me to thinking, could I do a low-budget version of "dithering" by changing the camera rotation a few times throughout the night? Last night was clear again, but with a full moon, so I did a crude test. I took 30 frames of M33 (wow is it hard to find with a nearly full moon, a duo-band filter, and being -15C outside!), rotated the camera a few degrees, and then let the camera run the rest of the night. Today, I ran the "extract HA" algorithm again and integrated the first 30 frames, the next 30 frames with the camera slightly rotated, the 60 consecutive frames after rotating the camera, and then the 60 frames with the initial 30 frames and the second set of 30 frames with the two camera rotated.

 

Here's a crop from the first integration of the 30 frames prior to rotating the camera. Please forgive the poor image quality, but there's quite a bit of noise in the background.

30_frames_before_rotation.jpg

 

Second set of frames after rotating the camera:

30_frames_after_rotation.jpg

 

Here's 60 frames after rotating the camera:

60_frames_after_rotation.jpg

 

Finally, here are the 60 frames including the 30 before and 30 after rotating the camera:

combined_stack_60_frames.jpg

 

To my eye, the 60 frames after rotating the camera still show some pattern noise across the image. But the combined image before and after, on my screen, looks much smoother and more uniform.

 

Just wanted to get the group's thoughts, for a simple set-up like mine, is this a reasonable approach to reduce some of the pattern noise that could otherwise be resolved with dithering? I find it much easier to use the camera rotator to change the rotation slightly than to try to make a small adjustment in declination with the SGP, but wanted to see if I was missing something (I probably am) obvious.

 

Thanks in advance!


  • Thaeland likes this

#2 Steve OK

Steve OK

    Surveyor 1

  • -----
  • Posts: 1546
  • Joined: 22 Sep 2007
  • Loc: OKC, OK

Posted 13 November 2019 - 04:45 PM

The only drawback I can see is that the effectiveness of the dithering would go to zero at the center of the frame where there is no angular displacement.  I don't use SGP, so I can't add anything about its ability to dither.  

 

Steve



#3 DubbelDerp

DubbelDerp

    Mariner 2

  • *****
  • topic starter
  • Posts: 292
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 13 November 2019 - 05:53 PM

Sorry, bad use of abbreviation on my part. SGP = skyguider pro in this case, so only one motorized axis. 
 

Agreed on the center of rotation... in my case, at the focal length of the AT60ED I usually get enough field drift over an hour that the center of rotation would move several pixels, so I think that if I did two different rotations it would minimize this effect? This scope has a camera rotator that moves the field flattener and camera together, so flats might not correct perfectly for vignetting, but hopefully it would be ok for dust motes. That would be the easier option, but I suppose it wouldn’t be too difficult to loosen the scope ring and rotate everything in unison. 



#4 whwang

whwang

    Mercury-Atlas

  • *****
  • Posts: 2910
  • Joined: 20 Mar 2013

Posted 13 November 2019 - 07:48 PM

It will be much better if you manually shift the telescope pointing a few times in a session, rather than rotating the camera a few times in a session. You do these by hand anyway, and I believe pushing the button of your mount controller (or rotating the knob) should be easier than rotating the camera.
  • elmiko, 17.5Dob and happylimpet like this

#5 Alen K

Alen K

    Apollo

  • -----
  • Posts: 1117
  • Joined: 25 Nov 2009

Posted 13 November 2019 - 08:08 PM

SGP has only one motorized axis and lacks even manual fine adjustment in declination. Trying to move just a little in declination without a fine control of some sort is probably a recipe for disaster. That would mean dithering in RA only, which is usually not recommended (worse than not dithering at all?). 


  • DubbelDerp likes this

#6 DubbelDerp

DubbelDerp

    Mariner 2

  • *****
  • topic starter
  • Posts: 292
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 13 November 2019 - 08:12 PM

That’s exactly what I was thinking. Without platesolving and goto, I’m really hesitant to touch the dec axis. But if I can do something reasonably close with only rotating the scope or camera.. I’m sure it’s not as good, but would it be better than nothing?

SGP has only one motorized axis and lacks even manual fine adjustment in declination. Trying to move just a little in declination without a fine control of some sort is probably a recipe for disaster. That would mean dithering in RA only, which is usually not recommended (worse than not dithering at all?). 



#7 whwang

whwang

    Mercury-Atlas

  • *****
  • Posts: 2910
  • Joined: 20 Mar 2013

Posted 13 November 2019 - 10:15 PM

Dither along just one axis isn't that bad, if the dither offset is sufficiently large and random.   With some flexure + imperfect tracking in the other axis, it may give you good results.  

 

The issue of rotating the camera is like what Steve said, the effect will be small or none in the center of the frame.  It can cause some ugly pattern.

 

It's unfortunate that your mount cannot effectively dither.  I think the best thing to do is to dither along RA, let the other axis naturally drift (by flexure or tracking error), and divide the session into two halves and rotate the camera 180 deg in between.  Small-angle rotation isn't really the right thing to do.

 

Cheers,

Wei-Hao


  • agavephoto, 17.5Dob and DubbelDerp like this

#8 17.5Dob

17.5Dob

    Fly Me to the Moon

  • ****-
  • Posts: 5392
  • Joined: 21 Mar 2013
  • Loc: Colorado,USA

Posted 13 November 2019 - 10:38 PM

A better solution would be to just dump the Canon camera, the source of all your FPN.....I've shot up  to 3- 4 hrs with my Nikon D5300 with no dithering, no problem.



#9 agavephoto

agavephoto

    Vostok 1

  • -----
  • Posts: 140
  • Joined: 14 Jul 2009

Posted 14 November 2019 - 12:40 AM

A better solution would be to just dump the Canon camera, the source of all your FPN.....I've shot up  to 3- 4 hrs with my Nikon D5300 with no dithering, no problem.

So, I shoot with a Canon, don't dither, and don't get the fixed pattern noise, on shorter or longer time scales as you mention ... because, I calibrate my data properly. If I don't calibrate the data, I will certainly see pattern noise, but this doesn't concern me much as I can test and understand the camera and the data it produces. 

 

EDIT - one day, I won't make typos ...


Edited by agavephoto, 14 November 2019 - 12:44 AM.

  • whwang and DubbelDerp like this

#10 Coconuts

Coconuts

    Ranger 4

  • *****
  • Posts: 332
  • Joined: 23 Sep 2012

Posted 14 November 2019 - 05:41 AM

The inevitable imperfection in polar alignment with a tracking mount such as the Sky Guider Pro will cause a built-in "dither" (more accurately, an offset) in Dec between subs.  The resulting declination drift would be significant over the course of an imaging session, but could still be entirely acceptable over the duration of any individual sub.  Such an offset between subs would be uniform rather than random, but changing the pauses between exposures would introduce variation in the offset.  Lengthening such pauses would also provide a knob by which to increase the declination drift offset, albeit at the expense of less overall imaging time.  One could even dial in an intentional level of polar misalignment to reach a target value of declination drift; for a quantitative look at how much you could get away with and not impact image quality, check out my post on that topic:

https://www.cloudyni...strophotography

 

All the best,

 

Kevin


  • DubbelDerp likes this

#11 happylimpet

happylimpet

    Skylab

  • *****
  • Posts: 4061
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 14 November 2019 - 06:07 AM

Rotation, nudges, drift, accidentally kicking the mount, its all good for dithering.

 

You just need numerous offsets so that each part of the image is recorded on different parts of the sensor.


  • DubbelDerp likes this

#12 DubbelDerp

DubbelDerp

    Mariner 2

  • *****
  • topic starter
  • Posts: 292
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 14 November 2019 - 08:43 AM

No, there's nothing wrong with my Canon camera. I can run all night without adjustments and get smooth data after stacking with no fixed pattern noise. In this unique case though, I'm trying to extract only the H-a data using a dual band H-a/OIII filter, and surprise, the red channel from the bayer matrix has a whole lot less signal. I would expect any DSLR that isn't mono-modded to suffer from poorer signal in the red channel.

 

This is just an exercise in trying to get the best H-a data I can with a sub-optimal camera (OSC DSLR) and sub-optimal mount (camera tracker.) I'm trying to see what kind of low-tech adjustments I can make to improve the quality of the data, and this thread has given me some good ideas. I suppose I can go back to taking wide-angle 50mm milky way shots... but what's the fun in that?


  • agavephoto and happylimpet like this

#13 DubbelDerp

DubbelDerp

    Mariner 2

  • *****
  • topic starter
  • Posts: 292
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 14 November 2019 - 08:46 AM

Dither along just one axis isn't that bad, if the dither offset is sufficiently large and random.   With some flexure + imperfect tracking in the other axis, it may give you good results.  

 

The issue of rotating the camera is like what Steve said, the effect will be small or none in the center of the frame.  It can cause some ugly pattern.

 

It's unfortunate that your mount cannot effectively dither.  I think the best thing to do is to dither along RA, let the other axis naturally drift (by flexure or tracking error), and divide the session into two halves and rotate the camera 180 deg in between.  Small-angle rotation isn't really the right thing to do.

 

Cheers,

Wei-Hao

 

Wei-Hao, thank you for the thoughtful reply. That makes a lot of sense.

 

It occurs to me that after rotating the camera, I did nudge the camera tracker in RA to re-center the target, since it had drifted a bit. This, combined with the rotation, would have caused a transverse shift across the sensor. It's quite possible that this by itself was what caused the reduction in noise that I saw across the image.

 

I appreciate the suggestions! I am budgeting for a more capable mount in the future, but at the moment, kids are expensive...



#14 happylimpet

happylimpet

    Skylab

  • *****
  • Posts: 4061
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 14 November 2019 - 09:27 AM

I suppose I can go back to taking wide-angle 50mm milky way shots... but what's the fun in that?

Exactly! Always fun to maximise the possibilities with the equipment you have. If I hadnt started doing DSO imaging with 2second subs and an uncooled ASI120MC I wouldnt have started on the road Im on now, and its been a lot of fun and Ive learnt a lot.

 

I appreciate the suggestions! I am budgeting for a more capable mount in the future, but at the moment, kids are expensive...

**** it, we've got our first due any day now....


  • DubbelDerp likes this

#15 DubbelDerp

DubbelDerp

    Mariner 2

  • *****
  • topic starter
  • Posts: 292
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 14 November 2019 - 09:56 AM

Exactly! Always fun to maximise the possibilities with the equipment you have. If I hadnt started doing DSO imaging with 2second subs and an uncooled ASI120MC I wouldnt have started on the road Im on now, and its been a lot of fun and Ive learnt a lot.

 

**** it, we've got our first due any day now....

Hey, congrats! So when you ghost us in the near future, we'll know why...


  • happylimpet likes this

#16 Alen K

Alen K

    Apollo

  • -----
  • Posts: 1117
  • Joined: 25 Nov 2009

Posted 14 November 2019 - 01:47 PM

A better solution would be to just dump the Canon camera, the source of all your FPN.....I've shot up  to 3- 4 hrs with my Nikon D5300 with no dithering, no problem.

 

You dithered in the past: https://www.cloudyni...g/#entry6307794

 

Are you saying you don't do it anymore? Or that you don't do it if your total exposure time is sufficiently long (3+ hours)?



#17 KLWalsh

KLWalsh

    Viking 1

  • -----
  • Posts: 608
  • Joined: 19 Feb 2014
  • Loc: North Georgia, USA

Posted 14 November 2019 - 09:28 PM

Rotating the camera slightly between exposures will help eliminate diffraction spikes from your secondary spider, if the scope has one, or from the iris diaphragm of your camera lens if you’re using a consumer camera lens.

#18 Coconuts

Coconuts

    Ranger 4

  • *****
  • Posts: 332
  • Joined: 23 Sep 2012

Posted 14 November 2019 - 10:25 PM

So I am still very early in my digital astrophotpgraphy efforts, and have so far only completed a couple of wide field photos using camera lenses. But I recently purchased a Takahashi e-130D as my first reflective astrograph, and this most recent post regarding secondary diffraction spikes being suppressed by camera angle rotation seemed relevant to a couple of questions that I had:

 

  • The Tak e-130D has a nice, built-in camera rotator.  But if I frame objects in the field by using it, do I have to re-take all of my flats?  Mirror scopes may differ materially from refractors in this regard. 
  • For secondary diffraction suppression via small field rotations, and assuming that you don't regard diffraction spikes as "benefits", what angle of dither (relative to focal length, of course) do you suggest as a starting point?

All the best,

 

Kevin



#19 17.5Dob

17.5Dob

    Fly Me to the Moon

  • ****-
  • Posts: 5392
  • Joined: 21 Mar 2013
  • Loc: Colorado,USA

Posted 14 November 2019 - 10:51 PM

So I am still very early in my digital astrophotpgraphy efforts, and have so far only completed a couple of wide field photos using camera lenses. But I recently purchased a Takahashi e-130D as my first reflective astrograph, and this most recent post regarding secondary diffraction spikes being suppressed by camera angle rotation seemed relevant to a couple of questions that I had:

 

  • The Tak e-130D has a nice, built-in camera rotator.  But if I frame objects in the field by using it, do I have to re-take all of my flats?  Mirror scopes may differ materially from refractors in this regard. 
  • For secondary diffraction suppression via small field rotations, and assuming that you don't regard diffraction spikes as "benefits", what angle of dither (relative to focal length, of course) do you suggest as a starting point?

All the best,

 

Kevin

ANY change in the camera/OTA coupling will require new flats......



#20 vidrazor

vidrazor

    Mariner 2

  • -----
  • Posts: 285
  • Joined: 31 Oct 2017

Posted 20 November 2019 - 02:13 PM

Curious, I tend to re-align Polaris and check focus every 100 frames or so on my unguided SkyGuider Pro. Wouldn't that offer any significant dithering? I certainly haven't seen it from my stacks from an M4/3 body, I still see walking noise.

#21 happylimpet

happylimpet

    Skylab

  • *****
  • Posts: 4061
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 22 November 2019 - 02:44 PM

Rotating the camera slightly between exposures will help eliminate diffraction spikes from your secondary spider, if the scope has one, or from the iris diaphragm of your camera lens if you’re using a consumer camera lens.

No it wont, unless you rotate the whole OTA. The spikes remain unrotated relative to the stars on camera rotation, and so will stack in the same places.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics