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

Substantial Difference in Noise Across DSLR image

  • Please log in to reply
15 replies to this topic

#1 ParagPA

ParagPA

    Sputnik

  • -----
  • topic starter
  • Posts: 46
  • Joined: 12 Aug 2019

Posted 11 September 2019 - 03:05 PM

Hello,

 

My setup: 

 

ED80T, CEM 40, EOS 80D, (50MM guidescope + ZWO ASI), Field Flattener

 

I took 30x 150s ISO 1600 lights and dark subs of M31 last night + 50 bias. 

 

Near full moon, but otherwise clear, temps around 30-32C (per CCD image)

 

Stacked in DSS, 90% best frames. Quick curve stretch in PI. 

 

Here's what I'm seeing that confuses me:

 

1) The noise on the left edge of the image is substantially less than everywhere else.

2) Stars on left side are also elongated - but nice and circular everywhere else - and tracking (PHD2) was good all night long

3) Strangely perfectly circular dust spot on upper left. 

 

Any ideas what's causing this? 

 

I looked at stretched versions of the dark and bias masters. Dark looks like random noise, bias looks mostly flat, but you can see some vertical banding. 

 

I didn't take any flats - but is this kind of difference in noise across an image normal w/out flats? 

 

Thanks...

Attached Thumbnails

  • M31_Stacked_Med.jpg

Edited by ParagPA, 11 September 2019 - 03:05 PM.


#2 bobzeq25

bobzeq25

    Hubble

  • *****
  • Posts: 16821
  • Joined: 27 Oct 2014

Posted 11 September 2019 - 03:19 PM

Flats would fix a lot of that.  Gradient reduction would help improve things even farther.

 

Elongated stars on one side can be optical issues and/or tilt in the optical train.  Stars are demanding targets.

 

Dust spots are frequently circular.  Again, the cure is flats.


  • zakry3323 likes this

#3 knight_parn

knight_parn

    Vostok 1

  • -----
  • Posts: 148
  • Joined: 03 Oct 2018
  • Loc: Peterborough, UK

Posted 11 September 2019 - 03:23 PM

Elongated stars on one side are most likely caused by tilt (not necessarily sensor tilt). I had a similar issue with my previous flimsy low profile t2 adapter when I took my picture of the western veil. Fixed it with a better built normal profile t2 adapter.


Edited by knight_parn, 11 September 2019 - 03:26 PM.

  • ParagPA likes this

#4 happylimpet

happylimpet

    Soyuz

  • *****
  • Posts: 3896
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 11 September 2019 - 03:25 PM

It may be a little pedantic but its useful to use the right terminology here. Thats not  a difference in noise, its a difference in level. And yes, its a job for flats!


  • ParagPA likes this

#5 ParagPA

ParagPA

    Sputnik

  • -----
  • topic starter
  • Posts: 46
  • Joined: 12 Aug 2019

Posted 11 September 2019 - 03:36 PM

It may be a little pedantic but its useful to use the right terminology here. Thats not  a difference in noise, its a difference in level. And yes, its a job for flats!

Pedantic is good! I appreciate it. It was much more noticeable to me as a difference in background noise vs. the impact on levels on the actual stars... But I get exactly what you're saying.

 

Just did a flat calibration in SGP and am taking 30 flats... We'll see how that helps the stacking.


  • happylimpet likes this

#6 ParagPA

ParagPA

    Sputnik

  • -----
  • topic starter
  • Posts: 46
  • Joined: 12 Aug 2019

Posted 11 September 2019 - 03:37 PM

Elongated stars on one side are most likely caused by tilt (not necessarily sensor tilt). I had a similar issue with my previous flimsy low profile t2 adapter when I took my picture of the western veil. Fixed it with a better built normal profile t2 adapter.

I'm using an Orion t2 adapter + field flattener... Which I thought would keep everything co-axial. But maybe not... 



#7 zakry3323

zakry3323

    Viking 1

  • *****
  • Posts: 941
  • Joined: 11 Apr 2016
  • Loc: Pittsburgh

Posted 11 September 2019 - 03:37 PM

I'm +1 for the culprit being tilt. The star elongation at the left and the lack of flat illumination across the field makes me think that there's some drop-off in light reaching that side of the sensor. You could get a similar drop-off in light from an OAG prism being positioned too closely to the edge of the sensor, but I don't think it would also cause elongated stars. It may be easier to diagnose if you list the equipment that you use, preferably in your signature :)



#8 ParagPA

ParagPA

    Sputnik

  • -----
  • topic starter
  • Posts: 46
  • Joined: 12 Aug 2019

Posted 11 September 2019 - 03:41 PM

I'm +1 for the culprit being tilt. The star elongation at the left and the lack of flat illumination across the field makes me think that there's some drop-off in light reaching that side of the sensor. You could get a similar drop-off in light from an OAG prism being positioned too closely to the edge of the sensor, but I don't think it would also cause elongated stars. It may be easier to diagnose if you list the equipment that you use, preferably in your signature smile.gif

Thanks! I will modify my signature. 

 

The equipment list was in my original post: 

 

ED80T, CEM 40, EOS 80D, (50MM guidescope + ZWO ASI), Field Flattener


  • zakry3323 likes this

#9 ParagPA

ParagPA

    Sputnik

  • -----
  • topic starter
  • Posts: 46
  • Joined: 12 Aug 2019

Posted 11 September 2019 - 03:54 PM

I'm +1 for the culprit being tilt. The star elongation at the left and the lack of flat illumination across the field makes me think that there's some drop-off in light reaching that side of the sensor. You could get a similar drop-off in light from an OAG prism being positioned too closely to the edge of the sensor, but I don't think it would also cause elongated stars. It may be easier to diagnose if you list the equipment that you use, preferably in your signature smile.gif

I'm wondering if it could be the focuser. Even though it's tightened down, I've loosened it in the past to rotate it, and I've noticed a bit of slop. 

 

It's the standard 2-step focuser that comes with the ED80T



#10 ParagPA

ParagPA

    Sputnik

  • -----
  • topic starter
  • Posts: 46
  • Joined: 12 Aug 2019

Posted 11 September 2019 - 04:04 PM

OK - I took 30 flats and re-stacked. This is what I'm getting. The circular spot is still there - and even more pronounced. 

 

Left image is original (stretched in PI), right image is re-stacked w/ flats (stretched in PI). 

 

I thought that by applying the flats, my levels would go down across the board and be closer to the left edge of the original image - instead it's gone up across the board. There's still a gradient in the right image, but it looks much more monotonic - brighter in bottom-right, decreasing near linearly towards top-left. 

 

That might argue for a misalignment somewhere in the optical chain also - since you'd expect a small angle to create a linear change in levels across the image, instead of the dark->bright->dark you see in the left image.

 

Thanks for the help!

 

 

Attached Thumbnails

  • Flats.jpg


#11 zakry3323

zakry3323

    Viking 1

  • *****
  • Posts: 941
  • Joined: 11 Apr 2016
  • Loc: Pittsburgh

Posted 11 September 2019 - 04:14 PM

OK - I took 30 flats and re-stacked. This is what I'm getting. The circular spot is still there - and even more pronounced. 

 

Left image is original (stretched in PI), right image is re-stacked w/ flats (stretched in PI). 

 

I thought that by applying the flats, my levels would go down across the board and be closer to the left edge of the original image - instead it's gone up across the board. There's still a gradient in the right image, but it looks much more monotonic - brighter in bottom-right, decreasing near linearly towards top-left. 

 

That might argue for a misalignment somewhere in the optical chain also - since you'd expect a small angle to create a linear change in levels across the image, instead of the dark->bright->dark you see in the left image.

 

Thanks for the help!

 

Quite an improvement! You still have lp gradients to deal with, but that's what ABE/DBE is for :)

You can check to confirm if tilt is really an issue with CCD Inspector. The trial should get you through hunting down the problem before you need to buy it :) https://www.ccdware....s/ccdinspector/



#12 bobzeq25

bobzeq25

    Hubble

  • *****
  • Posts: 16821
  • Joined: 27 Oct 2014

Posted 11 September 2019 - 06:05 PM

OK - I took 30 flats and re-stacked. This is what I'm getting. The circular spot is still there - and even more pronounced. 

 

Left image is original (stretched in PI), right image is re-stacked w/ flats (stretched in PI). 

 

I thought that by applying the flats, my levels would go down across the board and be closer to the left edge of the original image - instead it's gone up across the board. There's still a gradient in the right image, but it looks much more monotonic - brighter in bottom-right, decreasing near linearly towards top-left. 

 

That might argue for a misalignment somewhere in the optical chain also - since you'd expect a small angle to create a linear change in levels across the image, instead of the dark->bright->dark you see in the left image.

 

Thanks for the help!

Dust has an annoying tendency to move.  Particularly if you take the optical train off the mount, or if your camera has a sensor cleaning routine.



#13 OldManSky

OldManSky

    Surveyor 1

  • *****
  • Posts: 1706
  • Joined: 03 Jan 2019
  • Loc: Valley Center, CA USA

Posted 11 September 2019 - 06:08 PM

It may be a little pedantic but its useful to use the right terminology here. Thats not  a difference in noise, its a difference in level. And yes, its a job for flats!

I'll vote for "not pedantic at all."

Because that's really the point.  The left side has experienced some sort of vignetting (or the right side was "seeing" the moon and got swamped with signal -- but unwanted signal).

In either case, the noise across the image is (within random variation) the same.  

What varies is the S/N ratio -- 'cause there's constant noise, and differing signal.

 

I may look like "more noise."  It's not.  The noise is pretty much the same across the whole image.  What it IS, is what happylimpet said -- a difference in signal, hence S/N ratio.  Making the noise more visible without the signal to hide it.


  • happylimpet and ParagPA like this

#14 knight_parn

knight_parn

    Vostok 1

  • -----
  • Posts: 148
  • Joined: 03 Oct 2018
  • Loc: Peterborough, UK

Posted 12 September 2019 - 04:30 AM

OK - I took 30 flats and re-stacked. This is what I'm getting. The circular spot is still there - and even more pronounced. 

 

Left image is original (stretched in PI), right image is re-stacked w/ flats (stretched in PI). 

 

I thought that by applying the flats, my levels would go down across the board and be closer to the left edge of the original image - instead it's gone up across the board. There's still a gradient in the right image, but it looks much more monotonic - brighter in bottom-right, decreasing near linearly towards top-left. 

 

That might argue for a misalignment somewhere in the optical chain also - since you'd expect a small angle to create a linear change in levels across the image, instead of the dark->bright->dark you see in the left image.

 

Thanks for the help!

Flats should preferably be taken at the same focus distance as the lights and during the same imaging session before you take apart the setup (so the dust spots don't move). Otherwise their effectiveness are reduced.

 

Unless your focuser is of the same build quality as a FTF, I wouldn't recommend using its rotator. Personally I just the rotate the camera itself if I want to frame my target in a different angle since the FR/FF I use is attached to the focuser with a 2" nosepiece. A lot of people just rotate the whole scope within the cradles. Try re-tightening the focuser rotator screws and make sure the focuser does not wiggle.


Edited by knight_parn, 12 September 2019 - 04:32 AM.

  • ParagPA likes this

#15 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23591
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 12 September 2019 - 06:20 PM

It may be a little pedantic but its useful to use the right terminology here. Thats not  a difference in noise, its a difference in level. And yes, its a job for flats!

Actually, it is both. It is both  change in signal level and a change in noise. In fact, it is three things simultaneously...also a change in SNR. The darker areas have weaker signal, lower to the noise floor, thus swamp the read noise and FPN less, so are both noisy and low SNR. Technically speaking noise grows along with signal, just more slowly, so the central area may in fact may even have higher noise, but has a higher SNR (at least in the un-corrected frame without flats). 

 

Now, once flats are applied, the difference in level will disappear. Noise in the darker areas will be amplified along with the level, and thus will likely become more noisy and lower SNR than the center of the frame. 


  • ParagPA likes this

#16 happylimpet

happylimpet

    Soyuz

  • *****
  • Posts: 3896
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 13 September 2019 - 04:32 AM

Actually, it is both. It is both  change in signal level and a change in noise. In fact, it is three things simultaneously...also a change in SNR. The darker areas have weaker signal, lower to the noise floor, thus swamp the read noise and FPN less, so are both noisy and low SNR. Technically speaking noise grows along with signal, just more slowly, so the central area may in fact may even have higher noise, but has a higher SNR (at least in the un-corrected frame without flats). 

 

Now, once flats are applied, the difference in level will disappear. Noise in the darker areas will be amplified along with the level, and thus will likely become more noisy and lower SNR than the center of the frame. 

Very true, I was aware of this but was simplifying to get the important point across! I suspect the difference in level is actually very small, as a stretch has already been applied, and so the difference in noise will also be very small.  The difference in apparent level is therefore the dominant concern.


  • ParagPA likes this


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