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

Testing Oiii filter for the Moon's Impact

  • Please log in to reply
27 replies to this topic

#1 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 04 March 2015 - 03:02 AM

Another post here got a little off-topic as we considered the moon's impact on Oiii, and I figured I'd do a test tonight since the moon was out/bright and I'd pulled the gear inside last week because of severe weather (ie .. almost a full 1/10th of an inch of rain over the last week!  :grin: ) and I needed to break things back in.

 

So, after getting everything setup and running, I slewed off to the east, ran autofocus on my Oiii filter (3nm Astrodon), and then took a series of 10 minute guided subs. The first was 45 degrees from the moon, the second was 30 degrees from the moon, the third was 20 degrees, and the 4th was 10 degrees- but it experienced 'technical difficulties' and wasn't used.   :shocked:   I tried to keep the aimpoint for all the tests close to 45 degrees of altitude to keep things consistent.

 

I was mainly testing for gradient; looking at how close to the full moon I could image with my particular setup. Of note, my scope has a tiny lens shield - normally I add some foam to extend it but I ran into problems with it so I went with just the lens shield with the AT65EDQ.  I normally will shoot no closer than 40 degrees to the moon - but that's just a number I used without any real basis behind it. So tonight I tested.

 

So, using the three 10 minute subs (45, 30 and 20 degrees from the moon), I ran ABE to look at the gradient, and ran image stats and then stretched the image and ran AberationInspector.  Here's the results:

 

Degrees from the Moon
                        20          30            40

Mean             1827     1213         899.7
Median          1841     1217         901.3
avgDev         101.5      37.59        12.76
MAD             116.6      42.15        12.74
minimum       1644     1128         868.4
maximum      1970     1266         918.7

 

and here's the AberationInspector views:

 

20 degrees off:

O3_MT_20_mosaic.jpg

 

 

 

30 degrees off:

O3_MT_30_mosaic.jpg

 

 

 

45 degrees off ...

O3_MT_45_mosaic.jpg

 

 

 

 

Based on this testing, I'd probably be willing to shoot a lot closer to the moon. Although the gradient is noticeable at 30 degrees, it is smooth and easily corrected in PI.  20 degrees started showing problems, but I feel a better lens hood/shield would cut down on the glare enough to make it workable.  Any closer than 20 is pushing it with my system.

 

I had planned on some more tests, but the gremlins came out in force and I decided to shut down for the night - maybe more tomorrow. Hope this helps.


  • Footbag, David Ault, Madhup Rathi and 4 others like this

#2 terry59

terry59

    Cosmos

  • *****
  • Posts: 8962
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 04 March 2015 - 08:32 AM

I love testing equipment....good stuff Goofi :)



#3 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 04 March 2015 - 08:36 AM

Thanks Terry ... I kind of like testing stuff too, when the conditions aren't great for imaging.



#4 JJK

JJK

    Gemini

  • -----
  • Posts: 3462
  • Joined: 27 Apr 2008

Posted 04 March 2015 - 09:35 AM

While testing two FLI ProLine16803 cameras on my Ceravolo 300 at f/4.9 (1.26"/pixel, 1.43* x 1.43* FOV).  I recently imaged IC410 with the Moon perilously closeby (5 nm bandpass Astrodon H-a).  The results suggested the Moonlight had very little effect.  I usually use only 3 nm bandpass NB filters, but a team effort I'm joining likely wants to capture NII with the H-a filter.

 

The scope's baffling is pretty important.


  • Jon Rista likes this

#5 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 04 March 2015 - 10:21 AM

JJK, I agree baffling is important. I was testing Oiii because it's most impacted by moonlight.   Ha and Sii I don't have any problems with unless I'm pointed right at the moon. 

 

What started this was a quote from Don Goldman of Astrodon ...

 

 

There is no absolute cut-off where a narrowband filter will not be affected by moonlight. It is a continuum. The group is correct. SII, and H-a filters in the red region are not as affected by moonlight. However, moonlight, being mostly reflected sunlight, peaks near 500 nm, which is exactly where OIII is.  Thus, OIII is more problemmatic with moonlight. This is where the bandwidth (FWHM) of the filter comes into play. You're dealing with Signal-to-Noise (S/N). The Signal is largely due to the transmission of your filter. The Noise is related to many factors, but is largely controlled by the background signal.  Going from your 15 nm to an 8 nm narrowband filter will nearly halve your background ADU (counts. It is a linear relationship. If your 8 nm filter maintains the same high peak transmission of your 15 nm filter, than your S/N improves while having lower background signals and lower gradients. Repeat that again from 8 to 5 or 3 nm, and the improvement continues, with the caveat that you MUST maintain high transmission, which is much harder to do as the filter is made narrower. I often image with 5 nm H-a and SII and 3 nm OIII for this reason. I'll image any night when the moon is up as long as its not coming right down the "barrel". How far away perhaps depends upon the brightness of your target. You will see an improvement in structural contrast and will see background features pop out with narrower filters, also. Hope this helps.



#6 APshooter

APshooter

    Soyuz

  • *****
  • Posts: 3653
  • Joined: 14 Oct 2013
  • Loc: Camby, In.

Posted 04 March 2015 - 11:01 AM

Good to know, Goofi, I feel more confident now about my narrowband choice and shooting in moonlight.



#7 schmeah

schmeah

    Fly Me to the Moon

  • *****
  • Posts: 5108
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 04 March 2015 - 01:04 PM

Very interesting and most informative. I had been thinking about doing something similar.

Derek

#8 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 04 March 2015 - 03:12 PM

Thanks Derek and APShooter ....

 

I might follow-up with a filter comparison - shooting all three narrowband the same distance from the moon to compare/illustrate gradient impact. 



#9 Footbag

Footbag

    Cosmos

  • *****
  • Posts: 9115
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 04 March 2015 - 08:47 PM

Thanks Derek and APShooter ....

 

I might follow-up with a filter comparison - shooting all three narrowband the same distance from the moon to compare/illustrate gradient impact. 

 

Great comparison!

 

Can I ask what your bias pedestal was?  And this is f6.5?  Using ADU, bandwidth, f-ratio and distance to the moon, you can get a pretty good comparison between setups.  At this point ai almost want to confirm that linear relationship of background adu and filter pass.  Could it really be that good?  Your test shows considerable improvement over what I would expect from my 8nm OIII with my SCT.  And it would be all contrast.

 

The comparison of the different filters would also be nice!  



#10 David Ault

David Ault

    Gemini

  • -----
  • Posts: 3424
  • Joined: 25 Sep 2010
  • Loc: Georgetown, TX

Posted 04 March 2015 - 09:00 PM

Excellent write up Goofi!  I'm definitely curious what you find if you get the chance to do the other narrowband filters as well.

 

Regards,

David



#11 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23352
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 04 March 2015 - 09:14 PM

Thanks for the writeup, Goofi! This is good stuff.

 

The gradient at 20 degrees looks fairly strong..but I suspect that is partly the stretch. I there any chance you could do a normalized comparison, whereby you apply the exact same histogram stretch to all three images? That should give us a true apples-to-apples comparison. You would just run a screen stretch on one image...probably the 20 degree one to prevent an excessive stretch on the others, then do the histogram stretch, then apply that exact same histogram stretch to the other two, then run ab inspector. That would give us a visual idea of the differences in noise, and the background sky level should drop as well (which is not apparent in your current samples as it appears you did a separate screen stretch for each.)



#12 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 04 March 2015 - 10:04 PM

Ugghh.... I just started an hour long sub on Sh2-216 .... since the moon is still clearing the ridge to the east, let me run one and then I'll do the filter test (or I might hold off till tomorrow - weather will be good then and I can get a couple of this dim target before switching to the still-bright moon for the rest of the night).

 

@Adam - thanks, yes, those were taken at F6.5 with my usual setup (AT65EDQ, Atik One 6.0, Astrodon filters). I did that writeup pretty late last night, I was going to plot those stats but my eyes and tired old body gave out. I figured the raw data was good enough. It looked pretty linear to me eyeballing it ... I will add some more data points and do a filter comparison too.  I honestly expect the Ha and Sii filters to be minimally impacted until the moon's pretty close ...

 

@Jon - Good point Jon, thanks!  Since I've got an hour to kill now  :D  I'll try and get it done soon.  You're right, I did separate stretches on each and should have used the same stretch to make a better comparison.  I'm trying to remember what I did off the top of my head, but I'm pretty sure the image stats were on unstretched data.  Also, the moon's glow was obvious in PhD 20 degrees off, and made guiding a pain at 10degrees (about 1/2 my guider's view was washed out from the moon's glow).  One of my nit's with the AT65 is its dew shield is too short for my tastes, it really doesn't block any off-angle light.  I use a piece of black foam wrapped around it normally for a little extra dew prevention and to block that stray light. If I was using that I'm sure 20 degrees off the moon would look better.



#13 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 04 March 2015 - 10:33 PM

I was afraid of this ... but it makes sense to me based on the raw stats.

 

Because the background is brighter as you get closer to the moon, it takes  a completely different stretch.  So the stretch at 20 degrees off the moon leaves 30 and 45 stretched too little.

 

MoonTestNormalizedStretch-20.PNG

 

 

Choosing the middle one (30 degrees off the moon) leaves 45 under-stretched and 20 over-stretched ...

 

MoonTestNormalizedStretch-30.PNG

 

 

Note - the gradient is correctable in all of them, but with far less work at 30 and 45 degrees off than at 20.  With my setup last night, 10 degrees off the moon was just too close - I had effectively a direct light-path down the tube.  Again, what I found was that for my setup, I can push a lot closer to the full moon and still image with Oiii ... I'd be comfortable imaging bright targets (ie: Planetary Nebulas) 30 degrees from the moon.



#14 JJK

JJK

    Gemini

  • -----
  • Posts: 3462
  • Joined: 27 Apr 2008

Posted 05 March 2015 - 12:16 AM

Here's a shot of IC410 taken in 5 nm Hydrogen-alpha light (Astrodon) taken with a Ceravolo 300 @ f/4.9 + FLI ProLine 16803 (ten 10 min subs, no bias or flat field calibrations applied, no hot or cold pixel removal).   The Moon was pretty close by, but didn't affect the data much.

 

The image file suffers greatly from JPEG compression.

Attached Thumbnails

  • IC410_COS300_f49_PL16803_0123412_10x600s_5nm Ha_m30_AvgDarks_No Clipping_CN.jpg

Edited by JJK, 05 March 2015 - 12:37 AM.


#15 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23352
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 05 March 2015 - 02:28 AM

I was afraid of this ... but it makes sense to me based on the raw stats.
 
Because the background is brighter as you get closer to the moon, it takes  a completely different stretch.  So the stretch at 20 degrees off the moon leaves 30 and 45 stretched too little.

...
 
Note - the gradient is correctable in all of them, but with far less work at 30 and 45 degrees off than at 20.  With my setup last night, 10 degrees off the moon was just too close - I had effectively a direct light-path down the tube.  Again, what I found was that for my setup, I can push a lot closer to the full moon and still image with Oiii ... I'd be comfortable imaging bright targets (ie: Planetary Nebulas) 30 degrees from the moon.


It isn't just about gradients, though. It's about contrast. I think your test showed exactly what I expected...that your background sky level is significantly higher 20 degrees off the moon, which means your losing a lot of the contrast, and thus benefit, that an OIII filter provides. The difference, when you use the same stretch, seems extremely high to me...I expected that you would at least be able to see the 30 degree background sky when basing the stretch off the 20 degree sub...but both 30 and 45 degree subs are much, much darker. That's what you want, right? It means you can, and should, be exposing longer, getting a stronger object signal per sub, than is allowed or possible when imaging 20 degrees off the moon. I would still stick with targets 30-45 degrees away, personally...I wouldn't want to lose my contrast like you are with the 20 degree sub. 


  • Goofi likes this

#16 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 05 March 2015 - 09:45 AM

I thought about last night's post a little more, but will discuss it when I do the test with all three filters ...

 

Jon, you're correct about contrast.  With a brighter background, there's less contrast. Also, those were only 10 minute subs - I can't see taking 10 minute narrowband subs on any target (other than M42).  I picked that target set in part because there were no 'targets' (galaxies or nebulas) in the FoV ... just sparcely populated with stars. The way I wrap my  mind around this, the SNR has contrast built into it ... the signal (what I want) is in relation to noise (which takes into account the sky background).  20 degrees is too bright a background, SNR is too low, for me ... especially going after nebulas.  I could probably work with 30 degrees, especially with brighter targets.

 

 

Since I tend to image the same target over multiple nights, this ties back into the other thread: Should I take all my Oiii on one night and have one gradient? Or should I spread them over multiple nights and let the calibration and integration functions help normalize things (while possibly making more complex gradients to remove)?



#17 terry59

terry59

    Cosmos

  • *****
  • Posts: 8962
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 05 March 2015 - 10:19 AM

I thought about last night's post a little more, but will discuss it when I do the test with all three filters ...

 

Jon, you're correct about contrast.  With a brighter background, there's less contrast. Also, those were only 10 minute subs - I can't see taking 10 minute narrowband subs on any target (other than M42).  I picked that target set in part because there were no 'targets' (galaxies or nebulas) in the FoV ... just sparcely populated with stars. The way I wrap my  mind around this, the SNR has contrast built into it ... the signal (what I want) is in relation to noise (which takes into account the sky background).  20 degrees is too bright a background, SNR is too low, for me ... especially going after nebulas.  I could probably work with 30 degrees, especially with brighter targets.

 

 

Since I tend to image the same target over multiple nights, this ties back into the other thread: Should I take all my Oiii on one night and have one gradient? Or should I spread them over multiple nights and let the calibration and integration functions help normalize things (while possibly making more complex gradients to remove)?

I would look at WB and NB very differently, and I'm only talking here in relation to your situation. With RGB I combine the channels then process. With NB I process each set then blend so I would run one NB filter per session


Edited by terry59, 05 March 2015 - 10:22 AM.


#18 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 05 March 2015 - 10:36 AM

Thanks Terry   :)  ... that's how I process my narrowband - process first, combine later.  With one or two minor exceptions, I run one filter a night ... my processing is part of the reason why.



#19 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 10796
  • Joined: 14 Jan 2013

Posted 05 March 2015 - 07:41 PM

Remember that uniform (or well behaved gradient) signal can be subtracted out. What matters is the remaining noise.

 

Assuming that 100% of the ADU value is due to signal (and not bias offset - a bad assumption), then noise is the SQRT of the ADU value.

20 degrees off: SQRT(1827) = 42.74

40 degrees off SQRT(899) = 29.98

 

Difference = just 12.76 ADU. That may be just 5 or 6 electrons depending on the gain of the camera. In other words, in terms of importance, it is in the same ballpark as read noise.

 

I'm not saying that isn't important. Every little bit matters, especially in the faintest portions of an image.

 

BUT, I am saying that the importance is no where near as much as might be supposed just looking at the ADU values without considering the resultant noise.

 

EDIT: This is why even in broadband imaging, the Moon scares me much less than it does many other people. Back with film, it was much harder to deal with, but that is no longer the case.


Edited by Madratter, 05 March 2015 - 07:46 PM.

  • Goofi likes this

#20 josh smith

josh smith

    Fly Me to the Moon

  • *****
  • Posts: 5324
  • Joined: 13 Oct 2013
  • Loc: St Petersburg, Florida

Posted 05 March 2015 - 07:46 PM

Remember that uniform (or well behaved gradient) signal can be subtracted out. What matters is the remaining noise.

Assuming that 100% of the ADU value is due to signal (and not bias offset - a bad assumption), then noise is the SQRT of the ADU value.
20 degrees off: SQRT(1827) = 42.74
40 degrees off SQRT(899) = 29.98

Difference = just 12.76 ADU. That may be just 5 or 6 electrons depending on the gain of the camera. In other words, in terms of importance, it is in the same ballpark as read noise.

I'm not saying that isn't important. Every little bit matters, especially in the faintest portions of an image.

BUT, I am saying that the importance is no where near as much as might be supposed just looking at the ADU values without considering the resultant noise.


Could not have said it more elegantly myself.

#21 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23352
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 05 March 2015 - 08:16 PM

Since I tend to image the same target over multiple nights, this ties back into the other thread: Should I take all my Oiii on one night and have one gradient? Or should I spread them over multiple nights and let the calibration and integration functions help normalize things (while possibly making more complex gradients to remove)?

 

There is nothing that says you can't DBE your individual channels. You could always just do that, extract the background of OIII, Ha, and SII separately, get the flattest field you can, then combine. 



#22 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23352
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 05 March 2015 - 08:19 PM

 

Remember that uniform (or well behaved gradient) signal can be subtracted out. What matters is the remaining noise.

Assuming that 100% of the ADU value is due to signal (and not bias offset - a bad assumption), then noise is the SQRT of the ADU value.
20 degrees off: SQRT(1827) = 42.74
40 degrees off SQRT(899) = 29.98

Difference = just 12.76 ADU. That may be just 5 or 6 electrons depending on the gain of the camera. In other words, in terms of importance, it is in the same ballpark as read noise.

I'm not saying that isn't important. Every little bit matters, especially in the faintest portions of an image.

BUT, I am saying that the importance is no where near as much as might be supposed just looking at the ADU values without considering the resultant noise.


Could not have said it more elegantly myself.

 

 

Wouldn't the thing that determines whether 12.76 ADU matters or not depend on the strength of your object signal? For an object like M42, that's not likely to be a problem. What about for a fainter target? OIII on, say, Crescent nebula? To me, seems 12.76 ADU would be a rather huge difference with a target like Crescent and similar in OIII... At least, unless you were taking subs so long that you were able to completely swamp read noise with skyfog, which the moon at 20° off axis could well and truly hamper, unless I am mistaken.



#23 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 10796
  • Joined: 14 Jan 2013

Posted 05 March 2015 - 08:54 PM

Yes, as I mentioned, it does matter how strong the signal is that you are trying to get.

 

"I'm not saying that isn't important. Every little bit matters, especially in the faintest portions of an image."

 

But there is a HUGE difference between a delta of 12.76 ADU, and a delta of 927.3 ADU. You can't just look at the raw ADU values. You have to consider the noise difference. In this case, the importance of the Moon is about the same as the importance of the read noise. That is something you probably wouldn't expect unless you did the math.



#24 Goofi

Goofi

    Cosmos

  • *****
  • topic starter
  • Posts: 8137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 05 March 2015 - 08:55 PM

Madratter ... thanks for adding that point. It needed to be mentioned and you did it so clearly!

 

Jon, I do ABE/DBE on each channel separately, and then take a look when I combine to see if anything still remains.

 

A big part of this whole issue (Is the moon going to impact your imaging) will depend on your target. I haven't really brought that out since I felt it was understood.  As a for instance ... I'm shooting Sharpless 2-216 these days and it's taking 60 minute subs just to get a meaningful signal.  Every little bit of improvement I can get on SNR will matter with this target.  But if I were imaging something brighter, the Rosette Nebula for instance, I have more flexibility because the signal is pretty strong.  No, I'm not going to image the Rosette in Oiii with the Moon 15 degrees off, but if it was 20 degrees off I'd probably be willing to image with Ha.

 

Btw, I'm planning on testing Ha, Sii and Oiii tonight - I'll probably just use a target 30 degrees from the moon and 45 degrees altitude for the test.



#25 Footbag

Footbag

    Cosmos

  • *****
  • Posts: 9115
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 06 March 2015 - 09:41 AM

Remember that uniform (or well behaved gradient) signal can be subtracted out. What matters is the remaining noise.

 

Assuming that 100% of the ADU value is due to signal (and not bias offset - a bad assumption), then noise is the SQRT of the ADU value.

20 degrees off: SQRT(1827) = 42.74

40 degrees off SQRT(899) = 29.98

 

Difference = just 12.76 ADU. That may be just 5 or 6 electrons depending on the gain of the camera. In other words, in terms of importance, it is in the same ballpark as read noise.

 

I'm not saying that isn't important. Every little bit matters, especially in the faintest portions of an image.

 

BUT, I am saying that the importance is no where near as much as might be supposed just looking at the ADU values without considering the resultant noise.

 

EDIT: This is why even in broadband imaging, the Moon scares me much less than it does many other people. Back with film, it was much harder to deal with, but that is no longer the case.

 

That's a great point.  You really wouldn't expect that unless you did the math.  Actually, I'm trying to make that jive with my own experiences.  Basically, Moon & 8nm OIII. :thumbsdown:

 

But, and I'm still stuck on the relationship of the background ADU and filter band pass, Goofi is using the OIII. I'll assume OIII is the most susceptible to moonlight. But, at 3nm, it seems very resilient.  Unfortunately, OIII tends to be weak.  So maybe 12 ADU noise contribution could become a noticeable factor.  Take a 5nm filter, and the contribution almost doubles.  8nm and it goes further.  Wideband Blue(45nm)... It does seem like it could become a factor.  

 

I'm also wondering about the fact that that bias can be calibrated out.  Or at least a significant portion of it.  After calibration, though read noise is there to stay.  

 

Removing the bias pedestal should decrease the impact of of the moon, if I'm doing the math correctly.  We should take that into account.  It won't be very significant, but we're talking about a few electrons whether it's noise side or signal.  More accuracy wouldn't be a bad thing

 

This is a great topic.  




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