M 16 Ha

[Rodd]

My first attempt at Ha with the ASI 1600.  The target rises very late for me this time of year (2:00am at 25 degrees).  Unfortunately it never gets higher than about 35 degrees, and it is located in the worst part of my sky--within the light dome of a nearby city.  Only a few stars can typically be seen with the unaided eye in this part of my sky.  So LRGB is probably not likely to produce decent results.  

The image contains 28 5min subs.  collected with TOA 130. No calibration has been performed (I have not had time to collect the necessary frames).  The only obvious artifact from the lack of calibration was a small lobe of amp glow in the upper right hand corner that I cloned stamped out.  So, in case anyone notices, this is why the stars in the upper right hand area are not correct.  The background may be a bit clipped as well.  After calibration these issues will be corrected.  More data may not be necessary.  I compared this image with an image composed of just 7 5 min subs and the difference was pretty much limited to a bit of grain in the less bright regions of the FOV.

I have included a 20 sec sub and a 5 min sub.  Its amazing at how much Ha signal is present in these short subs--especially the 20 sec one.  I think that 100-200 20sec subs would make a respectable Ha stack. 

 

5 min sub

20 sec sub

[wimvb]

Lovely. What brand/type Ha filter do you use? The narrower the transmission peak, the better it will be at rejecting light pollution.

[Rodd]

17 minutes ago, wimvb said:

Lovely. What brand/type Ha filter do you use? The narrower the transmission peak, the better it will be at rejecting light pollution.

Thanks Wim--I use 3nm Astrodon filters.  Wow--on this screen I can really see the clone stamp area--that looks bad!  On my other screen I could not really see it.  Here's the  image with the amp glow--I think its better than the clone stamp.  

[Allinthehead]

Nice Rodd.

[Rodd]

2 minutes ago, Allinthehead said:

Nice Rodd.

Thanks Richard.  

Rodd

[Barry-Wilson]

Very interesting Rodd, and a good foundation for an Ha mono building there.

The pace of development of the CMOS cameras is quite breathtaking . . . almost seems like everyone has one!

[Allinthehead]

12 minutes ago, Barry-Wilson said:

The oace of development of the CMOS cameras is quite breathtaking . . . almost seems like everyone has one!

Some of the images produced by them has really come on recently as people are getting to grips with them and figuring out the best settings for their conditions. Like this. https://www.astrobin.com/full/288909/C/ or https://www.astrobin.com/338105/?nc=user

[Rodd]

24 minutes ago, Barry-Wilson said:

Very interesting Rodd, and a good foundation for an Ha mono building there.

The oace of development of the CMOS cameras is quite breathtaking . . . almost seems like everyone has one!

Thanks Barry--I will calibrate tonight and collect some OIII to see if I can at least get a bi-color.  Not sure if its the sensor or conditions, but my LRGB data is looking pretty frightful.  I had gotten to the point where I could render an LRGB image with a modicum of decency--but it seems I am back to my old self with this camera and Broadband.  Narrow band has always been easier for me

[pipnina]

Very nice image

What's your bortle/SQM? (roughly)

[Rodd]

1 minute ago, pipnina said:

Very nice image

What's your bortle/SQM? (roughly)

Accoring to the clearoutside website, my estimated sky quality is 19.5 (abutted by surrounded by 18.5), and my Bortle is at best a 5--but more closely reflected in 6 for all but a very small patch of sky.

[pipnina]

37 minutes ago, Rodd said:

Accoring to the clearoutside website, my estimated sky quality is 19.5 (abutted by surrounded by 18.5), and my Bortle is at best a 5--but more closely reflected in 6 for all but a very small patch of sky.

That sounds pretty similar to where I am. Except my dark and bright patches are the other way around (the light dome for me is north and I get somewhat better southern sky.)

Gives me some hope if I eventually get into telescopic AP.

[Rodd]

Well--the evening forecasted to be the best was the worst.  I was, unable to image--but I was able to collect calibration frames for M16.  Here is the fully calibrated image--what a difference.  The effects of calibration are strikingly visible in the individual subs.  Amp glow is gone.  This makes a halfway decent stand alone mono image--with only 2.5 hours of data.  I wonder if adding more data--say 5-6 hours worth, will increase dynamic range, or have I reached the point of diminishing returns.

 

 

[Allinthehead]

That's great Rodd. Adding 5-6 hours more will make a huge difference. I did the horsehead in ha with 60 second subs for 12 hrs and every couple of hours i added was significant. 720 frames, my record is over 1k. 

[Rodd]

2 minutes ago, Allinthehead said:

That's great Rodd. Adding 5-6 hours more will make a huge difference. I did the horsehead in ha with 60 second subs for 12 hrs and every couple of hours i added was significant. 720 frames, my record is over 1k. 

Thanks Richard.  So maybe 5min subs are not needed?  1,000 seems a bit much from a logistics standpoint to me.  I am coming to terms with the different style that is CMOS (many shorter subs instead of 20 or 30 long ones).  But at some point it defeats the idea for me...1,000 subs is definitely beyond the pail.  Maybe 1,000 total--but if its 1,000 per channel, I think I will stick to CCD .  15 30min subs seems so much simpler.  So far, I am super impressed with the Ha sensitivity.  LRGB, on the other hand, is testing my patience.

Rodd

[wimvb]

1 hour ago, Rodd said:

I wonder if adding more data--say 5-6 hours worth, will increase dynamic range

Yes, but you have a great image already. But unfortunately, dynamic range increases with number of subs, not total integration time.

1 hour ago, Rodd said:

or have I reached the point of diminishing returns

No, never. Add more if you can. More data will also decrease the noise left over after amp glow removal. Amp glow shows itself more at longer sub exposure time. That's why it may be better to shoot more and shorter subs. Just make sure that each sub is well above the read noise floor. If you start to see the read pattern of the camera in each sub, you should increase the exposure time. At least, that's my experience with the ASI174MM.

[Filroden]

That's a fine mono Ha result.

Like you Rodd, I find the 1600 great in Ha. I use 4 min subs and cannot complain. The camera really is sensitive. I can do automatic focusing in Ha using 8 second subs (albeit using 2x binning) and most targets can be faintly seen in the stretched result! I'm also frustrated by RGB. The gradient in each channel seems to be very different and irregular. When combined it leaves behind a terrible background. I still need to really learn how to use background extraction.

[Rodd]

3 hours ago, wimvb said:

No, never.

I guess what I am suggesting is that at some point, to see the effects of more data, you have to essentially double what you have.  Adding 20 subs when you already have 200 will not do very much.  Doubling 200 is asking for allot! I am only at 29--which is not very many--but the difference between 7 and 29 was not as significant as I expected ( the quality of 7 was way beyond what I expected). I am sure that if I collect 29 more--the difference will be very small (possibly worse depending on the quality of the new subs).  The details won't get sharper, but noise will be reduced.  Once the noise is low (not much noise now), more subs won't do much except allow me to remove some subs with high FWHM or eccentricity or whatever.  With Ha it seems like I don't need very many more (for this target).  With broadband, on the other hand (for any target) it doesn't matter how many I collect, it doesn't help.  I run into this with CCD as well--I'll add 10 more 30 min subs to the 12 I already have and see minimal difference.  Which brings me to a related issue...some people can make a great image with a remarkably small amount of data.  

Rodd

[Rodd]

2 hours ago, Filroden said:

That's a fine mono Ha result.

Like you Rodd, I find the 1600 great in Ha. I use 4 min subs and cannot complain. The camera really is sensitive. I can do automatic focusing in Ha using 8 second subs (albeit using 2x binning) and most targets can be faintly seen in the stretched result! I'm also frustrated by RGB. The gradient in each channel seems to be very different and irregular. When combined it leaves behind a terrible background. I still need to really learn how to use background extraction.

I am afraid the DBE does not help.  I find it ruins more images then it helps.  LRGB that is--for narrowband it usually does Improve the image--though gradients are rare for me in narrowband.  Is this camera good in OIII and SII as well?  The way its going now, I am seriously thinking of limiting this camera to narrowband.  The broadband results have just been terrible.  It's not so much gradients for me, but rather fog, or structures in the background that look like IFN where there is no IFN.  But the edges of the image are dark like they are supposed to be.  I am not using a OAG so there can be nothing interfering with the sensor.   

Rodd

 

[Filroden]

When I say background and gradients I don't mean simple linear gradients across the image but quite complex gradients that still seem "regular" but polynomial in definition rather than linear. I always suspected it could be complex due to reflections, e.g. the fences around the garden reflect the street lights so you get second order gradients from multiple directions. I like your analogy of IFN that isn't really there. Or its just my processing or calibration!

I still don't have OIII or SII filters but I cannot see why good narrowband filters at any frequency would not be similar to Ha. SII, being deep into red should be similar. OIII, where the worst of the light pollution peaks, could be more affected but nowhere near a much as RGB. Ha can be taken under moonlight, OIII not so much so I think the same might apply for bad light pollution.

[Filroden]

Here's an example of the background stretched about as far as I can take it (the top 80% of the histogram was fixed and only the lower 20% was stretched). I can see almost concentric patterns in red, green and blue. I'm now wondering if it's my flats? Are you seeing anything similar? (BTW, this is an LRGB image)

 

[glowingturnip]

nice pic.

It tickled me, because I have almost the exact same picture:

Eagle Nebula (M16) in Ha by Stuart Goodwin, on Flickr

 

I'm planning to add some OIII to mine this summer, see if I can make a decent bicolour

[Rodd]

6 minutes ago, Filroden said:

Are you seeing anything similar? (BTW, this is an LRGB image)

Well--hard to say--I never inverted the image and looked at it this way.  DBE does help with chromatic gradients.  You can test it by running it without impacting the image and it will generate a background map, including chroma gradients.  I don't think its your flats.  If they remove all the dust bunnies and donuts,  and do not create visible artifacts, tehy probably are working fine. I get color gradients in LRGB all the time.  I have always thought they were due to LP, or Moonlight. I find the color gradients the easiest to deal with--if DBE doesn't get rid of them, I mask off the high signal areas and reduce saturation using curves.  I know these are band aides and not really correct--but I have no other option.  I will collect 15 hours of data on an image and it will look like crap.  Someone else will make a stupendously amazing image of the same target with 3 hours.  I just don't get it.  Its easy to blame it on th sky--but the galaxy on Astrobin linked in one of the comments above in this thread was collected under light polluted skies with poor conditions.  

The "structural" problems I find more difficult to deal with --the fake IFN for the lack of a better description.  It may very well be the sky and the problems arise when I over stretch.  Every so often an LRGB image will come out free of these issues and I feel blessed.  That's why I did allot of narrowband work, I don't have such issues.....but I was finding the narrowband work not as respected as the LRGB work, so I decided to get better at LRGB.  The only real way to tell is to image from a truly dark sky site, which I have yet to do--but it is something I really want to do,

Rodd

[Rodd]

9 minutes ago, glowingturnip said:

I have almost the exact same picture:

A bit smaller FOV--but same orientation--almost exact.  What scope/camera did you use and how many subs?

Rodd

[wimvb]

2 hours ago, Rodd said:

I guess what I am suggesting is that at some point, to see the effects of more data, you have to essentially double what you have.  Adding 20 subs when you already have 200 will not do very much.  Doubling 200 is asking for allot! I am only at 29--which is not very many

It all depends on the subframe exposure time of course. With cooled cmos, you can either use many (hundreds+) short exposures at high gain or fewer (tens+) long exposures at low gain. There is an article/blog post or forum thread on the ZWO site about this. At low gain, you have more dynamic range, but also higher read noise. At high gain you have lower dynamic range and lower read noise. Dynamic range can only be gained by increasing the number of subs, not by increasing the exposure time (at constant camera gain). I believe you need to double the number of subs to gain one stop in range.

The final result will also depend on how much light pollution you have. Light pollution introduces sky noise. To compensate for this, you need to increase the total integration time. Again, doubling the integration time for each gain of one in magnitude, I believe.

That's why it's impossible to have general recommendation for "point of diminishing returns" or "optimum exposure time". In the end it's a personal judgement (and probably available clear-sky time) which determines when enough is enough. So, when you say you can't see much difference when adding data, you've probably reached your limit.

Btw, one way to keep "poor" data in the stack is to give each sub a weight factor. I've started doing this in SubframeSelector in PixInsight, and I have the impression that it increases the final image quality. My weight factor is based on snr, fwhm and eccentricity. It allows me to keep subs that weren't very good, but probably too good to discard.

[Rodd]

10 minutes ago, wimvb said:

It all depends on the subframe exposure time of course. With cooled cmos, you can either use many (hundreds+) short exposures at high gain or fewer (tens+) long exposures at low gain. There is an article/blog post or forum thread on the ZWO site about this. At low gain, you have more dynamic range, but also higher read noise. At high gain you have lower dynamic range and lower read noise. Dynamic range can only be gained by increasing the number of subs, not by increasing the exposure time (at constant camera gain). I believe you need to double the number of subs to gain one stop in range.

The final result will also depend on how much light pollution you have. Light pollution introduces sky noise. To compensate for this, you need to increase the total integration time. Again, doubling the integration time for each gain of one in magnitude, I believe.

That's why it's impossible to have general recommendation for "point of diminishing returns" or "optimum exposure time". In the end it's a personal judgement (and probably available clear-sky time) which determines when enough is enough. So, when you say you can't see much difference when adding data, you've probably reached your limit.

Btw, one way to keep "poor" data in the stack is to give each sub a weight factor. I've started doing this in SubframeSelector in PixInsight, and I have the impression that it increases the final image quality. My weight factor is based on snr, fwhm and eccentricity. It allows me to keep subs that weren't very good, but probably too good to discard.

I pretty much do the same--though less formally.  I look at those characteristics and if any are way outside what I want, I don't use the sub.  Then again, I have tried comparing using all subs and only the best, and the difference is very small--sometimes unnoticeable unless you zoom to 4-5:1.