Back to forums

Everything but the 183

Equipment CMOS
ZWO ASI183/QHY183 •16 replies•144 views
Howard Maron avatar
Howard Maron avatar
Jon (and others)
I don't own the "183". But I thought I might learn something important by posting here.

The rationale for the camera on your FSQ (Jon) is brilliant and I concur completely with the motivations for using this camera on your FSQ setup. I happen to have virtually the identical setup as yours including the Nitecrawler.  In fact, Jon, it was a posting of a photo of your FSQ/NiteCrawler that motivated me to acquire the system. I absolutely love telescope and focuser combo. It is so rock solid and "usable" particularly when paired with a great mount. My mount happens to be a 10Micron2000.

So why am I posting here? First and foremost it is to ask a question which I will eventually get to. But I first just wanted to extend accolades. I find it hard (perhaps emotionally?) to forego the large sensor size the telescope can support. So, on my FSQ I use what most others use with our telescopes: the predictable and "large" IMX455 and the  50x50 massively expensive filters it requires.  As I said, it is a lovely, almost "perfect" combo.

But what my system doesn't deliver is the potential of achieving the resolution sweet spot that the pixels size of the 183 provides. (Not to mention the terrific cost saving and bulk reduction). 

Here is the question: Given that one finds themselves in an area of "typical" seeing, do you still feel that you can benefit from the resolution of the FSQ/183 pairing?  And, if so, why is that? How do you best monitor this? Have you compared your images at this pixel size with your cameras that are using 3.76 microns? Though the question is on the surface very basic, I am sure it could lead to a rather complex answer. 

Thanks for indulging me. I am seriously interested in the answer to my question, both theoretically and practically.  As an aside, I also happen to have a DR350 in Chile and, in that location in particular,  I seriously would have loved to use a camera with 2.0 to 2.4 micron sensor size with that telescope. But, I didn't even consider using such a sensor. Rather, I immediately dismissed them because none felt "worthy" of riding on the back of the DR350 since their sensor size didn't utilize what the scope could deliver.  Now I wonder. 


-H
Jérémie avatar
Jérémie avatar
Hi Howard, 

I can’t answer your question as I believe the choice of camera you make depends on many paremeters and tradeoffs you are willing to make. Plus you have high-end telescopes that I have never used :-)

I just want to add in your ‘equation’ that you if you want to achieve the best resolution possible with your setup, you can partially overcome atmospheric turbulence using short poses techniques, usually applied when capturing planetary data. That’s a lot of work and data when applied to deep sky targets, but you should look at what Exaxe is doing and achieving on this forum (and probably other experts of that techniques) : https://www.astrobin.com/users/exaxe/ 

It allows him to capture details rarely seen with amateur gear. It applies to planetary nebulas (very bright), but he also does some wider field images that are gorgeous (galaxies for example). 

I would say that using this technique with sensors that are small in diagonal, but with relatively small pixel size, you indeed lose some potential of your nice telescope, but depending on the target, the gain in resolution may be worth. Again that’s a tradeoff (and I guess you can’t go to Chili every shooting session to change your camera…).
Helpful Respectful Engaging
Jon Rista avatar
Jon Rista avatar
Hi Howard,

I'm just getting back into the hobby here, after some years away. I am a bit new to ABin's forums, so I only accidentally happened to come across this thread (I don't know if there is any area where you get notified of new threads addressed to you?) 

Anyway, glad I did stumble upon it!

So, simple answer…I use the 183, only because its what I have right now. I would actually prefer to pair my system with an IMX455, in fact more specifically, I intend to pair it with a QHY600 at some point, hopefully soon. I think the larger sensor would indeed make better use of the large, flat, well-corrected image circle the FSQ106 has to offer. The 183 does a good job with the FSQ, but yes, its a much smaller field. 

For a deeper answer… The 2.4 micron pixels of the 183 are amazing, and I think they DO have the potential to improve resolution. That said, my honest opinion is, I'm not quite sure the FSQ106 actually makes the most out of the 183 pixel size. I used to image with a 600mm Canon supertelephoto lens. Despite some issues with the floating image stabilization group in that lens, it is in fact one of the sharpest optics I've ever used. It is more than capable of making the most out of the tiny pixels of the ASI183. 


In contrast, the resolving power of the FSQ106 is not as high. Its spot, especially as you move away from the center field, is higher than the Canon 600mm lens. In all honesty, especially with seeing added to the mix, I am not sure that I've actually been able to make the most of the small 2.4 micron pixels of the 183. At least, not to teh degree I was able to with the Canon lens. I DO think that the 183 might actually make for a good pairing if I added the extender to the FSQ106…when you do higher resolution imaging, I think there are added benefits to oversampling.

That's another point as well…if your goal is to image wider fields, then TBPH I think image scale, and being optimally sampled, is actually less meaningful than the size of the field. The IMX455 would DEFINITELY serve the goals of imaging big, beautiful wide field images with the FSQ106 a lot easier. Jereme makes a good point about short exposures, and that is a great way to improve resolution by cherry picking the best subs. That said, I still don't think you would be able to make the most of the 183's small pixels, as you would be spot size limited with the FSQ before you could, I think. 


SO, if you want to move to an IMX455 camera, I would say you should. I think it would be a better match, unless your goal is to image galaxies. And, if your goal IS to image galaxies, then, you probably want a different scope as well. ;)
Helpful Insightful Respectful Engaging
Howard Maron avatar
Howard Maron avatar
Thanks guys. Just to clarify I use the IMX455 on both the DR350 (the QHY600 UltraS/FLI Centerline) and on the FSQ (Moravian C3). I have used the latter system extensively and love it. The setup in Chile is just now becoming operational so I haven't had the advantage of really studying those images. 

Over time, I have learned a considerable amount of theory from you, Jon Rista, through your postings. So, when I saw that you, Jon, started this thread I decided to toss forward my question to see what I might learn. 

I think the FSQ run native at f/5 is a fantastic "lens". (Clearly, I am not a fan of it when paired with Tak's reducer). So my view is we are  "seeing" limited already even at a pixel size of 3.76 microns. But, is that, in fact, correct? Might (in practice) I be able to observe improvement in detail with a smaller pixel?  That is the question. Jeremie's point that shorter exposures might overcome some of the limitations of seeing just doesn't resonate with me.  Turbulence is just too rapid! Unless the exposure durations are in the sub-seconds category (as in video...as in so called, lucky imaging.....which is simply unfeasible for low light dso's) shorter exposures only overcome mount and tracking limitations, not seeing limitations.

My analysis could be entirely different when shooting from a location in the Atacama where there really can be long intervals of absolutely still atmosphere. But, here in the California desert, I can't imagine I would notice the resolution shift if my FSQ system was changed from what it is currently 1.46 "/px to what it would be with the 183 sensor, 0.93 "/px.

More specifically, the longer FL, DR350 in the Atacama is an entirely different matter. Currently, my pixel ratio with the QHY600 is .47"/px.  Nothing would make me happier to discover that the seeing there might be so good on occasion that I might sense that I am borderline "undersampled".  What a wonderful problem to have!  Well, time will tell.
Helpful Insightful Respectful Engaging
Dark Matters Astrophotography avatar
Dark Matters Astrophotography avatar
Well the 183 has smaller pixels, but it is also the older Sony design that was intended for security cameras, if I'm not mistaken. There's likely to be a big difference in FWC, QE, and other characteristics of the IMX455 that you would be missing compared to the current setup. These newer Sony chips were a huge step forward for CMOS in our hobby. You could look at using an extender on the FSQ and the larger pixels on the IMX455 or similar. That likely results in the best of both worlds.

Bill
Helpful Insightful Respectful Concise
Jon Rista avatar
Jon Rista avatar
Howard Maron:
Might (in practice) I be able to observe improvement in detail with a smaller pixel? That is the question. Jeremie's point that shorter exposures might overcome some of the limitations of seeing just doesn't resonate with me. Turbulence is just too rapid! Unless the exposure durations are in the sub-seconds category (as in video...as in so called, lucky imaging.....which is simply unfeasible for low light dso's) shorter exposures only overcome mount and tracking limitations, not seeing limitations.


I'm on my phone, so I'll respond to this for now. There is lucky imaging, and short exposure imaging. The goal of lucky imaging, using extremely short subsecond exposures is to achieve diffraction limited performance. Resolution over SNR.

Short exposure imaging on the other hand is a real thing. I did a fair amount of this back between 2016-2018. There ARE resolution benefits...IF you cull your subs properly. If you stack every sub, there is no benefit for short exposure imaging. Cull 30%, or more, and you can indeed improve resolution. Not diffraction limited but still improved.

With regards to seeing....think of it as a complex wave. Scintillating would be your limiting high frequency wave. There are other wavelets though, if larger scales. Even in bad seeing (ibeleive me, I know bad seeing...i live directly underneath the turbulent rocky mountains Jetstream), short exposure imaging will capture better quality subs during periods of convergence of the various scales of seeing areund the baseline level of scintillation.
Helpful Insightful Engaging
Jérémie avatar
Jérémie avatar
Just to clarify Howard’s doubts on using lucky imaging for DSO : turbulence can happen fast indeed, but it is mainly ‘random’. By shooting using short exposures such as 5s, 2s, 1s or 500ms, you’ll statistically find periods of several seconds where turbulence was low. By keeping between 10% to 50% of the best frames, you’ll have a virtual good seeing… Of course, it has advantages as well on the guiding / quality of mount side, as these constraints become less stringents.
Some references :
https://www.astrobin.com/groups/92/short-exposures-dso/
https://www.astrobin.com/forum/c/astrophotography/deep-sky/lucky-imaging-for-dso/
https://pixlimit.com/techniques.html
https://www.cedic.at/arc/c11/dwn/CEDIC11_FilippoCiferri.pdf
https://astrobiscuit.com/best-gear-for-lucky-imaging/
Helpful Concise
Howard Maron avatar
Howard Maron avatar
I have to thank you both very much. Jeremie, I will go through those links you provided.  I had no idea anybody had the huevos and the patience to apply this "planetary" astrophotography technique to weak DSOs. I am always impressed by the persistence and the determination of people in this "hobby".  
Thanks again,
H
Well Written Respectful Supportive
Jérémie avatar
Jérémie avatar
Yes, that’s indeed for passionate astrophotographers who wants to capture high resolved details on specific targets (planetary nebulas, some galaxies etc…). It seems it ermerged a few years back in a research group at Cambridge, with tests on the Palomar telescope.

To be honest have been looking at it for 2 years but did not give it a go due to ‘lazyness’ : I mainly shoot with an Asiair, and this technique would require me to use my laptop outside at night, which I am quite reluctant to do…  and indeed, it requires some processing time.

Here is another interesting link I forgot to share, but unfortunately it is in French (the automatic translation on Youtube is OK-ish, though) : Siril now allows one to manage directly  the stacking of SER files for DSO lucky imaging. Here is the video / tutorial : https://www.youtube.com/watch?v=es5fTWWnqT8 , with explanations from Stephane Gonzales (Exaxe on astrobin) who has a lot of practice with this technique.

I’ll have to find some momentum and passion again to give it a try :-)
Helpful Engaging
Jon Rista avatar
Jon Rista avatar
Howard Maron:
I have to thank you both very much. Jeremie, I will go through those links you provided.  I had no idea anybody had the huevos and the patience to apply this "planetary" astrophotography technique to weak DSOs. I am always impressed by the persistence and the determination of people in this "hobby".  
Thanks again,
H

With modern sensors, especially with HCG modes where you might have 1.2-1.5e- read noise, people are already not really NEEDING long exposures. Sometimes, you can get away with 30 seconds anyway. FWIW, that is a "short" exposure, and it CAN get you better resolution with effective culling of the softest subs. You don't have to get down to 1, 5, 10 seconds to benefit from short exposures. The key, really, is in the culling, and capturing a LOT of subs. Hundreds, or more. One you have say 500 subs, 1000 subs...you can often then cull the CRAP out of those sets, keeping the best 50-60% or so. You can then still get a good quality DSO (since you should be swamping read noise well enough, even though its only 30s, you also only have 1-1.5e- read noise). If you want the best of both worlds, you can acquire even more subs, cull aggressively, and you could still have 1000+ subs, which is 8-9 hours of data. Do that per channel, you have yourself a real nice DSO image, with sharp details.
Helpful
Howard Maron avatar
Howard Maron avatar
Thanks Jon. Yet, there are many inconveniences of extremely short exposures. But I have to put that in perspective: It is the software that is suffering the inconvenience, not the user……it's doing all the work. (my attempt at humor). Technically even the culling could be done by software.  When all is said and done, one thing is undeniable: there are far greater opportunities to acquire great images on less than "perfect" equipment.  This is great news for the vast majority of people who otherwise could not afford this hobby.

Thanks for sharing your ideas/thoughts.
h
Respectful Supportive
Jon Rista avatar
Jon Rista avatar
Yeah…interstingly, this is kind of a snagging point with CMOS technology. Smaller pixels, smaller wells. A KAF-16803 has over 100ke- in terms of REAL physical well depth, and ~1e- read noise per square micron. For a CCD from its timeperiod, it was a phenomenal sensor. It ALLOWED very long exposures with the gain the manufacturers configure to achieve that 9e- read noise per 9 micron pixel with 100ke- FWC. 

In contrast, most CMOS cameras have less than 50ke- FWC at the HCG modes most people use. Sometimes a lot less. If you are using a CMOS camera at the HCG cutover gain, then you actually are going to be using the dynamic range of the sensor poorly, unless you use short exposures. Clip more stars, all that. 

The "advice" all over is to just use the HCG mode. But then people are using several minute exposures and the images have mad clipped stars, very limited dynamic range, etc.

IF you really want longer exposures with a CMOS camera, then lower gains is generally the way to do it. You won't be able to optimize your stack nearly as much, but, you can generally tune a CMOS camera to support any exposure and/or any sub count for your stacks that you want to. Even with lower gain settings, you can still manage fairly shortish exposures as well (minute or two), and still benefit from some lighter culling, to optimize the stack. 

This is one of the nice things about CMOS cameras, at least as they are designed thus far (QSI's new cameras might change this a bit), they are extremely flexible, adaptable cameras to a wide variety of use cases. Something CCDs weren't all that good at. 

FWIW, so many of these sensors are oversampling a lot of fields anyway. Even though its done in software, 2x2 binning could save you on disk space and processing time, without losing much resolution, allowing you to capture, cull and process 1000+ subs in a lot less time than at native resolution. In fact, at moderate to longer focal lengths, you probably wouldn't lose any detail at all in the end, since a 3.75 micron pixel or smaller will oversample most stars under "average 2" seeing" by nearly 3x at just a little over 1m focal length!!
Helpful Insightful
Howard Maron avatar
Howard Maron avatar
Jon,
Thanks yet again. I love the way you think.
I am excited about the fact my DR350 is going to go live this week at Obstech. I should have been there for setup but I farmed it out to Matt Dieterrich who I trust greatly. (Side note: I just hate the idea that I delegated that process out.....it's not the way I like to do things. But Chile is frickin' far and the trip isn't easy for me.  At least I had the DR system up and running here first for about a year. But I still feel doing it this way is way too hands off for my taste).
Anyway, I am super excited to actually "see" the kind of seeing I hope to experience down there. 

I do have one hopefully quick question: If one wanted to automate culling what metric(s), what parameter(s), which index(indices) would you follow/measure? Similarly, (and perhaps this is an odd question) what ROI (how small a field and from where on the image would you derive the parameter(s)?  

I hate referring to what we do as a hobby because, frankly, the term almost denigrates the actual science that we deploy. Nevertheless, this is one great hobby. I even used to love it back in the film days when what we produced was frankly pure shlock. Today, the options are boundless. And, thankfully, so much more fun. 

One last aside. My favorite images from yesteryear that I used to cherish were those using the FSQ106N with a 11000m. They were grossly undersampled! Yet, I admired the beauty of some of those "masterpieces" the most.

-H
Jon Rista avatar
Jon Rista avatar
Howard Maron:
Jon,
Thanks yet again. I love the way you think.
I am excited about the fact my DR350 is going to go live this week at Obstech. I should have been there for setup but I farmed it out to Matt Dieterrich who I trust greatly. (Side note: I just hate the idea that I delegated that process out.....it's not the way I like to do things. But Chile is frickin' far and the trip isn't easy for me.  At least I had the DR system up and running here first for about a year. But I still feel doing it this way is way too hands off for my taste).
Anyway, I am super excited to actually "see" the kind of seeing I hope to experience down there. 

I do have one hopefully quick question: If one wanted to automate culling what metric(s), what parameter(s), which index(indices) would you follow/measure? Similarly, (and perhaps this is an odd question) what ROI (how small a field and from where on the image would you derive the parameter(s)?  

I hate referring to what we do as a hobby because, frankly, the term almost denigrates the actual science that we deploy. Nevertheless, this is one great hobby. I even used to love it back in the film days when what we produced was frankly pure shlock. Today, the options are boundless. And, thankfully, so much more fun. 

One last aside. My favorite images from yesteryear that I used to cherish were those using the FSQ106N with a 11000m. They were grossly undersampled! Yet, I admired the beauty of some of those "masterpieces" the most.

-H

You would use measurements and statistics for the things that matter to you, to automate culling. If your goal is to maximize detail, key metric would probably be FWHM. Fit a PSF to the stars of your frames, sort the results, reject anything beyond a certain range of criteria. Or maybe keep the best X%. If you want the best star quality and color, FWHM would probably become a secondary factor, and saturation ratio might become more important. If you care about background SNR most, then that would be your criteria. 

As for where within the field...as long as your field is flat and generally has good quality throughout, then all of it. Averaging criteria from throughout the field might help normalize the results a bit. If you have stronger vignetting, for example, you might want to over-weight SNR from the periphery of the field where its more shaded, and keep the frames that have better weighted SNR in the periphery. The center would always be pretty good, the periphery will suffer faster if something changes (sky brightness, passing thin clouds, etc.) If your scope starts to lose star quality as you approach the corners and your main criteria is FWHM, you might want to use a more constrained region, especially if your normal practice is to crop out the corners anyway. 

I also don't really like using the term hobby...but, I honestly don't know what else to use... Is there a better term? Endeavor doesn't quite cut it, I don't think. Not sure what else we could use. But I agree!  

I hear you too, about the big pixel cameras and the FSQ106. I always loved Rogelio Bernal Andreo's images, KAF-16803+FSQ106, times TWO! And he usually used a dark site that was around 21.6mag/sq" IIRC. He did not always produce those glass smooth backgrounds that many of those big pixel imagers were able to, but...he pulled out details fainter than almost anyone else. A bit noisy, but incredible stuff. He is also the only guys I know of, to get lots of color in his IFN images. You would think IFN was gray due to how many images depict it that way these days. RBA was doing IFN imaging long before most, and his depict it as dark and mid browns, tans, with yellow and blue reflections, and a lot more structure than we often see today. I miss seeing images THAT deep, of things rarely seen. We do see a lot more IFN these days, and often with smoother noise, but rarely with the incredible depth and detail that RBA achieved.
Helpful Insightful
Howard Maron avatar
Howard Maron avatar
I have really enjoyed this discussion. The sharing of ideas (and, yes, friendship) with some very extraordinary people makes this craft particularly special.


("Craft" ain't it either, but it's getting closer. …… hmmmm, "pursuit"???  While good,  is merely sidestepping)


On Andreo's images, what then do you feel was his secret sauce if one were to replicate his images as a goal?
Jon Rista avatar
Jon Rista avatar
Howard Maron:
I have really enjoyed this discussion. The sharing of ideas (and, yes, friendship) with some very extraordinary people makes this craft particularly special.


("Craft" ain't it either, but it's getting closer. ...... hmmmm, "pursuit"???  While good,  is merely sidestepping)


On Andreo's images, what then do you feel was his secret sauce if one were to replicate his images as a goal?

RBA used a dual FSQ106+KAF-16803 imaging rig, under skies that are about as dark as you can find on earth. He acquired tons of integration time through those two scopes. Before he redesigned his web site, he used to have all the acquisition details for each of his images.

His key, was to just stick with the fundamentals. Get LOTS of data, as much as possible. Go DEEP on all your exposures, RGB, L, NB, whatever. USE dark skies, don't mess around with light pollution.

RBA was doing his thing with a lot of his most spectacular images 10+ years ago, and the processing tools and advice were not nearly as extensive or widespread (or diverse, good or bad) as thy are today. He developed many of his own techniques for pulling out extremely faint details. When he did, however, he had deep enough and enough data, particularly on the color side of things, that he could pull pretty rich color out of things we today, generally render colorless:

https://www.deepskycolors.com/product/north-celestial-pole/

https://www.deepskycolors.com/product/ifn-and-the-galaxies/

https://www.deepskycolors.com/product/integrated-flux-world/

Today, people not only render IFN as gray, but actually talk about "why" it "is" gray, how its the integrated starlight of all the light in the galaxy, blah blah. Its all a bunch of bunk. Look at the images above. IFN is anything but gray!! It is very rare these days, that we see anyone render deep details with color. Sometimes you might see an image with IFN that renders some parts of it very slightly tan. 

We've greatly weakened color these days, but radically over-emphasising the importance of the L channel. Further, we are weakening EVERYTHING because people openly talk about how AI processing tools eliminate the NEED to even get more than just a little bit of data. That all just belies a grave and deep misunderstanding about what astrophotography is, and what our cameras are capturing. 

 I think, if we are to take a cue from RBA, its to stick to the fundamentals. Get a lot of deep, strong signal...with your cameras....under dark skies, if you can.
Helpful Insightful Engaging
Howard Maron avatar
Howard Maron avatar
Thank you.
These are such good points. I love it when credible thinkers just call out the BS.  There is considerable groupthink and some of it is just noise that gets very loud. 

I don't know why I am this story comes to mind. In a sense, it is completely off-topic. Anyway………I once went on a photo (wilderness) shoot with a group of friends and acquaintances.  Most were very well-healed Microsoft folks. They were wonderful and experienced amateur photographers with the finest of equipment and all were technically brilliant. We all shot in RAW. Who doesn't? Well, invited on our trip was also a renowned professional photographer. He always shot in JPEG!  Principally, he said, it was for ease and practicality.  We all produced great images. Yet, his were in another realm completely. Ours were wonderful, yet indistinguishable. His were stunning and notably "his".