What conclusions can one make looking at the Nina imaging tab [Deep Sky] Acquisition techniques · Anderl · ... · 14 · 410 · 1

Anderl 4.52
...
· 
·  Share link
good night and clear skies fellow astrophotographers, 

my gear is doing its work rn and because I have nothing else to do I asked myself what one can conclude looking at the following. 


Bildschirmfoto 2025-03-20 um 22.02.56.png


chatgpt told me following 


| Parameter                     | Value                                   |
|-------------------------------|-----------------------------------------|
| Telescope                     | Esprit 120 ED, 840 mm focal length, f/7 |
| Camera                        | IMX571 Mono (3.76 µm pixel size)        |
| Pixel scale                   | ~0.92"/px (optimal)                     |
| Seeing                        | 1.7–1.9" (excellent)                    |
| Guiding RMS (total)           | 0.30" (excellent)                       |
| HFR (Image sharpness)         | ~1.8 px (very good)                     |
| Sky quality (Bortle scale)    | ~Bortle 4 (very good conditions)        |
| Recommended exposure (Lum)    | Currently 200 s, optimal ~300–400 s possible |
| Strehl (estimated, Esprit 120)| >0.95 (excellent)                       |


this makes me wonder: 

are things like optical quality, seeing, sky quality etc. things one (or ai) can conclude given the above information?

cs
anderl
Like
Frank777 7.94
...
· 
·  Share link
What information did you feed ChatGPT? How did you do that?
Like
jimmythechicken 19.44
...
· 
·  6 likes
·  Share link
Anderl:
this makes me wonder: 

are things like optical quality, seeing, sky quality etc. things one (or ai) can conclude given the above information?

No. Absolutely not. It has no idea what your seeing is, it has no idea whether your HFR is good or bad (this is system dependent), it misidentified your guide error, it has no idea what Bortle scale you are imaging in, it has no idea how long of exposures you can take, it definitely has no idea what your Strehl is, it doesn't know anything about so called "optimal" pixel scale, and without prior information it has no idea what telescope you're using. LLMs are frankly useless for this kind of stuff and accepting anything they will say as truth is a recipe for confusion when its inevitably dead wrong. 

LLMs (ChatGPT, etc.) are not astrophotographers, they do not know anything about astrophotography. They will regurgitate the mean of the information commonly available on the internet blindly and indiscriminately. There is zero point asking it technical questions or about technical concepts for which it has no real knowledge.
Like
WhooptieDo 10.40
...
· 
·  4 likes
·  Share link
ChatGPT knows absolutely nothing about astrophotography.   Period.
Like
AstroRBA 4.98
...
· 
·  1 like
·  Share link
Anderl:
are things like optical quality, seeing, sky quality etc. things one (or ai) can conclude given the above information?


Could it have picked up your location from your PC and concluded Bortle 4 skies ? and by the same method, found the current seeing conditions ? Could it have guessed the camera by pixel size and then mono by noticing the luminace filter ? - don't know how it got the scope correct?
Like
jmdl101 1.81
...
· 
·  3 likes
·  Share link
Looking at your Nina screenshot makes me conclude that I am jealous of how nice everything is looking, haha. I would love to have hfr values below 2 and guiding below 0.5" consistently. With only a few hundred pixels at maximum, I do agree with the Ai in increasing to 300s, I typically try to stay below a few thousand at max value on luminance.
Edited ...
Like
warble_master 12.34
...
· 
·  2 likes
·  Share link
LLMs like ChatGPT are token generators, they do not possess understanding of the subject matter.  They can speak authoritatively while speaking absolute falsehoods.  I like using them to draft letters and such, but they are useless for understanding NINA's imaging screen.
Like
Anderl 4.52
Topic starter
...
· 
·  Share link
Frank Alvaro:
What information did you feed ChatGPT? How did you do that?


Charles Hagen:
Anderl:
this makes me wonder: 

are things like optical quality, seeing, sky quality etc. things one (or ai) can conclude given the above information?

No. Absolutely not. It has no idea what your seeing is, it has no idea whether your HFR is good or bad (this is system dependent), it misidentified your guide error, it has no idea what Bortle scale you are imaging in, it has no idea how long of exposures you can take, it definitely has no idea what your Strehl is, it doesn't know anything about so called "optimal" pixel scale, and without prior information it has no idea what telescope you're using. LLMs are frankly useless for this kind of stuff and accepting anything they will say as truth is a recipe for confusion when its inevitably dead wrong. 

LLMs (ChatGPT, etc.) are not astrophotographers, they do not know anything about astrophotography. They will regurgitate the mean of the information commonly available on the internet blindly and indiscriminately. There is zero point asking it technical questions or about technical concepts for which it has no real knowledge.


Anderl:
are things like optical quality, seeing, sky quality etc. things one (or ai) can conclude given the above information?


Could it have picked up your location from your PC and concluded Bortle 4 skies ? and by the same method, found the current seeing conditions ? Could it have guessed the camera by pixel size and then mono by noticing the luminace filter ? - don't know how it got the scope correct?

Hey guys,

nina knows about my gear because my prompt was „i am using an esprit 120 together with a imx571m camera. What things can you tell me about my astrophotography/rate the quality of imaging?“ the nina screenshot was uploaded together with this Message.
Like
ScottBadger 7.63
...
· 
·  1 like
·  Share link
Charles Hagen:
LLMs (ChatGPT, etc.) are not astrophotographers, they do not know anything about astrophotography. They will regurgitate the mean of the information commonly available on the internet blindly and indiscriminately. There is zero point asking it technical questions or about technical concepts for which it has no real knowledge.

Given the VERY wide range of responses, often conflicting, any AP question to real people gets, isn't a 'mean' of that information possibly useful? LLM's can be wrong, but I'm not sure any randomly chosen answer from the community has a higher likelihood of being right.... As a quick test, I asked CGPT for acquisition suggestions given my location and gear (response below). I give it a B. It didn't recommend binning, and at my image scale and seeing that would be pretty standard, M42 is only available for about a half hour between dark and 30deg and M57 doesn't rise above 30deg until 2:30am, so the target list isn't great, and the fifth suggestion for exposure balance is muddy. On the other hand, the most useful part of its answer, the exposure length suggestions, is pretty spot on. So, not the best response possible, but I wouldn't call it totally bereft of any AP insight.

So, still buyer beware, but same as seeking answers anywhere else.

Cheers,
Scott



Below is a set of suggestions tailored to your equipment and local conditions. Of course, each target can behave differently, so treat these as starting points to be refined by your own tests.Recommended DSOs

Given Jackson’s dark (Bortle 3) skies, your moderate aperture, and 3″ average seeing, you have a wide range of options. Consider targets such as:
    [*]
    M42 (Orion Nebula):
    A bright, well‐defined nebula that works well with LRGB. Its high surface brightness may even allow for slightly shorter exposures.
    [*]
    M57 (Ring Nebula) or M27 (Dumbbell Nebula):
    Both planetary nebulae offer fine structure. They’re forgiving on a 9.25″ scope and provide a good testbed for color balance in LRGB.
    [*]
    M51 (Whirlpool Galaxy):
    A fainter galaxy that benefits from longer integrations; its spiral structure can be teased out by stacking many subframes.



Other targets—such as open clusters (e.g., the Beehive) or even portions of larger nebulae like the Rosette—can also be rewarding, but the above are excellent starting points given your setup.Suggested Exposure Lengths

Your exposure times will ultimately depend on your camera’s dynamic range and the brightness of the target. As a general guideline:
    [*]
    For Luminance (L) Frames:
    300 seconds per subframe is a good starting point.
    • For very bright targets (like M42), consider trying 180–240 sec subframes to avoid saturation.
    [*]
    For Color (R, G, B) Frames:
    300 seconds per subframe works well in many cases, providing a balanced SNR across channels.
    • If you’re imaging a fainter galaxy (like M51), you might extend these to 300–600 seconds per subframe—but always check that stars remain tight given your 3″ seeing.



Keep in mind that these are per-subframe times. You’ll likely want to acquire a series (say, 15–30 subframes per filter) to later stack and increase your overall signal-to-noise ratio.Additional Acquisition Techniques

To make the most of your imaging session:
    [*]
    Calibration Frames:
    Darks, flats, and bias frames are essential to correct for sensor noise, vignetting, and other artifacts.
    [*]
    Dithering:
    • Introduce small offsets between exposures. Dithering helps reduce fixed-pattern noise and improves the final stacked image.
    [*]
    Accurate Focusing:
    • Use a Bahtinov mask or your camera’s live view to ensure sharp stars; even small focus errors can be critical when stacking many subframes.
    [*]
    Autoguiding:
    • Even though the iOptron CEM70 is capable, autoguiding can further ensure that your exposures (especially those longer than 5 minutes) remain perfectly tracked.
    [*]
    Experiment with Exposure Balance:
    • If you find that your luminance frames are giving you too much detail compared to the RGB channels (or vice versa), adjust the exposure times to balance the dynamic range. Some imagers even opt for a slightly longer integration on the luminance channel for extra detail, while keeping color exposures in check.
    [*]
    Software Considerations:
    • Use stacking software (such as DeepSkyStacker or PixInsight) to align and combine your subframes. Techniques like “drizzle” can help improve resolution in post-processing.



These guidelines should get you started on a productive night of imaging. Remember that the key is to experiment a bit—the optimal settings can vary by target and by the specific sky conditions on any given night. Happy imaging!
Edited ...
Like
KGoodwin 4.71
...
· 
·  1 like
·  Share link
If you are well sampled or over sampled, and your guiding is impeccable, and your optics are excellent, and your focus is perfect, then you can get a pretty good idea of seeing from the FWHM of your images.  Star count can give you a pretty good idea of transparency relative to other times you've shot the same object, but only if the moon is down and there isn't a lot of light pollution influencing the background.  Point being there are a lot of ifs to conclude much of anything objective from what your subs are telling you.
Like
Gondola 8.11
...
· 
·  1 like
·  Share link
I'm curious, what sort of information were you looking for?
Like
jimmythechicken 19.44
...
· 
·  1 like
·  Share link
Scott Badger:

I think you have just proven my point. Even with probably the least technical of questions you could have asked it - "What should I image tonight?" - its response lacked insight or critical thinking at even the most fundamental level. As predicted, it gave a mild response with the most generic and common messier objects possible with the most generic and common exposure length followed by the most generic possible additional recommendations yet it still managed to mess up in several places. Now extrapolate this to real a real technically challenging question (recalling that it still has literally zero idea what any of this is) and you can imagine how spectacularly it will fall on its face. These tools are just simply not useful in any capacity for astrophotography. 

The issue is that people do not understand that this is just an over-confident blind regurgitation of random information - they think its some tailored, highly intelligent response and they trust it. For a beginner, the type of person who is likely to be asking these questions in the first place, it is especially problematic as they cannot identify what parts are sound logic and what aren't.
Edited ...
Like
jonnybravo0311 8.79
...
· 
·  1 like
·  Share link
Here's what I conclude by looking at the screenshot…

1. Guiding is very good. You have no spikes in either RA or Dec. Everything is nicely controlled. Your total RMS is 0.44", which is well under your image scale.
2. Stars appear to be round throughout.
3. Your median is about 135 ADU above your min value.
4. Very few saturated pixels (only 273).
5. HFR starts to increase immediately after an AF run, then oscillates between 1.9 and 1.93 before the next AF run, which brings it back down to the low 1.8s, but it immediately starts to rise again. Granted, the swing is pretty minimal, and the impact is pretty exaggerated by the way the graph is presented. I'm going to assume the first AF run was a filter switch, since the number of stars shot up from ~400 to ~1000. 

Overall, I'd be happy with this imaging run. My only area of concern is the HFR jumping up immediately after a focus run - and that's a minor concern at best. I also don't typically shoot at gain 100 / offset 100 for broadband targets, but those settings seem to be working for you, given the stats (min/median/max ADU values).
Like
Anderl 4.52
Topic starter
...
· 
·  1 like
·  Share link
Jonny Bravo:
Here's what I conclude by looking at the screenshot...

1. Guiding is very good. You have no spikes in either RA or Dec. Everything is nicely controlled. Your total RMS is 0.44", which is well under your image scale.
2. Stars appear to be round throughout.
3. Your median is about 135 ADU above your min value.
4. Very few saturated pixels (only 273).
5. HFR starts to increase immediately after an AF run, then oscillates between 1.9 and 1.93 before the next AF run, which brings it back down to the low 1.8s, but it immediately starts to rise again. Granted, the swing is pretty minimal, and the impact is pretty exaggerated by the way the graph is presented. I'm going to assume the first AF run was a filter switch, since the number of stars shot up from ~400 to ~1000. 

Overall, I'd be happy with this imaging run. My only area of concern is the HFR jumping up immediately after a focus run - and that's a minor concern at best. I also don't typically shoot at gain 100 / offset 100 for broadband targets, but those settings seem to be working for you, given the stats (min/median/max ADU values).

Hey Jonny, 👋 

thx!
i am indeed happy with that night. A few more nights like that and i can start to process the data.

any reason you shoot at other gain/offset values?

The first af run really was after a filter change. I am always starting my imaging sessions as soon as i find a guidestars. To bright for lum but ok for a few ha shots. 
regarding the hfr jumps. Always was like that for me. Only thing i can think of causing that is my sesto senso 2 but it is indeed to little to really care about.
Like
jonnybravo0311 8.79
...
· 
·  1 like
·  Share link
Anderl:
any reason you shoot at other gain/offset values?


I own a Player One Poseidon, so it's not completely an apples-to-apples comparison. Having written that, when I shoot broadband, I typically use a gain of 0 and an offset of 25. My skies are bright (Bortle 7 at best), so I swamp the read noise very quickly. I like to get as much depth in the well as I can, and that is at gain 0. Also, if I'm shooting with the SharpStar 15028HNT, it's a fast scope and collects light very quickly. Shooting at unity gain (which is 125 on my camera) means I'm taking extremely short subs, especially with the luminance filter.

As I wrote in my first response, the screenshot shows you're doing just fine with 200" subs at your gain/offset settings. I suppose you could experiment, but as my dear ol' pappy used to say, "if it ain't brok, don't fix it"
Like
 
Register or login to create to post a reply.