7.16
...
·
·
2
likes
|
|---|
Arun H:Christian Großmann: Hi Aron, I see these discussions everywhere  . Let's say, you are right. I do not agree totally, but that was discussed on many other places before. But I get the point. So I ask myself, what is the benefit of having a stop more dynamic range in this case? The narrowband signals, especially OIII and SII are usually so weak, that you barely need the additional stop (I don't say it can't be useful in some cases). Broadband, however, is a completely diffferent story.These days, we have really great sensors and a lot of equipment with great properties. The discussions about the details heat up sometimes and from a scientific point of view there is always a justification. People everywhere try to use their settings based on technical details and I'm fine with that. I've done so for many years in real life photography and astro photography as well and still do in most cases. But I may argue, that there is the art as well. Common people looking at the final images don't care about what gain was used or how many stops of light the sensor was able to record. We as astro photographers look for those small details and sometimes over-analyze these things IMHO. Of course, we may be able to get fainter details or other things with those technical stuff. But if you live in a standard light polluted area (maybe bortle 3 or worse), you may barely benefit from those things. How many percent may your image improve? Please, please don't get me wrong! I am not saying, that you are not right. You are and so are many others. But in my opinion, the art itself falls short too many times. So I decided to use the tools I have and sometimes compromise. It helped a lot to keep me doing the things I love and reach a point of satisfaction. If I only think about the technical details, there is always a "better". And that's the biggest motivation killer of all. At least for me. We are doing a hobby which is related to a lot to these technical details. But they are not everthing astro photography stands for. You are here for a long time now. Your images clearly show a lot of experience, aren't they? I really like them and always did. But I doubt the images look worse if you would've taken your subs at gain 180 and not at 120. There are so many things that are much more important and have much more impact on the final result. I am sure that you agree. I'd love to discuss these things while having a beer with you and all the other geeks around here. But we went a bit off topic. So lets just say we are weighting things from a different perspective  . So @Claudio Boicu has one more option to choose from...CS |
2.41
...
·
|
|---|
| (deleted) |
|
7.16
...
·
|
|---|
CVZ_Astrophotography: mine too... even on the QHY294M |
4.83
...
·
·
1
like
|
|---|
Christian Großmann:CVZ_Astrophotography: I have the QHY294M and for OIII and SII the flats look similiar to yours - I once asked and was told, that the reason for this strange pattern is coming from two sources - coatings of the optics and the filter itself. No filter, I was told is perfectly even regarding transmission of the desired wavelengths and also no coating is perfect. As result, if you take an image of a more or less evenly illuminated surface - like for flats - the overall transmission is inhomogeneuous as well. If you deduct this from your actual light-sub, it should level these inhomogeneities out. I hope, you can understand my nonperfect English, it is my third language..... :-) CS Georg |
|
7.16
...
·
|
|---|
Georg N. Nyman: That totally make sense to me. I never thought about it but now you told us, it seems quite obvious. |
13.22
...
·
·
2
likes
|
|---|
Christian Großmann: Hi Christian , I don't think you will get any disagreement from me that the most important thing is to have fun and take images you are happy with. So, if doing it a certain way works for you, I definitely would be respectful of that! That is why I said "Obviously, if Gain 180 works for you, feel free to use it"! So the only purpose of my post was to make a technical point. The OP or others are certainly free to choose as they see fit, as long as the images work for them. CS, Arun |
2.10
...
·
|
|---|
Arun H:Christian Großmann: I do not want to flog a dead horse here, but I appreciate information from people who are more knowledgeable than I am, so here's a question. If I photograph m42 or m31, where there is a wide range of brightness, I can understand why a large dynamic range would be beneficial. However, what about a situation where the object is quite faint, such as the squid nebula(sh2-129)? It is very faint, and depending upon your optics, you could easily be shooting 600sec subs, and that is generally where I draw the limit. So if by boosting your gain, you keep your subs to a length you are comfortable with, does DR matter in this case? The subject is very faint, so DR would not seem to be important to me, but then, I am analyzing from a position of limited knowledge. Thoughts? |
|
13.22
...
·
·
3
likes
|
|---|
So if by boosting your gain, you keep your subs to a length you are comfortable with, does DR matter in this case? The subject is very faint, so DR would not seem to be important to me, but then, I am analyzing from a position of limited knowledge. Thoughts? Hello Scott, Indeed you are correct that DR (or more correctly available full well capacity) is less important for really faint subjects, so long as you don't mind stars getting saturated (of course, stars can be dealt with separately these days by taking short RGB exposures). But I think out is instructive to look at the actual noise numbers of the ASI 294MM:  You can see that, between gain 120 and 180, the read noise is hardly changing at all!! So sure, you can image at Gain 180 for a faint subject. But is the 10% or so change in read noise really meaningful? By keeping Gain 120, you don't really have to worry about whether you are imaging a very faint, or slightly brighter subject. Edit: Here is some more detail. For any sensor, there are two components to read noise:
Component 1, relative to your signal remains the same whatever gain you choose. When you increase Gain, what you are doing is amplifying your signal before it is converted to digital, and hence increasing your SNR relative to noise component 2. This is why read noise decreases with gain. The cost is, because you are limited in how many digital bits you have available, you are throwing away bright signal because there is not enough bit depth to keep it after amplification. For sensors like the 294MM, you can see that, after Gain 120, the read noise really does not change very much after Gain 120. This is because the analog to digital conversion noise is already very low, so the dominant noise component is the sensor read noise Component 1. So you are sacrificing full well capacity for almost no gain in noise. |
|
7.16
...
·
|
|---|
|
I'd like to throw in another fact. It was already mentioned, that gain just amplifies the signal. So basically you have a copy of the signal with a higher amplitude. But if this is done before the A to D conversion, the signal fits the range of the converter better. So lets say, you are not cutting signal on the top, which should be the case for faint stuff. You basically are able to record the signal with more resolution. So lets say, we have 16 bit converter. This results in 65536 brightness values. With a lower gain, the same signal is represented on the low end of this range. Let's say you get values between 0 and 20000. That is, what you have to stretch with software. If you increase the gain and amplify your signal 3 times, you will end up with values between 0 and 60000 instead. So You won't have to stretch the result as much as before in software and get a higher resolution of the things that was missed in the low gain image between two numbers. This seems to be beneficial, because you used the optimal range of your AD converter. However, this is true only for ideal amplifiers and ad conveters. In reality, this relation is not as simple and we have to compromise some things. But even if you are not able to get 3 times the quality out of your high gain image, you may benefit from the gain setting. That's one reason it exists. In astro photography, we usually are working on the extreme low end of the histogram and gain may help us to increase the resolution of the conversion result. But it has its downsides as well. So one has to decide, which setting he/she uses. So in case of gain 120 to 180: yes, the read noise decreases only by a small amout. But the signal is amplified by 50% before the AD conversion and the resulting sub has better data to work with. This all changes, if you amplify the signal to the point it clips. Then you have to decide, which is more important to you. An example is the ghost nebula in cassiopeia. It is really close to the star Navy and you have to choose to either get the faint parts of the nebula and a blown out star or you keep the gain setting and exposure time so low, that you will not blow out the star. Then you will have a really hard time to process the nebula itself (if it even comes out of the noise). EDIT: We have two kinds of gain here. The analog gain (camera gain before the AD conversion) has basically infinite resolution. You use this to your advantage to spread the signal over the range of the AD converter. After the conversion, you use kind of a digital gain in post processing. If you amplify here, you are just increasing the step between two values without adding something in between. So what can be done before the conversion will bring you better results afterwards. |
|
13.22
...
·
·
1
like
|
|---|
Christian Großmann: Hi Christian, If you are referring to quantization error (the fact that when you amplify the signal, it can be stored at greater resolution in digital than when not) - this effect is actually already included in the noise graph. From the 294 MM manual: "Read noise includes pixel diode noise, circuit noise and ADC quantization error noise, and the lower the better" (Page 6)." So yes, there is this benefit to high gain, but it is already taken into effect in the noise graph. One of the great benefits of modern sensors is the really low noise levels (including quantization error) that allows you to get the benefits of higher dynamic range. There are certainly benefits of using higher gains/ISOs, in applications like sports or machine vision, where you do not have much processing time available. But that's not really the case in many other applications where you do have time to process. |
|
13.22
...
·
·
1
like
|
|---|
|
Since the topic of quantization error was brought up, I thought I'd share one additional reference that I read many years ago that I still refer to from time to time - Emil Martinec's exceptional writeup on noise, which I have linked here. It is very technical but has a wealth of information. What you will see is that there really is no benefit to using more bits to represent the data than the noise level. At Gain 120, the read noise is approximately 2 electrons. The 294MM sensor's gain, at Gain 120, is 1 e-/ADU. That is, you are already recording your data at a finer resolution than your noise - and that is not even including the other sources of noise, which are probably much more important than read noise - such as photon shot noise and dark current noise. If you increase the gain and amplify your signal 3 times, you will end up with values between 0 and 60000 instead. So You won't have to stretch the result as much as before in software and get a higher resolution of the things that was missed in the low gain image between two numbers. So this is incorrect for the reason I mentioned above, at Gain 120 and above. You are already quantizing your data at a higher resolution than your noise allows. The stacking and processing in PS and PixInsight is done in floating point, so the absolute number will not matter, just the number of significant digits. The analog gain (camera gain before the AD conversion) has basically infinite resolution. I think what you mean is that the resolution of the recorded signal (not gain) before AD conversion is infinite. This is also not correct. Because, once again, you are limited in practical resolution by noise. It would be similar to using a micrometer which has 20 digital digits on display, but has an error of 0.01 mm. Everything after the second decimal place is meaningless! |
|
7.16
...
·
·
1
like
|
|---|
Arun H: Hi Arun, this is not what i meant. But it doesn't matter. The website you linked to is indeed very interesting. This changes some things, although there are still some issues I have after quickly reading through some of the points. But I have to spend some more time on this topic first. Maybe they clear up while digging deeper into this matter. So thanks for the link. |
1.43
...
·
·
1
like
|
|---|
Arun H: 🎯 That's exactly what I was going to say. The 294 needs to have at least 3-4 second flats with the ADU at about 30% |
0.00
...
·
·
1
like
|
|---|
|
Not entirely sure what's causing the really odd look of the flat frames (outside of the normal 294 look...) I shoot my flats at the same gain as I shoot my light frames. (108 for narrowband in bin 1, 120 for bin 2.) I only shoot long enough frames to reach an adu of around 25k~ ish. I don't worry about it being super precise. Just in the same ballpark. My previous set of flats were 5.5 seconds long (I try to adjust lighting/time to achieve the same length of exposures so I only have to use one set of dark frames, since the bin 1 file sizes are huge. For LRGB, typically I shot at gain 0 (bin 1), but feel like I might experiment more with it, but honestly if I want to shoot in LRGB I'm likely to shoot luminance in bin 1 and gain 0, and RGB in bin 2. I feel like it's worked well for me, though I suppose it's purely up to the user in question. For narrowband, I shot at various settings one night, to find out what I thought worked well, and landed at 108 for myself (baader 6.5nm filters.) |
2.61
...
·
·
3
likes
|
|---|
|
Hello, quick update, i used the tshirt into the sky method, and the sii filter actually looked like a flat! took the same with ha and oiii, although the oiii still had that “cut off” part on the bottom, but it wasnt as noticeable. So i just went with it and did some gradient removal in APP, and also some other adjustments, and somehow got an image. Its not the flat field i hoped for but i was just desperate to see my first light with that camera. On monday my flat panel should arrive and im gonna retake a set of flats for all the filters, and i hope that it will improve. Anyways here is the image i came up with, short of 4 hours integration time, SHO. Despite the problems im still happy with the final image.  |
|
2.61
...
·
|
|---|
|
hello everybody, I was away on vacation for like a month, but now ive come back, hooked up the flat panel from Deepskydad and took some flats. for LRGB there is no problem with them, but on the SHO filters, there is still the "cut off" part. someone also suggested to shoot flats between 5-10 seconds, so I did that, but the results are the same. I tried turning the rotation of the camera a bit (loosen the thread up a bit and cover it with some black electrical tape so no lightleaks are there) but again, same outcome. Im really running out of things to try so if anyone has any more tips on what I should do, please don't hesitate to post. |
|
7.16
...
·
|
|---|
| As you turned your camera, did the position of the cut off change the same amount or did it stay where it was before? |
|
2.61
...
·
|
|---|
Christian Großmann: it changed.. and it got dimmer and dimmer, and I actually got "normal" looking flats for the 294mm. I guess it has something to do with the camera then? or is it something in-between? |
|
7.16
...
·
|
|---|
Claudio Boicu: No, the exact other thing is true. If it has changed, the camera is not the problem. Would you please do a sky flat for us as well? If you put the shirt on, you can just use the sky as flat panel. It doesn't matter that much, if there are actually clouds as long as they have not too much contrast. The thought is, to see if the straight edge disappears. We won't use the flat, of course. If it dissolves, it must be the flat panel. If not, then there must be something else. |
|
7.16
...
·
|
|---|
| If it got dimmer, it could be an edge in the imaging train, which is likely the case. If this edge gets more parrallel to the long side of the sensor, it just flips out of the frame and disappears. There is something mechanically happening here. |
|
2.61
...
·
|
|---|
Christian Großmann: in the past before my holiday I took a sky flat and the edge got noticeably dimmer. but il retake the flats with each filter and doing skyflats. |
|
7.16
...
·
·
1
like
|
|---|
|
If the edge is still there, then it must be something in the scope. The edge is really straight. It must be a straight part in the imaging train somewhere. The filters are round and the filter wheel has round connections and holes. I assume you use round screw on filters as well. The window in front of the sensor of the 294MM is rectangular. But if the sensor is in the middle of the glass, this should work without issues. Then, when rotating the camera, the edge should be at the same position. Because that's not the case, the camera might be ok. Is there maybe something straight in the tube itself that may be hit by reflections before bringing light on the sensor? Usually I only have round parts in my scopes. If you used a Newton scope, I would guess its the focuser tube going to far into the main tube. But it's a refractor that works different. If you look at your sensor, does it look clean? I hear people with ZWO cameras having leaks. There can be thin films of something on the sensor that looks like humidity or have rainbow colors. The edges can be straight as well. But then the edge should've been at the same position if you rotate the cam. So no solution here. This is weird! |
|
2.61
...
·
|
|---|
this is the sky flat for Ha. |
|
2.61
...
·
|
|---|
|
I threaded off the filter wheel with the camera and it doesn't seem like is dead center, its slightly off that might be the problem I think, because if I rotate it, the edge turns with it. I recalibrated my filter wheel using Nina but still slightly off center. any suggestions on how to fix it? or do I need to contact my reseller and get a replacement? idk if it is useful but back when I got my camera I still didn't have my filters so I used a oiii 2" filter that I had lying around and with a filter drawer, the flats looked perfect, so I guess having a off center filter wheel that is 1.25", it has little to no wiggle room for error and it would work when the filters are positioned perfectly in the center. |
|
7.16
...
·
·
1
like
|
|---|
| When I look at the edge, it must be near the camera. The closer to the sensor, the more defined the edge should be. From experience, I might guess it's about 4-5 cm in front of the sensor. But that's just the ballpark. It is not on the front lens of the scope, I assume. But what exactly it is, I have no idea... |
