4.42
#...
·
·
2
likes
|
|---|
|
Hello everyone, Lately I have been thinking on which of the following - Dual-Narrowband filter+OSC and Ha/O3+Mono - is most time efficient for shooting and I'm starting to reach the conclusion that Dual-Narrowband filter+OSC may be a better option. This is a bit against astrophotography common sense but still I'd like to share my thougts for your appreciation. Let's suppose you have 10 hours to shoot a given target and you want to capture Ha and O3. For that you have: i) an OSC camera and a dual-narrowband filter (DNB) ii) a mono camera and an Ha and O3 filter The sensors of each camera are the same and the transmitance of each bandpass is equivalent between each filter and the DNB filter. In the OSC camera you'll have: 1) 10 hours in Ha, collected in 1/4 of the sensor 2) 10 hours of O3, collected in 3/4 of the sensor When debayering, software will interpolate the missing pixels. Hence, in the end, you'll have two masters, 10 hours each. In the mono camera, you'll have: 3) 5 hours in Ha, collected in 100% of the sensor 4) 5 hours in O3, collected in 100% of the sensor There's no debayering and in the end you'll have two masters, 5 hours each. In this situation, having only half of the exposure will yield a 33% more noisier image, which makes me think the best approach to follow would be to shot OSC+DNB filters. Would you agree with this? I would really like to have your opinion to confirm/dismiss my thinking. Cheers, André |
10.79
#...
·
·
3
likes
|
|---|
|
Hi Andre, I admit, I haven‘t elaborated the numbers in detail for your suggested application but I once did a calculation for LRGB vs. OSC. It turned out that for R the SNR for the OSC was about 80% than the SNR if it were shot with mono. For green, the OSC had a slight edge over the mono. While it may turn out that the SNRs are quite comparable, my point is this: The OSC data will be interpolated and *might* miss detail due to reduced resolution. If resolution matters depends on the image scale, on the other hand. If you oversample, you can afford the loss of resolution and shoot OSC for sure. Another point: if you chose OSC + multi-band, you cannot distinguish Ha from S2 in case your filter transmits both. Personally, for NB imaging, I‘m clearly favoring mono and filters for each frequency. CS! Björn |
7.22
#...
·
|
|---|
| In short that would be correct but you need to factor in the double losses from both the dye filter and the DNB filter in the case of the OSC. Obviously this does not occur in the case of the mono camera. The typical dye filter losses run around 5%-10% for high end sensors so there you go, you would end up with 9 hours per NB channel in the case of the OSC camera vs 5 hours per NB channel in the mono case. One slight twist concerning Björn remark about sampling is that while you can oversample to accommodate the loss in resolution due to the OSC structure this accommodation comes at a cost: longer exposures. |
6.62
#...
·
·
22
likes
|
|---|
|
In my opinion, the whole mono vs osc debate comes down to weather. Do you live in an area where clouds are not an issue and you can shoot every night? Mono. Do you live in an area with only 14 clear nights per year? Osc. Of course you can use mono in the latter, but it could take many weeks or years to capture required data. There’s nothing wrong with that, but not everyone has that kind of dedication or time. Where I live, most clear nights are during the summer where we only have about 3 hours of night time at our disposal. During fall and winter, there is a consistent cloud cover that only lets us peek through maybe once or twice per month. These are very difficult circumstances for this hobby throughout the year, but osc can help mitigate a lot of these issues (at the expense of reduced SNR). If time is your enemy, osc is definitely the more efficient choice. |
|
7.22
#...
·
·
5
likes
|
|---|
Guillermo de Miranda: Not to say the least expensive |
3.94
#...
·
·
3
likes
|
|---|
|
I think I would put more focus on the quality of filters you can get instead of crunching exposure numbers. To give you an example, I do own both the L-eXtreme Filter and an Antlia 3nm Filter set. For targets without too bright stars the L-eXtreme works wonders and given that your OSC Camera has enough megapixels you will be very happy with the results. I was quite happy with the 16 Megapixels of my old ASI071 Pro but now that I have 26 Megapixels I can easily live with the possibility that I might loose a little of resolution because of the bayer matrix. So for that use-case your calculations are quite right minus the fact that I usually throw away blue as it has a bad Signal to noise ratio compared to the green channel for Oiii Now on to the things I think that are worth thinking about: To get true SHO images (which are a lot easier to process to a pleasant form in my opinion) you'd need an extra Sii Filter to have complete signal coverage which overall increases your exposure time because you now need two instead of just one exposure compared to three exposures for mono. When you are experienced in combining only Ha and Oiii channels then this may not be relevant for you, but for me the struggle of making something pleasant with only H+O on popular targets was too much, with the added Sii things suddenly became a lot easier. But this of course has a lot to do with personal taste and experience in processing. If you are a fan of RGB stars then the OSC camera helps a lot, you get them with just one more exposure compared to 6 (or 7 with lum)(or 5 with only RGB+HO) exposures that you would need with a mono camera. So here the time savings are more relevant. Things get a little messy when you have bright stars in the picture, the Halos that the L-eXtreme produces (at least mine) in Oiii are quite visible. For the bright targets I at some point decided to get the Antlia Filters, I still use my OSC Camera to shoot but am a lot more pleased with the very reduced halos. Again here, when you are a good at Post-Processing then you will be able to get rid of the Halos and the limitation will not apply. There is also a lot of development going on, Antlia has released the 5nm ALP-T filter which is said to be much better at halos and Optolong has spread a few samples of a new Ultimate filter that is supposed to be 3nm. I have the ALP-T on order and when it holds it's promise then my prefered way of shooting will likely be to do 1xRGB for the Stars, 1xALP-T for Ha and Oiii and 1xAntlia Sii which will be a very relevant time saver over having to use 6-7 filters on a mono camera. And yes, a big plus to the thought that when you have a limited number of clear skies you likely make the best out of that with an OSC Camera. |
7.75
#...
·
·
9
likes
|
|---|
|
A while ago in another thread I claimed that at some point in AP you have to define your level of ambition. If you aim for maximum quality I believe you have to go for mono + filters at the price of significant investments of time and money. If - like me - you're happy with some half decent looking images to stick at your walls at home OSC wins hands down. If you take the time and energy to try to calculate what is better in terms of SNR, resolution and efficiency of exposure time, this may be an indication that you belong to the first group. Nothing wrong with belonging to either group, both is a viable decision. Just don't complain that you don't get an apod if you decided to join the scond group  .Have fun and clear skies Wolfgang |
1.81
#...
·
·
1
like
|
|---|
Andre Vilhena: I am afraid it is not as simple as that. Just try to think about the total number of photons you will really collect/register . If the same number of let say "red" photons hits the area of let say 2 x 2 pixels on OSC within a given time and they are randomly spread over the surface. An OSC camera will collect only 25% of them . Mono camera with the same pixel size will collect them all. For OSC camera the debayering process will try to estimate how many photons other 3 pixels should have collected and what should have been their color. Therefore Debayered image is somewhat "smoothened " already , when you get it as "raw" . |
5.73
#...
·
·
4
likes
|
|---|
|
As mentioned OSC vs Mono depends on your personal time/comittment/budget consideration. Or you wanting easier or better? The article by Tony Hallas in Astronomy Feb. issue claims that the QHY410C OSC is capable of achieving Mono results. That's fine for Tony, he images from upper California in dark skies. That want work for me in my Bortel 7 skies. One may argue that the new Bi-band/Tri-band filters can compensate for that. Yes & No. Ha requires different band pass and exposure times as OIII. 5nm or 6nm will gather very good Ha S/N. But OIII needs 3nm to smooth the background noise in Bortle 7 skies. Also the OIII exposures needs to be longer to compete with the stronger Ha signal. With Ha, OIII & SII, it's not a one size fits all, especially when it comes to the moon. As far as freeing session time, being able to image during moon light is important. One needs 3nm in OIII to do that well. Expensive, but the best equipment to free up imaging time is a remote permanent set up. No way around it. No other equipment will remove burden as effectively from an imaging session. I don't agree that exposure time equates into photos gathered (sigal) equally in a Bayer array and a non-Bayer array. The Red filtered pixels on a Bayer array will only gather 1/3 of the photons. I understand that most of the photons from an emission nebula are in the Red spectrum, but a mono chip will gather all of them. Not just 1/3. So, to assume that 10 hours of exposure renders 10 hours of Ha segnal, I think is incorrect. It's more like 3.33 hours of signal. Where the mono Ha & OIII gets a full 5 hours each. Each to there own. Lynn K. |
|
7.22
#...
·
·
2
likes
|
|---|
Lynn K: Signal is intensive not extensive. Claiming otherwise would be ridiculous. Each pixel "sees" a fraction of sky and that's that. How many are there is pointless and as long as you satisfy the Nyquist sampling criterion you are in, the original signal can be reconstructed. |
4.57
#...
·
·
1
like
|
|---|
| An interesting exercise would be to take a set of Ha and Oiii narrowband frames, align and crop them them, then use PixelMath to discard 3 out of each 4 pixels of Ha to create a R channel and 3 out of 4 Oiii to create B and then 2 out of 4 to create G (using a valid RGGB Bayer mapping). These could then be combined to create an OSC image from the two narrow-bands and perform a direct comparison of narrowband vs pseudo dual-band OSC with the same set of data (the transmittance of the R,G,B filters could be emulated with a suitable scaling factor too). |
|
0.00
#...
·
·
2
likes
|
|---|
Guillermo de Miranda: Makes sense to me. I do OSC, and have been considering to try Mono, but when I factor in the hours needed and the particulars of where I live, I keep thinking that it's not worth the investment and additional effort. As it is, I am getting results I would not have dreamed of years ago with OSC! |
|
11.01
#...
·
·
2
likes
|
|---|
| I'm personally erring on the side of mono+filters even though I live in an area that has clouds 80% most of the time. I say that because I like to have separate channels that I can manipulate easily in post-processing, that are focussed individually and that I can choose different exposures on. I know it is more time-consuming, however having moved from a mirrorless camera to a dedicated mono camera I am having much more fun creating the final image. |
#...
·
·
4
likes
|
|---|
|
One other issue (that may be mentioned above but I didn' see it?) is that O3 images are more sensitive to light pollution and moonlight than are HA. Therefore I like to reserve those rare clear and moonless hours for O3 or RGB and do HA and SII even when the moon is up. This tailoring to conditions isn't really an option for an OSC plus L extreme. Also you don't necessarily want to be restricted to equal HA and O3 acquisition times? i.e. O3 SNR is usually lower than HA (except for planetary nebulae) and so it often makes most sense to image O3 for longer than HA So overall I tend to use mono camera for narrowband beacuse it is more flexible-- but prefer OSC for RGB because of time constraints and the simplicity, |
|
6.89
#...
·
·
2
likes
|
|---|
andrea tasselli:Lynn K: Hi Andrea, I will be delighted to be wrong, but think this is too optimistic, specifically in narrowband for red where you are basically sampling only 1 out of every four pixels and the other three convey zero information (unlike in broadband where they still give you parts of L). Sampling is 2x2 pixels wide so the reconstructible signal is details 4x4 pixels wide which depending on the situation might or might not be okay. But of course light gathering efficiency is only 25% so SNRingly speaking this is not the same thing at all as using real larger pixels sized 2x2. So the mono sensor can capture detail 4 times smaller (same resolution) or give you same detail but 2 times better SNR (at a quarter of data storage requirements) at bin2. Again, this is for narrowband red signal. Not for the general case. Cheers, Dimitris |
7.14
#...
·
·
3
likes
|
|---|
|
With (presumably) halo-free Antlia dual-band filter OSC imaging with 2600MC is very close to basic mono usage. What you give up is mainly: flexibility - how long to expose each line, what to expose with the moon up. SII - very inefficient until an OIII/SII dual band filter comes out (it will eventually be produced). With such OIII/SII filter on the market I'd say OSC will have almost caught up with mono on narrowband other than flexibility and moonlight issues. The advantage would be painless and efficient gathering of RGB data. BUT So what does mono still have over OSC? Luminance, if you at all care about faint reflection objects and galaxies, mono is still king. Its advantage is tremendous and won't be soon overcome, if ever. Writing as an owner of both 2600MC and 2600MM. |
|
7.14
#...
·
·
5
likes
|
|---|
|
I see a lot of people bringing up over and over the bayer matrix (25% yo!) and how much more resolution mono has -- it's not really that important. With CFA drizzle you recover pretty much all of it. Bayer matrix is a red herring in 2022. Here's a link to L-Extreme Red Channel OSC vs Mono H 7nm comparison. The resolution advantage of mono isn't too pronounced. gallery_352013_17466_369009.png (1490×954) (cloudynights.com) |
|
3.94
#...
·
|
|---|
|
Thank you for the comparison picture! I the MC version drizzeled or is it simply the debayered version? If drizzeled, would you mind adding a 3rd picture with OSC debayered only? Another relevant information would be if you need to increase exposure time with the color cam to get the (roughly) same result. Quantum Efficiency of the Mono Camera seems to be (a little) higher, but both are very close so I am asking myself if this results in a relevant difference. |
|
7.14
#...
·
·
2
likes
|
|---|
|
While I don't think I have it as I did that comparison a while ago I remember it is definitely more smeared, as expected. CFA Drizzle on Dual-Narrowband data is a must to recover the resolution and get rid of interpolation. Once that is performed (on well dithered data) the integration is almost as good as from dedicated mono camera Ha filter. When it comes to narrowband the quality you get from MC+Dual Narrowband is tremendous for much less than mono setup. The only thing missing is the OIII/SII and it HAS to come out eventually, I can't believe we don't have one yet available! This is what Juan has to say about it. I really am surprised this isn't common practice/knowledge. Bayer drizzle instead of de-Bayering with OSC | PixInsight Forum |
|
6.89
#...
·
·
4
likes
|
|---|
I see a lot of people bringing up over and over the bayer matrix (25% yo!) and how much more resolution mono has -- it's not really that important. With CFA drizzle you recover pretty much all of it. Bayer matrix is a red herring in 2022. Hi, The Nyquist-Shannon theorem is fundamental in digital signal processing and gives you the maximum possible resolution of a continuous signal that you can reconstruct from discrete samples. In our case, the signal is the image and the samples are the pixels. As a theorem, it is not something that can be overcome algorithmically. Instead, it is what the best algorithm can possibly hope for. Regardless of year. To make it absolutely clear: that 25% resolution you get in R with a Bayer device is not something you can work your way up from, reconstructing better detail. It is, as a matter of fact, the BEST (accurate) result any conceivable reconstruction will ever be able to come up with, starting from way worse. Using CFA drizzle does not change that. For CFA drizzle to work you must have dithered, which means at least 2X exposures (so that each red pixel "falls" on different places) and typically more. So you are not really sampling with superpixels 4x4 large, you are sampling twice and thrice more fine grained, consequently spending twice and thrice more time imaging. CFA drizzle simply makes sure the extra data is geometrically arranged and combined in a way that tends to approach what is theoretically possible (and avoids the small "leakage" of information between colour channels inherent in Bayer demosaicing algorithms, which is a good thing but irrelevant to narrowband). Now, I don't think anybody said 1000 well dithered subs from a Bayer device cannot produce something as good as 250 not-so-well dithered subs from a mono device, or even better. But this is not really apples to apples, is it? And note: you can take 1000 well dithered subs with the mono and drizzle those. I guarantee you the result will absolutely blow away that from the 1000 Bayer subs. So, the simple fact remains you cannot get the same result using just 1/4th of the samples. The only exception that I can think of is when your incoming signal is inherently low resolution. Say you have bad seeing or you are using a small telescope or you have tracking errors or whatever, in that case it truly does not matter if you are bayering or not because stuff is too low res to tell the difference. Your example seems to fall under that category. Nevertheless, one cannot help noticing that the subs in the OSC image are 180 seconds long while the mono image is only 120. Cheers, Dimitris |
|
7.22
#...
·
·
3
likes
|
|---|
| The recovery of resolution by drizzling does not fall foul of the N-S theorem. Since the sampling occurs at spatially diffrent locations (and mind you, they don't need to be unit-pixel or more apart) at different times, it is irrelevant whether that is from a Bayer pattern, X-Trans (forgot those?) or mono. Is the mono resolution better? Sure, even because it is going to have a better SNR than equally exposed CFA. It is worth it? Maybe or maybe not. I have done LRGBHaOIIIHbetaUVIR compositions for more than I care to remember but now I chose to go OSC and all too happy about it. Will I go back to mono shooting as mainstay of my imaging even if the sky conditions were better than currently are here? I doubt it. |
|
7.14
#...
·
·
2
likes
|
|---|
andrea tasselli: That's the right path to follow honestly unless one wants to go all-in. Having used both up to now I can say that the big showstopper for OSC was halos that came with the L-Extreme filter. Barring that it's very viable way to go shooting narrowband. Antlia's 5nm filter is supposed to make that issue go away. For galaxies and other faint L stuff, mono still can't be beat. This topic goes back years and years, it's the eternal mono vs osc debate...bottom line, I own both so I don't have a horse in this race, and mono is my main tool. I do recognize that OSC is closest it's ever been. |
3.34
#...
·
|
|---|
| One factor only mentioned once in this thread but I think is quite significant is the ability to focus each filter separately. I am currently experimenting with optolong 7nm SHOs on a ZWO071mc vs an STC duo. I know it’s not exactly apples for apples because the Optolongs have a tighter bandwidth. I am running a 200mm f/4 newt with a GPU f/4 coma corrector. I haven’t had enough clear skies to do enough experimenting yet to say what the results are but I will post them when I do. I am also aware that this isn’t comparing mono vs OSC but you’d imagine if OSC with NB filters is sharper, the result can only be better again with mono pixels. |
|
7.14
#...
·
·
1
like
|
|---|
| Yes for people who do NB with achromats or the ubiquitous budget "apo" doublets the ability of mono to focus separately oxygen and hydrogen lines is an advantage. |
#...
·
·
1
like
|
|---|
|
UK sky conditions over the last week or so perfectly illustrate a practical aspect of using separate versus combined filters. Clear dark skies on 5 th Jan allowed me to take 3h of good mono O3 frames once the moon had set. Of course it would equally been possible to have imaged the same object for 3h with an OSC camera plus L-extreme and to have done so would have provided HA as well O3 data. Clear skies on 13 th Jan allowed me to take 3h of HA of the same object but this time within 30 degrees or so of a waxing gibbous moon. The same 3h of course would equally have been possible with an OSC and L extreme - but then I'd argue that only the 'HA' channel would really have been useful with the SNR of the O3 channel too compromised by moonlight. You can argue it either way. On the first night the OSC plus L-extreme would have been an efficient solution collecting 3h of both HA and O3 in the same time that the mono could only collect O3 frames - thus a picture obtained quickly from just one night's work. On the second night however the L-extreme plus OSC effectively reverts to only providing useful additional HA data - and doing so less well in terms of focus, resolution and time needed to achieve a given SNR than a mono camera plus single band filter. Moonlit but clear skies when the L-extreme won't be working to its full advantage are rather common. So I'd argue that over any given typical period of a few weeks the observing opportunities most likely to arise in practice mean that the mono camera plus single filter solution will always come out top in terms of the quality of the final picture obtained . |
