3.81
#...
·
 |
|---|
|
Sorry for bringen that back but after reading the things in here i am not smarter. can somebody tell me in short how much i lose or win with a duo narrowband osc setup against mono sho? |
7.56
#...
·
·
2
likes
|
|---|
Andi: Money? |
7.16
#...
·
·
1
like
|
|---|
Rafał Szwejkowski: What if I tell you that this OIII/SII Filter exists ? ;-) Have a look here: https://youtu.be/n9sD4nYuDZg |
2.97
#...
·
·
1
like
|
|---|
| You are correct on the time/SNR comparison (if all else were truly equal, which often isn't the case), but you have to consider high-frequency detail, which is not captured when you involve demosaicing/interpolation. The way around that is CFA Drizzle, but then you're giving up the time advantage as you want be putting all of the signal into the same pixels anymore. |
|
2.97
#...
·
|
|---|
Lynn K: It does matter if we're talking about detail. But it doesn't matter when we're talking purely about SNR which is a statistic that refers to the pixels you sample not the pixels you don't (no one is leaving the dry pixels in their final image and counting them as noise). The pixels that capture photons will have a certain measurable SNR. Demosaicing will fill in the remaining pixels and they would be expected to have a similar statistical profile as the "real" pixels. It follows that the final image as a whole will too. So when people say that demosaiced OSC images will have a lower SNR for the same amount of time, they're mistaken unless all they mean are the very real efficiency and transmission issues cause by the physical presence of the matrix. But those things are impacting the pixels that actual do capture light and therefore that bleeds over into the interpolation. It's not because 1/4 of the pixels capture red photons etc. Actually SNR inferiority is probably on the order of 10-20% not 75% (or 50% in the case of G). Loss of high-frequency detail is a very important consideration but that's a different measure of quality than I think the OP was referring to. |
3.87
#...
·
|
|---|
Sascha Wyss:Rafał Szwejkowski: There are 2 Sii/Oiii filters that I know of, the IDAS NB3 and a 6nm one made by Askar. I will be adding the NB3 to my setup as I have had excellent results with the IDAS NBZ and no halos. If I had a backyard observatory, great weather and a ton of cash, I’d have a nice mono setup. But I don’t so it’s OSC for my personal gear. I buy mono data from a few remote observatories to augment what I can capture locally and to get objects I can’t see from the cloudy north. I won’t for a second pretend my budget uncooled mirrorless camera could rival the best mono gear but the price difference is staggering and with OSC I always get an image even if my budget and Mother Nature conspire against me to keep it from being exactly what I want. Just an example but a QHY 600m plus a set of ok filters and a filter wheel would alone cost more than double what I have invested in my whole flawed but capable full frame OSC rig. |
|
7.16
#...
·
|
|---|
Joe Linington: Just because you are happy with one IDAS does not mean all are superior - I would suggest you make a more in-deapth comparison before you buy.... The Askar has a narrower bandwidth, also no halos what so ever (that is the main premise the colormagic filters are sold for) and they are readily available. To find the IDAS NB3 in a 2 inch version is an adventure on its own and I heard somewhere they are discontinuing them.... Just saying - there is a reason the NB3 never really took off..... ;-) |
0.00
#...
·
·
1
like
|
|---|
=16pxThere are 2 Sii/Oiii filters that I know of, the IDAS NB3 and a 6nm one made by Askar. I will be adding the NB3 to my setup as I have had excellent results with the IDAS NBZ and no halos. I have used the NBZ/NB3 combo, and can tell you that the NB3 is not in the same league as the NBZ. Its Oiii band is enormous at least double that of the NBZ, and it shows a ton of light pollution compared to the NBZ. You will not be able to combine the Oiii from the NBZ and NB3 together because the NB3 is so much different. You can still use the Sii of course, but the Askar filter is very close to the NBZ's bands. I just switched to the D1/D2 filter set and I can say with certainty the images as just as nice as the NBZ/NB3, at least to f3 so far. You can see my last few images here, they are done with the Askar filters. |
|
0.00
#...
·
|
|---|
#...
·
|
|---|
Sascha Wyss: Aside from the point made above https://www.astrobin.com/users/neverfox/ by Roman that the OSC approach inherently delivers less detail -- unless you opt to drizzle - which process inherently sacrifices SNR -- which loss can only be made up for by imaging for considerably longer. ----the two further points raised by Sascha about flexibility and moonlight are also important. Some objects ..e.g. the veil nebula, the lion nebula and the crescent nebula do exhibit pretty strong OIII signals and for these being stuck with a 1:1 ratio of Ha to OIII frames (especially since OSC cameras are inherently more sensitive to the OIII) actually works pretty well and dual narrow band plus OSC camera combination produces some fine images. However quite a few HII regions are really relatively very weak in OIII and/or SII as compared to HA. For these it often makes sense to image the weaker channels for longer relative to the HA. HA is fine to image under almost any lighting. However with 6 and 12 nm OIII filters scattered moonlight does significantly degrade the SNR --- I expect that 3 and 4nm narrower band filters are less susceptible. So having a mono camera I have always shot HA anytime and reserved the darker nights for OIII. Overall - under my typical observing condidtions - I'd expect to get images from both set ups after a similar number of elapsed nights but that the mono would win out in final quality - and deliver good OOH and SHO images of a wider range of objects. Tim |
|
7.56
#...
·
·
1
like
|
|---|
Tim Hawkes: I'd have to diagree on this, and so is the creator of PI. There is a very moderate loss in SNR when you do a Bayer drizzle at 1:1 scale, which appears to produce better stars colours and a tad better detail in high SNR regions. Low SNR regions suffer a bit more but something, somewhere is got to give... |
|
2.97
#...
·
|
|---|
andrea tasselli:Tim Hawkes: I'm happy for someone to set that straight. I've been meaning to quantify the impact but I don't have the matching mono/OSC data to do it. I'd like to understand more how it achieves that without interpolation because it sounds a bit like getting something for nothing and, as you say, something, somewhere... If it's not costing you SNR to restore the detail, I have to expect it's costing something (and something more than just the processing time of Drizzle). Or is it just that the detail restoration is expected to be in proportion to the (average) SNR hit? |
|
7.56
#...
·
·
1
like
|
|---|
Roman Pearah:andrea tasselli:Tim Hawkes: Happy to oblige, since I can do it and did it already. Here is a side-by-side comparison of the same integrated light, once with and once without Bayer Drizzle:   Note that this is the stacked up light of nearly 4 hours of imaging. STF-ed the same way. |
|
2.97
#...
·
·
1
like
|
|---|
andrea tasselli: Thanks a bunch! To me the Drizzled one looks to have quite a nice improvement in high-frequency detail in the nebula and a loss of fainter features at the extremities. The latter is, in my book, evidence of a lower SNR even though the noise looks smoother. But that smoother correlated noise profile is often the case, even with 2x Drizzle in mono, so it can look a bit like SNR improvement even though you are quite literally cutting the image scale 2x in the 2x Drizzle case (meaning you know you're cutting SNR just on principle). I'd be really curious to compare the total sky time of a mono image shot on the same system and similar conditions that (based on some measurement, perhaps subjectively) seems to match the Bayer Drizzled result. My hypotheses have been something along these lines: OSC Demosaic vs Mono at equivalent SNR would require ~3x (all else being equal) more sky time for mono due to the fact that you aren't collecting channel data in parallel. The mono result, however, who demonstrate better detail. A mono image of equivalent sky time might still be regarded by most as a better image. OSC Demosaic vs OSC Bayer Drizzle at equivalent SNR would require ~3x (all else being equal) more sky time from the Drizzled data. The Drizzled result, however, who demonstrate better detail (as I see above). OSC Bayer Drizzle vs Mono at equivalent SNR would require roughly the same sky time and produce roughly similar detail, i.e. you trade your parallelism advantage for said detail. I'm curious if you disagree with any of these from experience, with the understanding that, in practice, full 3x differences aren't likely going to materialize. |
|
7.56
#...
·
·
1
like
|
|---|
| At the moment I cannot quantify reliably the difference in SNR between the two cases but I'll wager isn't anywhere near 3x and Juan Conejero made that clear in one post on the PI forum. Does drizzle cause a (modest) loss of signal amplitude? It does. Does it make for better images? It does, as far as I am concerned. Would be a mono imager create a better image, for equal effective times (i.e. about 3x times longer), all other things being equal? It would get my nod. Would I want to do it? I won't. |
#...
·
|
|---|
andrea tasselli: No argument about star colours - I use 1x1 Bayer drizzle as well --but I've never noticed andf wasn't aware andrea tasselli:Tim Hawkes: Yes you are right - I had forgotten about Bayer x1 CFA drizzle despite the fact that I use it routinely for my OSC data in PI |
|
2.97
#...
·
|
|---|
andrea tasselli: Did you link to Juan's post earlier? I'd love to check it out. |
|
7.56
#...
·
·
1
like
|
|---|
Roman Pearah:andrea tasselli: I cannot find that comment I was referring to but this makes the case rather strongly and if you can read between the lines... https://pixinsight.com/forum/index.php?threads/bayer-drizzle-instead-of-de-bayering-with-osc.12996/ |
1.51
#...
·
|
|---|
| I’m speaking from limited experience but I have 2600mc with L Ultimate and recently acquired first mono camera 2600mm with Optolong sho 3nm filter set. My location is bortle 8 and comparing same target between the 2 with similar exposure the mono provides more detail and better quality image. It does take more processing time comes down to personal preference. See the rosette images on my page most recent is mono and 2nd is OSC. |
#...
·
|
|---|
Roman Pearah: I would have thought 1.5X rather than 3X ? i.e. the Mono camera needs 3 separate sky times for HA, OIII and SII while the OSC camera needs two separate sky times for HA / OIII and then another for OIII/ SII. |
11.91
#...
·
·
1
like
|
|---|
Is it really being suggested that mono with filters would require 3x or even 1.5x the time as an OSC with a dual narrowband filter? That claim is simply not supported by anything Juan said and certainly not in the link provided.Roman Pearah: This is a misunderstanding of how noise and SNR works. Noise =uncertainty. Repeat the same acquisition again, and the same pixel will have a slightly different value on account of noise. Saying that demosaicing will fill in the remaining pixels with a similar statistical profile is similar to saying that one can fill in the value of the next coin toss based on the value of the current coin toss result. Demosaicing can certainly interpolate the value of an unsampled pixel, but the uncertainty caused by acquiring less light will not be eliminated. Were this not so, one could simply take the result of a mono camera break into four separate channels, and get four times the frames. A detailed statistics based comparison of OSC versus mono for color acquisition has been done here: https://www.cloudynights.com/topic/682340-monochrome-vs-one-shot-color-%E2%80%93-by-the-numbers-please/ For dual narrowband, we are making use of all the pixels on the sensor. At best, the integration times will be identical - neglecting considerations of filter efficiencies and assuming bandwidths are the same. If there is a different conclusion to be made here, I would like to see the statistics behind it. |
|
7.56
#...
·
|
|---|
Arun H: It can surely be said that the probability of the outcome of the next toss of coin is equal to the probability of the present toss landing either a head or a tail, since they are both 50% anyway. |
|
11.91
#...
·
|
|---|
andrea tasselli: That is something I can agree with. The better analogy is estimating the bias in a coin through an experiment involving hundreds of coin tosses. Halving them will impact our uncertainty in the estimate of the bias . |
|
2.97
#...
·
|
|---|
Tim Hawkes:Roman Pearah: I wasn't actually speaking about tri-band narrowband, but the more general case of RGB. |
|
11.91
#...
·
|
|---|
Roman Pearah: That case is also covered in the analysis linked. For pure RGB (not luminance) signal collection, OSCs have an advantage over monos primarily because of the bandwidth of the filters used - the red pixel in a OSC, for example, covers a greater range of wavelengths than a red filter used in a mono camera because it has a much less sharp cutoff. The downside of that is that you become a bit more susceptible to light pollution in each channel. At a dark site, an OSC would definitely have an advantage for RGB signal collection. But that changes once you start accounting for luminance data. Here, the advantage of the mono is huge - and again covered in detail in the link I provided. |
