Mono really is faster

[vineyard]

While I save for a fuller frame camera, the eternal debate of mono vs OSC (+dual band filters) wages on in my head.  And a point often made is that mono is actually faster not slower.  I think I just saw an example of this.

Two images below - the colour is about 4.5-5h of OSC over 4 nights last autumn (w dual-stacked filters to synthesise a dual-band) on NGC 6888.  The red one is only 1h48from last night via a 6nm HA filter and the same camera 294MCPro on the same target.

The difference in data gathered is shocking.  Clearly a tighter NB filter will be helping (vs synthetic dual-band), but this is as a semi-mono w c 60% less capture time than the OSC - so I'd imagine a proper mono would have captured even more data?  And btw this is on full moon night at a time of year when astro dark doesn't exist (even if the moon wasn't out).

Mono really is faster!

[edarter]

I'm no expert but does stacking filters work? Surely the filter which the light hits first is preventing the wavelength of the 2nd filter getting through?

[The Lazy Astronomer]

Not sure whether this will be of interest/useful or not, but, just for comparison, here is a single 3 minute sub with the mono version of your camera and a 6nm Ha filter. All I've done is calibrate, bin x2 and stretch a bit PI:

 

[ollypenrice]

I'm not sure how many people argue that mono is faster, but those who do are correct! It is far more common to hear people argue the reverse, that they 'don't have time' for LRGB so prefer one shot colour. The problem is that 'one shot colour' does not exist. What a so-called one shot colour camera delivers is really more like 'quarter of a shot red, half a shot green and quarter of a shot blue.' 

However, edarter makes an interesting point. I'd have thought that the first NB filter in the system will block the light the second one is capable of passing - unless they overlap.  I'm now off to see if the internet can tell me how dual band filters work...  My initial guess is that the filtration of each passband will be somehow dependent on the other.

Olly

Edited June 15, 2022 by ollypenrice

[tomato]

Unless I’m missing something, I can’t see how stacking narrow band pass filters will work. The first filter is only letting light through of a specific wavelength which the next filter will then block. 
 

However, I do agree mono is faster, but for whatever reasons based on posts on this forum, there does seem to be a lot more IMX571 colour cameras in circulation than mono versions.

[vineyard]

I think it would depend on which filters you stack @edarter & @tomato.  If I had 2 proper NB filters, I wouldn't stack them b/c they'll be working on different parts of the spectrum as you say.

I can't remember where I first read of this double-stacking but I think it was in some obscure CN thread.  Based on the logic described therein, I double-stacked an Astronomik UHC & CLSCCD filter.  I've pasted their respective band-passes below - I think b/c they work on v similar parts of the spectrum (as you surmised @ollypenrice) but w slightly tighter passes in the UHC, the combined effect is a bit closer to a dual-band like L-Enhance (still not as tight as that & nowhere near as tight as an L-Extreme).

The interesting thing is that in solar, double-stacking HA etalons does tighten the bandpass but perhaps that's because of a different physics being used wrt the filtration mechanism?

Thanks too for that image @The Lazy Astronomer - v v helpful.  I'll try and do a Pixelmath comarison later, but at quick glance I think there's more finer detail in that than even w the OSC+NB, which would be another validation of mono?

[edarter]

Interesting thread, and to be clear I'm not arguing whether mono is faster or not, I get the theory behind that and I'm still in a quandry as to whether to go mono or not when I finally purchase a dedicated astro cam (mono would be significantly more money though!). I'm just curious as to how stacking filters can work unless they overlap in their bandpass somehow. Even then though, I would have thought its only the overlapping portion that actually makes it to the camera sensor? Watching for the result of Olly's investigation! 🙂

Ed

[vlaiv]

22 hours ago, vineyard said:

Mono really is faster!

Mono is faster, but your example can't be used to demonstrate that. You can't show that mono is faster by using OSC camera only.

What you have demonstrated is that removal of LP increases SNR even if it decreases signal somewhat - that is the point of using certain filters.

14 hours ago, edarter said:

I'm no expert but does stacking filters work? Surely the filter which the light hits first is preventing the wavelength of the 2nd filter getting through?

With filters - it is pretty simple. You multiply transmission graphs.

It goes like this - for each wavelength, you check out transmission in both (or multiple) filter graphs and read of percentage passed.

Use percentage in 0-1 range (so not 80% but 0.8) and multiply.

If there is even single filter with 0% transmission - whole stack will block that wavelength as X * Y * Z * 0 = 0, no matter what X, Y and Z are ...

Back on topic mono vs color.

Color is faster in other aspects. We must first define what "fast" means.

If we define fast to be something like time taken to reach target SNR given certain parameters (sampling rate, aperture, scope type, etc - pretty much everything except OSC vs Mono) then Mono is always going to be faster. In fact, besides being faster - it is much more flexible.

But, there is other type of speed that we need to consider. Imagine someone not having a permanent setup - and most amateur astronomers don't in fact have permanent setup.

This means that per imaging session for LRGB you need to include 4x focusing, 4x flats.

If flats happen to be of different exposure - that also means 4x flat darks

If one is not very versed in all of that - above can amount to maybe additional hour or so of "not imaging" during imaging time.

Over the course of whole evening in wintertime, or several evenings - that can be compensated by speed of Mono system (time to target SNR kind of speed), but if one is imaging for only 1-2 hours, well - half of that is going to go on focusing and flats and all of that after filter change.

Automated filter wheel improves this speed (not all people have it so less people will see mono as faster)

Automated focuser will improve this speed (not all people have it so again less people is likely to see mono as faster)

Not all have flat panels, ...

It is easy to see why people think OSC is faster.

One more thing - comparing two same sensor cameras in Mono and OSC version is comparing apples and oranges - and often people don't see / understand that.

What you want to do when comparing data from OSC / Mono same sensor is either:

Use super pixel mode to debayer OSC and bin Mono x2 (to get the same sampling rate)

or

Use regular debayering and bin Mono x2 and then scale it up by x2 to again get the same format of data.

It simple fact that OSC samples at half the frequency of Mono and this needs to be addressed when comparing data. It also means that if Mono is properly sampled, OSC will be less sharp for same sensor (nothing to do with antialias filters on DSLR - this is true for dedicated astro cameras as well).

[vineyard]

Thanks @vlaiv - nuanced explanations such as yours is part of why SGL is so rich to learn via.

From my clumsily phrased perspective, I was just really surprised at the difference in the two images.  For my crude mental model, in the OSC+double stack, all the 4 Bayer arrayed pixels are being used for 4.5-5h yet so much of the structure still doesn't come close to emerging - whereas in the OSC+NB (which to my mind means akin to a mono camera with 1/4 of the pixels, albeit with interpolation between the pixels which a pure mono wouldn't need to do) one night gave so much richer data.  Maybe "NB is faster" is a better formulation for what has surprised me.

I hear you on the set-up & flats - people's mileage will vary but for me <2h in one night vs 4 nights to get 4-5h is much more convenient, especially with that much difference in data.  (I'll admit I'm lazy and if I haven't messed with the setup & no big changes have happened wrt dust bunnies, I just use the same flats for a few nights - I suppose slight focus changes night-to-night should mean I should take new flats).

Cheers!

[edarter]

Thanks @vlaiv, I get what you are saying about the transmission. Looking more closely at the transmission graphs in @vineyard post it looks to me that this does work if the CLS filter is the first one that light hits, as this has wider bandpass ranges than the UHC. The UHC's ranges seem to fall within those of the CLS. But doesn't that mean that the CLS is effectively doing nothing? In my simple head the CLS is filtering out a lot of the surplus light but then its further filtered by the UHC. That is the one controlling what gets to the sensor? I suspect there is something going on here I still don't understand!

This is of course also assuming that the graphs are accurate and do not transmit any light out of these bands! If thats not the case and there is a small amount of transmission outside of these bands then all bets are off! 

Ed

[vineyard]

15 minutes ago, edarter said:

But doesn't that mean that the CLS is effectively doing nothing? 

If the Astronomik curves are correct, the CLS is cutting off the long-pass.  The Astronomik UHC seems to be somewhat akin to a long-pass filter above 630nm so the CLS would cut that off at about 700nm.  (The Baader UHC curve seems to have a shut off in it unlike the Astronomik UHC, at least going by Semrock curves: https://searchlight.semrock.com/?sid=a08a1af9-84ee-49d2-959d-153d7e7c0eb8#)

[vlaiv]

4 minutes ago, edarter said:

Thanks @vlaiv, I get what you are saying about the transmission. Looking more closely at the transmission graphs in @vineyard post it looks to me that this does work if the CLS filter is the first one that light hits, as this has wider bandpass ranges than the UHC. The UHC's ranges seem to fall within those of the CLS. But doesn't that mean that the CLS is effectively doing nothing? In my simple head the CLS is filtering out a lot of the surplus light but then its further filtered by the UHC. That is the one controlling what gets to the sensor? I suspect there is something going on here I still don't understand!

This is of course also assuming that the graphs are accurate and do not transmit any light out of these bands! If thats not the case and there is a small amount of transmission outside of these bands then all bets are off! 

Ed

Well, best course of action is to actually digitize those graphs given in images and then put results in say spreadsheet, multiply values and produce resulting graph.

You can use online tool like this to read graph values and save as CSV or similar:

https://automeris.io/WebPlotDigitizer/

It is possible to do crude assessment of what resulting transmission will be over range of frequencies. It is important to note that usually two things happen - first, resulting peak transmission is the same or (more often) reduced. It can never be increased.

Similarly - width of filter pass bands is the same or (more often) reduced. It can never be widened by stacking.

For this brief analysis of stack of two filters - we can divide resulting filter in two regions. Region around 500nm (OIII zone) and Region around 656nm - Ha zone.

CLS filter is simply wider around OIII as it starts around 450nm and finishes at ~520nm, while UHC goes from 480nm to 510nm. Both filters have peak transmission around 95% in this range.

Stack will simply behave as narrower of the two - like UHC filter with peak transmission around 90% (0.95 * 0.95 = 0.9025).

As far as OIII zone is concerned - stack performs little worse than UHC alone

Then there is Ha zone - here both filters start at 640nm - but UHC does not finish after Ha signal - it continues. CLS does drop off just before 700nm and that is a good thing because it removes all wavelengths above that. Again - peak transmission will be smaller then either of filters - at about 90%

Overall stack behaves like UHC filter with IR cut off applied and somewhat lower transmission.

Similar effect can be obtained by stacking UHC with regular IR/CUT filter, or even better - Astronomik L3.

 

 

[vlaiv]

Here is interesting comparison:

http://www.astrosurf.com/buil/filters/curves.htm

Note this graph:

Astronomik UHC is combined with IR cut filter (this is important for imaging, but not for visual as we are not sensitive to light about 700nm anyway).

[Lee_P]

4 hours ago, vlaiv said:

If we define fast to be something like time taken to reach target SNR given certain parameters

Just building on this comment, I think that if you want a "fast" system (being able to gather data quickly), then I really recommend making your equipment set-up and pack-down times as speedy as possible. If it's quick and easy to set your kit up, you're more likely to actually do it and then gather data. This might seem like a tangent to the OP's queries about mono / OSC, but I'd argue that it can have a bigger impact on how fast you produce images. As a reference point, I use a DIY pier, small refractor and ASIAIR Plus; it takes under five minutes for me to get imaging, and in the last 12 months I've produced around 20 images with integration times in the range of 16 - 24 hours. And this is from cloudy South-West UK.  

[The Lazy Astronomer]

And for those of us without a permanent pier, maybe consider some weight training 😅

Seriously though, l don't consider myself particularly strong, but I leave my eqiupment permanently set up and just move the whole lot in and out of the garden in one go (100mm refracter on an eq6r*) - quick polar align in Sharpcap and l can be imaging in 10 minutes.

*side note: what does the "r" denote? It seems like a naming structure oddly borrowed from the automotive industry

[vlaiv]

10 minutes ago, The Lazy Astronomer said:

*side note: what does the "r" denote? It seems like a naming structure oddly borrowed from the automotive industry

Rrrreally good version?

R stands for version with belt fitted out of the box, but why did they decide for R moniker is beyond me.

[saac]

I've never really thought about the speed angle OSC v mono, honestly never really occurred to me. I use OSC primarily due to the paucity of consecutive clear nights.  If I get a four hour image run one night I count myself lucky; the next clear night may be the following day but more likely a month away.  I also prefer the simpler setup, less equipment and simple workflow of OSC.  Definitely a personal thing I think.  Interesting points though. 

Jim

[ollypenrice]

7 hours ago, vlaiv said:

 

This means that per imaging session for LRGB you need to include 4x focusing, 4x flats.

 

Not necessarily true in practice. My experience suggests that the need to refocus comes from the optics, not the filters. This just means that you can refocus per colour with mono and can't with OSC. It also means that you can't when shooting luminance - at least, all you can do is compromise somewhere across the colour correction of the optics. Anyway, being a hare out of focus is really no big deal in RGB. After all, you can bin colour and still get a sharp result. Likewise I rarely find that there's any need to shoot flats per filter. It's possible to get a dust bunny from a contaminant on one filter but I've found over thousands of hours of imaging that I very rarely do. Nearly all my imaging is stacked using the luminance flat for everything. In reality it works.  Because I don't use robotic focus on my own rigs I sometimes just scroll LRGB, LRGB, at the luminance focus and that works fine, though I do prefer to optimize the altitude when shooting L and blue if the target is low enough for this to make a significant difference.

The book says flats and focus per filter, I know, but sometimes it can be safely ignored. (I know you won't like this! )

Olly

Edit, I agree that OSC is a boon for the heroic mobile imager...

Edited June 15, 2022 by ollypenrice

[vlaiv]

1 minute ago, ollypenrice said:

dust bunny from a contaminant on one filter but I've found over thousands of hours of imaging that I very rarely do.

Because you use permanent setup.

Much more likely if you take your setup apart after every session.

[ollypenrice]

2 minutes ago, vlaiv said:

Because you use permanent setup.

Much more likely if you take your setup apart after every session.

Possibly, though with an electric wheel how would dust get in if the camera is never removed? I'd have thought this possible for many.

Olly

Edited June 15, 2022 by ollypenrice

[vlaiv]

2 minutes ago, ollypenrice said:

ossibly, though with an electric wheel how would dust get in if the camera is never removed? I'd have thought this possible for many.

Say you have multiple scopes that share filter wheel and camera, or perhaps you have 5 position filter wheel that you use for both LRGB and narrowband imaging and need to change filters from time to time.

[The Lazy Astronomer]

8 minutes ago, ollypenrice said:

Possibly, though with an electric wheel how would dust get in if the camera is never removed? I'd have thought this possible for many.

Olly

 

4 minutes ago, vlaiv said:

Say you have multiple scopes that share filter wheel and camera, or perhaps you have 5 position filter wheel that you use for both LRGB and narrowband imaging and need to change filters from time to time.

Or surely an open tube design like a newt or RC would be liable to collect some new dust motes from time to time? 

[ollypenrice]

10 hours ago, vlaiv said:

Say you have multiple scopes that share filter wheel and camera, or perhaps you have 5 position filter wheel that you use for both LRGB and narrowband imaging and need to change filters from time to time.

True.

Olly

[Macavity]

For we Imaging Dabblers, the Bi-Color (UHC) images of e.g. the Veil Nebula
are (neon) "electrifying"! That would have been (may yet be) my way to go!
Then there's stuff about using "Synth Green" etc. But that's "another thing"? 😏

For some reason, the idea of double... or triple (fairly) narrow band filters
appeals to me. That said, much vaunted, "filter design" is... serendipitious? 🥳

[rickwayne]

My personal experience is that the ratio of acceptable-to-me subexposures increased dramatically when I started doing mono, and the flexibility is VERY appealing -- HaLRGB, RGB only, LRGB, etc. To say nothing of simply being able to collect good data quickly in my Bortle 7 back yard, or in full moonlight. It is more of a hassle -- all kinds of little things. The filter wheel makes it a bit more challenging to work with my OAG. I have to clean the things. Imaging sequence construction in Ekos is a little more complex. Processing, likewise. But overall, I am super glad that I made the jump. If nothing else, I got a couple years' worth of higher-quality imaging that would have been difficult to achieve with the cameras that I could have afforded when I started. The new sensors make OSC much more attractive to me.

I wish I could do without per-filter flats, but my sadly-abused LRGB set (I got it used -- I did not do the abusing!) eliminates that option for me. There are ALWAYS artifacts that need removal, and I cannot remove them from the filters. 

Concur on the focusing, I haven't been bothering with per-filter focusing with my new scope, for which I haven't yet built an autofocuser. I do check with each filter change but so long as I nail the luminance or red focus, it's OK for me. But since thermal effects will mandate at least some refocusing during the night anyway, I'm going to refocus per-filter again when I get a rig going. Heh -- one advantage of an f/8 scope over an f/4.5 is that it's a little less psychotic about its critical focus zone. (The people with Hyperstars and RASAs must have WAY more patience than I do.)

Assuming the autofocus works, I lose maybe 2 minutes per iteration. If I time filter switches to my temperature-dependent refocusing, that's two minutes per hour. Acceptable. Likewise flats, I'm running 20 frames per filter and each takes about 3-5 seconds to download, so 100" x # of filters used. Something like 7 minutes, plus the fiddle-faddle of getting the scope pointed at the sky and the LED tracing panel connected. Call it 10. I could modify my teardown workflow, take the scope off the mount first so that I can pack up the mount while I'm shooting flats, that would be a tad more efficient I guess. For me it's cheap insurance against having to Photoshop some weirdness out of my subs.