"Double duo-band"

[dx_ron]

A CN poster made a nice SHO image using an OSC camera and two different duo-band filters: https://www.cloudynights.com/topic/8309 ... p=12000143

The first filter is a normal Ha/Oiii. The second is Hutech's IDAS NB3 (https://www.sciencecenter.net/hutech/id ... ex-nb3.htm), which passes Sii and Oiii. So the 2-filter process is similar to the couple of people who have used a duo-band followed by an S filter, with the difference being that the blue and green pixels are collecting additional Oiii data while you capture the S.

I haven't decided to try this yet, but I'm curious if one could properly present the stacks to ST, with Ha and S each coming from just one, but O coming from both stacks. (I also want to see some results from something not as blazingly bright as NGC 7000 before I think about jumping into this)

[Mike in Rancho]

Nice. That is just the kind of crazy stuff I like. And often do - all the time! Once I get my mono, EFW, and a full complement of filters all situated, I likely won't as much. But it's still fun stuff.

I remember a much more in-depth convo we all had on CN regarding this issue. Included therein was some discussion of some special and different extraction methods offered in Siril (and maybe APP). Sounded almost like a single channel superpixel followed by a drizzle. Honestly I still don't buy into it, but that might be another way about things you could explore, particularly as a Siril user.

There are a number of caveats and limitations to this, which I guess I can save to the end, but...hey things might still turn out good enough, right?

Here's how I would go about it (you can use ST, but for the main extractions I would use Siril or ASTAP, to maintain 32-bit precision):

1. Stack the duoband containing the Ha band. Call the result "pre-Ha."
2. Stack the duoband containing the SII band, to the same reference as the Ha. Call the result "pre-SII."
3. Stack everything, again to the same reference as previously. Call the result "pre-OIII."
4. Extract the red channel from pre-Ha, and call it "Ha."
5. Extract the red channel from pre-SII, and call it "SII."
6. Extract the green and blue channels from pre-OIII, and call them "pre-OIII-G" and "pre-OIII-B."
7. In ST, composite pre-OIII-G and pre-OIII-B into the G and B channels, and select L+Synth L from R(2xG)B, mono. I'm a little hazy on color ch interpolation being on or off. I don't think it will make a big difference though for this file. Hmmm? I believe Ivo mentioned before that it just puts a little something into the open channel(s). I don't believe it will greatly affect the creation of this blended 2xG and B mono file. Anyway, save that file as "OIII."
8. Now you can composite the three mono SII, Ha, and OIII files into ST. Query whether you want to weight the Synth L, either per exposure times or artificially to give a boost to something that may need it, like SII. Of course SII may be lower SNR and so that could be increasing some noise. Same with OIII which of course also tends to be filled with some extra noise - especially if maybe an L-eNhance was used.

Caveats and concerns:

a. Not always, but sometimes stuff like this ends up requiring a separate Wipe for each channel, due to gradient differences (or in cases I had myself, flats calibration issues too). Don't do an initial AutoDev (or revert) so that after Wipe you save the linear file.
a1. If you do a, you may also need to pre-crop and re-save all the files (in order to get Wipe to work) to eliminate any stacking/alignment edges.
b. Unfortunately at this time (hint hint! ) ST only saves in 16-bit TIFF, so when then recompositing the 32-bit precision is lost I believe. Alas. But I've still come up with nice images. And if you go straight to composite of SII, Ha, and OIII from above, the loss to 16-bit only affects the OIII.
c. You would want the two duobands to have identical OIII passes, or at least really really (really) similar.
d. There will still be Bayer bleed-over or contamination. Of course we always deal with this using OSC with duoband. However, from my own experience and discussions I've had with Norm and others, the extracted Ha and OIII files are not usually going to be heavily affected by this. You can see this by looking at the camera's channel response curves and comparing or overlaying it against the duoband pass charts. Ha is usually the strongest emission, but its contamination of the B and G channels tends to be at a weak spot in the response curves, so you can get generally clean OIII. Meanwhile, Ha is going to dominate the R channel, so the OIII contamination therein will also be relatively weak. However, with relatively weak SII, there is much more of a chance of OIII contamination in the R channel to be extracted. This is not the case of course with a discrete SII filter that is then R channel extracted. But that's just theory - I don't have a duoband with SII in lieu of Ha, so have never experimented to see what the reality of the data ends up being.

That's all I can think of at the moment. And of course YMMV and different targets are likely to be either very amenable to this technique, or perhaps not so much.

[dx_ron]

Thanks for the very detailed description. Quite the dance - but it should still all not amount to much time compared to how much time is invested in collecting the data.

I haven't found the passband specs for the NB3 filter. The likeliest way to match the Oiii passbands and response would be to pair it with the same company's NB2 Ha/Oiii filter - but I just ordered an Antlia ALP-T (which has 5nm passbands).

Will I try it? Not sure. Should I go full mono? Yeah, but - and you know full well what the 'but' is.

[hixx]

Hi dx_ron,
Using different Filters from different vendors has another problem: glass thickness '& refraction index become important. Depending on the optical train, the focussing distance might differ. If an APO or Newton with flattener/corrector was used both filters may have slightly different working distances resulting in different focus points and slightly geometrical distortions. The filters should be homofocal instead. I would think they should be one Vendor preferably.
However on an OSC, you could achieve the same result using just the Ha/O-III duo plus a seperate S-II No need for the S-II/ O-II duo, because the O-III signal is least prone to noise.
clear skies,
Jochen

[Stefan B]

However on an OSC, you could achieve the same result using just the Ha/O-III duo plus a seperate S-II No need for the S-II/ O-II duo, because the O-III signal is least prone to noise.
clear skies,
Jochen

Hi Jochen,

are you sure about that? I've always heared that OIII is the one you should shoot during new moon since it is most affected by light pollution and most prone to noise.

Due to that using a second dual nb filter and doubling time on OIII appears like a smart idea to me.

Regards
Stefan

[Stefan B]

Hi Mike!

3. Stack everything, again to the same reference as previously. Call the result "pre-OIII."

Great idea, but maybe not so easy... You would need separate flats for each filter I guess. That's not a problem in DSS with the different tabs containing sets of matching lights and calibration frames but in ASTAP that's not possible as far as I know. You would have to precalibrate the light frames, right? That's certainly increasing the pre-processing time. Or do you have another idea?

Regards
Stefan

[hixx]

Hi Stefan,

are you sure about that? I've always heared that OIII is the one you should shoot during new moon since it is most affected by light pollution and most prone to noise. Due to that using a second dual nb filter and doubling time on OIII appears like a smart idea to me.

Yes You are right about the moon, which affects shot noise. My point was that O-III band is usually a stronger signal hence having a better SNR. Of course with moon or under light polluted skies more integration time might not hurt as You said, but my statement was pertaining to reasonable conditions. But even doubling the number of O-III frames using the S-II / O-III duo would (on paper) cater for not more than a factor of 1.4 in terms of SNR, In reality it's usually a bit less. I do not say this method is bad, I just doubt it adds a lot bang for the buck. At least the IDAS douband is not cheap. (I do have the NB-1). For this purpose, I'd suggest the NB-1 & NB-3, but not pairing it with the Antlia ALP-T, which has a thickness of 2mmm vs. the 2.5mm Of the IDAS filters. You'd be way off focus and correction. Another problem with the NB-2& NB-3 is, they do not block infrared, so You'd need an extra IR cut. You're probably cheaper with a set of homofocal Ha/O-III duo and an S-II filter by the same vendor.


cheers,
Jochen

[Mike in Rancho]

Hi Mike!

3. Stack everything, again to the same reference as previously. Call the result "pre-OIII."

Great idea, but maybe not so easy... You would need separate flats for each filter I guess. That's not a problem in DSS with the different tabs containing sets of matching lights and calibration frames but in ASTAP that's not possible as far as I know. You would have to precalibrate the light frames, right? That's certainly increasing the pre-processing time. Or do you have another idea?

Regards
Stefan

Hi Stefan,

Yeah perhaps it is a bit clunky compared to DSS, but ASTAP does have a "classify by..." checkbox. Date is the one I would generally use, and I think in most cases you would probably be using one duo filter one night, and then the second duo another night. I'm not quite sure at what point ASTAP decides to change the "date," but I haven't had a problem with one session going past midnight on one night, and then another imaging session being the next night (but still technically the same day, before midnight).

ASTAP can do this for both flats and darks.

If you can pull it off, it still seems a reasonable idea to me. Nothing wrong with doubling the OIII, which may be stronger than SII, but in almost all cases will be far weaker, and lower SNR, than Ha. Plus that teal area seems more prone to picking up junk, and thus could use some more integration. And I'll also throw in that one might be more likely to acquire more time on the OIII-SII duo, in order to try to pick up the weakest SII, compared to the OIII-Ha duo. Thus, you would not be matching the Ha exposure and thus "only" be getting a 1.4x OIII improvement.

I'm also not quite following the focus issue. I mean, one always has to set the focus, and it is wise to check with any filter change. Depending on the optics, all duobands run some risk of red and blue not precisely focusing either or both bands.

Edit: Yes, precalibration is another option. But ASTAP isn't necessarily optimal for any of these situations, due to the inability to select your own reference. You can register afterwards though, even though that is not best practice as Ivo has told us.

[dx_ron]

Hi Stefan,
At least the IDAS douband is not cheap. (I do have the NB-1). For this purpose, I'd suggest the NB-1 & NB-3, but not pairing it with the Antlia ALP-T, which has a thickness of 2mmm vs. the 2.5mm Of the IDAS filters. You'd be way off focus and correction. Another problem with the NB-2& NB-3 is, they do not block infrared, so You'd need an extra IR cut. You're probably cheaper with a set of homofocal Ha/O-III duo and an S-II filter by the same vendor.


cheers,
Jochen

The NB3 isn't "cheap", but it is quite a bit less expensive than the L-eXtreme or ALP-T.

The thickness of the NB3 also needs the thickness of the 2nd UV/IR blocking filter added. And you need a way to mechanically stack filters, which I do not currently have. But, like Mike, I don't understand why it matters if they are not parfocal with the ALP-T. The NB3 would have separate flats anyway (and Siril has no trouble with that sort of separate calibration).

[Stefan B]

Hi everybody,

Yeah perhaps it is a bit clunky compared to DSS, but ASTAP does have a "classify by..." checkbox. Date is the one I would generally use, and I think in most cases you would probably be using one duo filter one night, and then the second duo another night.

Oh yes, you're right, Mike, I forgot about that. But I use the same flats for half a year or so. Thus, it won't work. Will have to check what other options are available. If I understand this (https://www.hnsky.org/astap.htm#operation) right, you can classify by filter. So that could work...

I do not say this method is bad, I just doubt it adds a lot bang for the buck. At least the IDAS douband is not cheap. (I do have the NB-1). For this purpose, I'd suggest the NB-1 & NB-3, but not pairing it with the Antlia ALP-T, which has a thickness of 2mmm vs. the 2.5mm Of the IDAS filters. You'd be way off focus and correction. Another problem with the NB-2& NB-3 is, they do not block infrared, so You'd need an extra IR cut. You're probably cheaper with a set of homofocal Ha/O-III duo and an S-II filter by the same vendor.

Mmhhh... I have the Optolong l-eNhance. The Optolong SII is available for 230 EUR and the NB3 for 170 EUR. So I would save 50 EUR AND get more signal on OIII. With a SII single bandpass filter I'd waste the blue and green light and three quarters of the sensor. Using two dual band filters still seems like a reasonable and economic idea to me.

Homofocality isn't very important to me since I do not have a filter wheel and have to disassemble the image train during switching anyway. And I have to refocus then of course.

Regards
Stefan