Is SII worth the bother!

[MartinB]

This discussion arose in GordonH's Soul neb thread http://stargazerslounge.com/index.php/topic,34744.0.html

I think Gordon's image is excellent for all sorts of reasons (way better than my low res Rosette), but I'm still not convinced about SII

Some different views were expressed. However, nothing illustrates an imaging problem like an image. Here is a narrow band Rosette of mine from a couple of years ago. It was taken using an SXVH9 and a ZS66 with a reducer making the scope around F4 270mm focal length. 13nm Astronomic narrow band filter 7x10mins for each channel. SII - red Ha - green and OIII blue.

(click to enlarge)

There are lots of "problems" with this pic quite apart from the narrow band issues and I'm itching to have another go this winter. I just have to decide what combination of filters to use. So here is my narrow band problem

This image has equal time given to each channel. If you combine the channels before stretching the Ha completely swamps the other 2 channels. OIII has a bit of influence SII zilch other than helping with the star and background colour. I can pull out the OIII blue fairly easily by stretching the blue channel and this works quite well.

I cannot get any sensible contribution from the SII channel. I can stretch it hard and tone down the Ha but then up comes the noise and tell tale magenta background and star colour. With an image mangling effort I can get the SII to show up in some small areas of the nebula. I have applied star layering techniques and differential stretching, background bias correction, the works but the bottom line is that to get the SII contribute anything requires dumbing down the Ha channel noise wise to match the SII.

I have had identical experiences with other targets such as the Bubble and Elephant's trunk.

The image above shows the best I could come up with at the time. As you will see, there is no red SII nebulosity! I haven't posted it but the best image I could produce with this data was using just the Ha and OIII and create a synthetic green. By adjusting the hue you can come up with a hubble palate if you want to.

I have seen images from Rob Gendler which do show small wisps of SII despite not using over long exposures in the SII channel but the overall wow factor of those images still doesn't come from the SII. The fantastic narrow band images on the web have hours and hours of exposure time, typically more than 10 hours of SII. It would probably take me a few winters to manage that!

So if I do go for the Rosette in narrow band I am thinking of exposing SII:OIII:Ha 4:2:1 but is it really worth sacrificing all that time to get a few wisps of pink? I might just stick with Steve Cannistras bicolour technique http://www.starrywonders.com/bicolortechniquenew.html

My mind isn't made up though, am I missing a trick? Are there any SII rich targets out there.

[barkis]

Is the SII filter a disappointment to you Martin, because it seems to lend nothing to Narrowband colour simulation.?

I thought planetary and gaseous nebula benefited from SII in some way.

I ask simply because the filter is one I bought along with OIII and Ha. I didn't get the Hb filter because I thought having the

SII would do. Perhaps I was a bit naive buying without advice, because I based the selection more on what I was gleaning

from posts on SGL, rather than any personal knowledge of my own. :oops:

Ron.

[narrowbandpaul]

i also use the SXV H9, with fairly fast scopes <f6.

Douglas and I's greatest success was IC1795 in S[iI] Ha O[iII], with 2 hours S[iI] 1 hour Ha 1 hour 40mins O[iII]

with 20min subs, and 4 dark frames, taken with an FS128 at F6.

long subs yes, but they were at f6....

Richard crisp uses at least a ratio of 2:2:1 S[iI][iII]:Ha, because the Oxy and Sulphur data will! require more stretching to boost the signal to somewhere around that of Ha...

try using S[iI] again but with a ratio of at least 2:2:1.

the way I look at things is that Ha is a background, 'uninteresting' species, on top of which you add compression information (S[iI]) and high energy (hot) information (O[iII]).

The image, I will try and post later, has a good SNR in S[iI] but not quite as good in O[iII], but reds and blues are clearly seen, meaning each filter is being represented.

In theory the S[iI] and O[iII] look nice, giving a 3d image, i think. But the reality is you need to focus your imaging time on good subs with O[iII] and S[iI], because, lets be honest, Ha is a doddle to get a good signal for (its really abundant)

you can almost ignore Ha, put it to the back of your mind, focus your imaging talent on good SNR in both S[iI] and O[iII]...

so good darks help....especially when trying to stretch the sulph and oxy data.

Flat fields help too, but you may not get enough signal to enter the FPN limited noise regime, so flats may only correct vignetting (which is good) and may not give the SNR increase to the image, because your signal is in the shot noise regime (which is good)...but thats a different point.

If you have flats...definately use them, they may very well give an SNR boost since FPN can kick in after only 500e- signal.

So flats and darks will reduce noise...

and double the number of subs for sulph and oxy will boost signal, as will longer subs say from 10mins to 15mins.

2:2:1 for S:O:Ha should work well, with darks and flats...

I honestly find O[iII] a bit weaker than S[iI]...but they are both weak of diffuse neb, no doubt.

Barkis...i personally think you made the correct choice with regard sulph rather than Hb...

A nebula can have a certain brightness in ha, a certain brightness in O[iII] and a certain brightness in S[iI]...each brightness is arbitrary...one nebula might have a weak sulphur, but strong O[iII] or vice versa, or they can be both strong or both weak...

moreover the brightness of the S[iI] and O[iII] lines compared to the brightness of ha is completely arbitrary...that is...they are uncorrelated...the brightness of one line doesnt give any information about another.

In this way each filter is unknown when you image it....it could come off bright or faint...and so the nebula and have an arbitrary look to them...lots of blue, or no blue, lots of red, no red completely arbitrary....

but if you use Ha and Hb for the same object...the abitraryness is gone. When you image in Ha, you will instantly know where the Hb is and how bright it will be. This is because, the Hb is coming from the same stuff as Ha (namely Hydrogen) so has the same location, the brightness is 2.92x weaker (approx) because the transition for Hb is less probable.

So they are correleated species...in this scheme images tend to look the same...mostly red since Ha is 3x stronger than H. Some people prefer the look...thats fine.

I personally dont like having two correlated species in a NB image. Hb in itself is fine, but not used in conjunction with other hydrogen emission lines...

to have 3 independant emissions, the elements must be different. Eg Sulph oxy and Hydrogen... or Hydrogen, helium, oxygen...its all good, brightnesses completely independant...you dont know whats coming before its arrived.

Perhaps i will get some stick for my Hb comments, but I really dont believe in Hb when used with Ha, so I'm out.

As for good S[iI] objects...i would suspect supernovae remnants as they create all the heavier elements, so i would expect to find some decent sulphur.

Thats my two pennies...

Basically give sulph and oxy double the exposure than Ha, longer subs if possible, but 10 is OK for F4, and some darks, and flats if you have them....

I will try and get the image of IC1795 up later, but am at uni just now.

note though that the oxy was fainter than sulphur (should we ditch oxy?...I like the look of objects with oxy and sulph filters), but there was 20mins less exposure time thanks to clouds...grrr

Paul

[MartinB]

Ron, the only way for you to find out whether the SII is for you is to try it and see. I'm coming to the conclusion that you may be able to enhance your images with OIII but only if you are prepared to spend a lot of time collecting that smidgeon.

Thanks for your reply Paul. As you will see from this Rosette image, the SII signal doesn't come close to the OIII. This has always been my experience but I'm sure there are targets that differ.

Flats, darks, fixed pattern noise, chip QE, I feel pretty well up to speed with all that stuff. I've also considered band shift at fast focal ratios but the problem still applies at F7.

I would be really interested in seeing your individual SII OIII and Ha data channels along with the overall exposure time info.

[NickH]

OIII absolutely. I have H-Beta and SII (coming in a few days), as some objects will do it justice, and I got fed up synthing channels in CS2, but agree totally that the 2:2:1 rule of thumb should be adhered to.. with SII, OIII and H-A

All depends on the objects I guess... but that's a good rule anyway

[barkis]

Thanks guys, a little more knowledge gathered.

Ron.

[MartinB]

Nick, why do you feel 2:2:1 should be adhered to? The Rosette above clearly shows there is much stronger signal in OIII than SII so my thinking is 4:2:1 would be more balanced.

[GordonH]

Hi

Just as an after thought with the 2:2:1 for SII:OIII:Ha, sometimes you may have to consider as much as 5:3:1 or different depending on the ratios of the 3 emission lines in a given subject. As to Martin's question as to whether it is worth it or not, that is a very difficult question to answer. If you have a remote controlled system high up where ther are a plethora of clear nights then it isn't too much of a problem, if you are imaging in the UK then it is a different issue. The only answer I can suggest is give it a try ie try for really long exposure time even if it is over several weeks, months etc and if the end result is worth it then so is the time put in. Don't forget that the Mona Lisa wasn't painted in a day.

I am certainly going to give this a try with a subject at some point to see what is achievable

Best wishes

Gordon

[NickH]

Nick, why do you feel 2:2:1 should be adhered to? The Rosette above clearly shows there is much stronger signal in OIII than SII so my thinking is 4:2:1 would be more balanced.

Sorry i meant adhered to as a kind of baseline, other objects will need different ratios, but 2:2:1 as a base rule of thumb I think is good

[MartinB]

Thanks Nick.

Gordon, what constellation is Mona Lisa in?

[narrowbandpaul]

a min of 2:2:1...the more the better for those weak oxy and sulph channels.

[narrowbandpaul]

for the IC1795 6*20mins S2, 5*20mins O[iII] 3*20mins Ha...total 4 hours 40mins. at f6....tak fs128

will try and post 2nite.

tis our best image so far.

[GordonH]

Gordon, what constellation is Mona Lisa in?

Hi Martin

I think it is in Leo(nardo Da Vinci) :laughing3:

Best wishes

Gordon

[narrowbandpaul]

as promised here is a tricolour nb image of ic1795 with contribution from all 3 channels. exposure as above.

[narrowbandpaul]

(click to enlarge)

[barkis]

Well that looks a bit classy to me. I just wish I could give an appraisal worthy of it, but lack of experience on my part excludes that. It obviously has all the attributes associated with the data streams captured, and processed. I'm sure more informed officionados will follow and give you a better response Paul. I do like it though.

Ron.

[narrowbandpaul]

thanks ron...

it does show a contribution from each filter

[MartinB]

That's a great image Paul. There is definitely SII and OIII in there indicated by the yellow outer araas and also a hint of pure SII over to the right. I also agree that there is a nice 3D effect. Certainly an image to be proud of.

Casting a very close eye over the image the Ha areas show better s/n ratio than the OIII SII and some of the stars are showing magenta which suggests the SII has been pused and the Ha held back when stretching the histogram. Would you be able to post the individual channel images? I'm interested in seeing how the s/n compares.

Hi Martin

I think it is in Leo(nardo Da Vinci) :laughing3:

Best wishes

Gordon

[narrowbandpaul]

if you do an S/N for each of the filters I would be very interested...

presumably you want the unprocessed, but darked and stacked, data?

if you want fits data..better for analysis, then I can email them to you

yes the S[iI] and oxy have been pushed...but thats normal unless you have like several more hours

[MartinB]

yes the S[iI] and oxy have been pushed...but thats normal unless you have like several more hours

That's the whole issue Paul and the difficulty with narrow band imaging in the UK. I'll pm you cos I would love to have a look at a decent narrow band data set.

[RichieJarvis]

A very interesting topic...

A question for you though - as I am new at this ere narrowband imaging. I only have Ha and OIII at the moment. I've been doing exactly the same length and number of subs for both filters at the moment.

It sounds like the advice for this combination should be 2:1 OIII:Ha, right? But does that mean 2 x exposures for each Ha exposure of the same length, or doubling the length of the OIII exposure?

Should it be:

2 x 5 minute OIII

1 x 5 minute Ha

or

1 x 10 minute OIII

1 x 5 minute Ha

Thanks from an apprentice imager!

Cheers,

Richie

[narrowbandpaul]

shouldnt matter, but keeping the subs the same length is better for ease for using dark frames

[GordonH]

Hi

I asked this same question to Richard Crisp a while ago and his view was that if the bandwidth of the filters are the same ie if they are all 13nm then the exposure length is the same for each subframe and you take more subframes for the SII and OIII than you would for the Ha so for example if the suggested ratio was 5:1:2 for SII:Ha:OIII (HST palette) on a given subject then typically the exposures would be 30x10 minutes for SII, 6x10 minutes for Ha and 12x10 minutes for OIII. Having more subframes for the SII and OIII allows them to be stretched more before noise becomes apparent. I am not an expert Paul, but I think it does make a difference when you use longer subframes as opposed to more subframes of the same length as using longer subframes increases the signal whereas more subframes does not increase the signal, it increases the signal to noise ratio which I believe is different

Best wishes

Gordon

As an afterthought I was thinking about this and why not try for longer subframes for the weaker channels, what is the worst thing that could happen, I mean you could end up with a really aesthetically pleasing image through experimentation (this is what I believe astroimaging is all about, especially when we are dealling with wavelengths and colours that we cannot see visibly through a telescope) so does it really matter what colour palette we end up with as long as it looks nice to the viewer. Again I think I am going back to the point of art, give ten different artists a subject to paint with identical materials and I can guarantee you the result will be ten different paintings, much the same as astroimaging. Wouldn't it be boring if we all came up with identical images of M42?

Best wishes

Gordon :wave:

[narrowbandpaul]

the SNR of an image increases with root N (N= no of mages) and also increases with root t (t= length of sub) as long as you are in the shot noise regime. If you are FPN limited then averaging frames and increasing sub length wont do much...

but for the low signal experienced when imaging the weaker channel, the camera 'should' be shot noise limited...

thats my understanding

however, I do advocate longer subs if possible...

our average is 15-20mins subs...which works quite well

Paul

[RichieJarvis]

I am doing 5 minutes and 10 minutes atm.....

I don't think its quite as clear tonight

Maybe i should reduce my sub length?

Heres an example OIII frame - what do you think?