20 replies to this topic

[gerdastro]

For a multi night project, is there any strong reason not to finish all intended imaging for one filter first before moving on to the next filter in a given night?
A simplified example would be: night 1: only OIII -- night 2&3: only Ha -- night 4: only SII -- night 5: only OIII

 

If instead every night I would take 1-2 hours each Ha, OIII and SII, I would lose more imaging time to EAF refocus steps after each filter change.

And flats would need to be taken for all filters used, which can also eat into valuable imaging time and on top creates many more calibration files. This saving of imaging time/disk space is even more severe for LRGB, leave alone LRGB+Ha.

 

Additionally, concentrating on fewer filters per night allows for better specialization: e.g. use the moonless nights for OIII exclusively instead of having to 'waste' it on adding Ha which one could take any time.

Lastly, I like the simplicity of only having to think of 1, max 2 filters per night - less things to confuse and thus less errors to be made.

 

In fact, I have changed to minimizing filter changes within a night since summer and not noticed any issues, but without long term experience I lack a reference to compare to in order to recognize potential drawbacks.

 

 

At present I can only think of two problems:

 

Not spreading time evenly over filters, one might end up not having signal for a particular filter for a long time
Personally, for me that is a non-issue. If the project is interrupted by clouds - even for months - while annoying, I can live with that and just continue at a later date or even the next season.

 

Ending up with very unbalanced signal acquisition between filters due to changing conditions
This point I can not judge and would love to hear some opinions. Obviously conditions can change between nights. If one was to take e.g. R and G in one night with good conditions, then B and L under slightly different circumstances … would it lead to any significant issues, or should it be fine as long as the differences are not dramatic?

[David R.]

Just curious - are you using NINA? I use filter offsets which really speeds things up. 

 

Regarding your last question about unbalanced signal due to conditions, this was the only thing that made me go "hmmm..." - I am also curious what folks have to say on this. 

[daveco2]

I like to average light pollution, moon illumination, and sky glow over all filters evenly. So, I rotate filters throughout the night as the target transits.  My usual protocol is 300 sec exposure, 12 subs in Ha, then 12 in Oiii, and 12 in Sii, for a total of 3 hours.  Then repeat.  The order may change so Oiii and Sii are taken more toward the zenith where the signal is strongest; and the usually much stronger Ha taken at more of the slant range where transmission is poorest.  I don’t know definitively if any of this makes a difference, but it makes sense to me to continue doing it.

 

I think this scheme also gives the best chance of getting signal in all channels before weather moves in during the night or blanks everything out for the next few days or weeks, which has happened more than once.  I like to check signal levels and normalization on even a partial data set after the first night’s data acquisition.

Edited by daveco2, 14 November 2024 - 01:53 AM.

[AstroVagabond]

I'm a NINA user and let the Target Scheduler plugin by Tom Palmer manage and run all my imaging. If I'm shooting narrowband my sequence looks like this:

 

Red, Red, Green, Green, Blue, Blue, Ha, Ha, SII, SII, OIII, OIII, Dither

 

I like to capture RGB for stars (20 minutes per RGB filter at 60 second exposures) and then 300 second exposures for narrowband.

 

I prefer as you can see to rotate through the filters vs. dedicating one filter per night.

 

Each imager should use an exposure plan scheme that best fits their needs.

[gerdastro]

I am using an ASIAir, so no filter offsets. It definitely sounds logical to me that an even signal collection over all filters will get more balanced results - but I really do wonder whether generally, this would be detectable assuming conditions do not change extremely. Currently I am still assuming in most cases it would go unnoticed.

 

As already mentioned, for each filter used flats have to be taken. I had a few very short nights of just 2-3 hours in the beginning of the hobby, for which I then ended up with about 45 min each Ha, SII and OIII - in other words, 9 subs each. Still, 30 flats each had to be taken. It just feels so inefficient and while nowadays storage space virtually costs nothing, I still prefer to be lean in my process, including the amount of necessary files.

 

Then again, if there is reasonable evidence that it makes a noticeable difference for the resulting image quality, obviously I do not want to compromise on that.

[psandelle]

I’m a pessimist, so a lot of time I shoot a single narrowband filter per night, Ha first. That way, if I only get one night of imaging (weather, mechanical, a bear eats me), I will have a great image. A second night I get my OIII, then I have a beautiful bi-color image. A third night, if I need SII, voila; if not, I can pick up more Ha and OIII, and my RGB stars (which I try and get whenever). So, great image of a different kind regardless of the nightly circumstances. Of course, if I ONLY have one night, I try and shoot them all and hope at a later date I can collect more data. And thus the pessimist wins! 

 

Paul

[WadeH237]

Are you dithering?

 

If you are dithering and guiding, keep in mind that the time to re-acquire and center the guide star after a dither takes time.  I am a strong advocate of dithering on every exposure in a stack, unless I have multiple hundreds of subs per channel.  In that case, I may dither after two or three subs per channel.  I'd also note that I am not personally guiding, so my recover times after a dither are pretty short, but still there.

 

If you rotate through all filters, you might actually save time versus imaging repeatedly through the same filter.  Here is a short example of what I mean.  Assume that you are doing an LRGB image and taking 3 subs per channel.  If you image all the luminance first, then all the red, then all the green, and finally all the blue, then your sequence looks like this:

 

L, <dither>, L, <dither>,L, <dither>, R, <dither>, R, <dither>,R, <dither>, G, <dither>, G, <dither>,G, <dither>, B, <dither>, B, <dither>,B

 

If you cycle through the filters, then you can dramatically reduce the number of dithers.  For example, this sequence gets the same result:

 

L, R, G, B, <dither>, L, R, G, B, <dither>, L, R, G, B

 

I do notice that you are using an ASIAir, which doesn't support filter offsets.  Personally, that would be a total show stopper for me, unless I was using a one-shot-color camera.  There is no way that I would want to do a full focus run on every filter change.  That's totally nuts, especially with narrow band filters (I always focus on luminance and use offsets for the other filters).  In my opinion. that makes the ASIAir unsuitable for use with a mono camera.  YMMV.

[gerdastro]

I’m a pessimist, so a lot of time I shoot a single narrowband filter per night, Ha first. That way, if I only get one night of imaging (weather, mechanical, a bear eats me), I will have a great image. A second night I get my OIII, then I have a beautiful bi-color image. A third night, if I need SII, voila; if not, I can pick up more Ha and OIII, and my RGB stars (which I try and get whenever). So, great image of a different kind regardless of the nightly circumstances. Of course, if I ONLY have one night, I try and shoot them all and hope at a later date I can collect more data. And thus the pessimist wins! 

 

Paul

You seem to be doing something very similar to what I do. I am curious though: have you ever shot all filters in one night, then added evenly over following nights - and compare this to a result you would get if you do it the way you are doing it now?

[gerdastro]

Are you dithering?

 

If you are dithering and guiding, keep in mind that the time to re-acquire and center the guide star after a dither takes time.  I am a strong advocate of dithering on every exposure in a stack, unless I have multiple hundreds of subs per channel.  In that case, I may dither after two or three subs per channel.  I'd also note that I am not personally guiding, so my recover times after a dither are pretty short, but still there.

 

If you rotate through all filters, you might actually save time versus imaging repeatedly through the same filter.  Here is a short example of what I mean.  Assume that you are doing an LRGB image and taking 3 subs per channel.  If you image all the luminance first, then all the red, then all the green, and finally all the blue, then your sequence looks like this:

 

L, <dither>, L, <dither>,L, <dither>, R, <dither>, R, <dither>,R, <dither>, G, <dither>, G, <dither>,G, <dither>, B, <dither>, B, <dither>,B

 

If you cycle through the filters, then you can dramatically reduce the number of dithers.  For example, this sequence gets the same result:

 

L, R, G, B, <dither>, L, R, G, B, <dither>, L, R, G, B

 

I do notice that you are using an ASIAir, which doesn't support filter offsets.  Personally, that would be a total show stopper for me, unless I was using a one-shot-color camera.  There is no way that I would want to do a full focus run on every filter change.  That's totally nuts, especially with narrow band filters (I always focus on luminance and use offsets for the other filters).  In my opinion. that makes the ASIAir unsuitable for use with a mono camera.  YMMV.

For SHO my subs are 300s and I dither after each frame. For L and RGB I have only just started out and have not really found my sweet spot yet. Somewhere on the board it was suggested to just dither every 300s in general, so if L is 60s -> every 5 subs; if RGB is 120s -> every 2-3 subs. Other voices have suggested that for L (60s), every 7-9 subs should be enough, so I am still figuring out my best mileage.

 

Anyway, I am also guiding and you brought up a good point. Frankly, I was not aware of the pros of filter offsets until after I had my rig purchased, with the ASIAir as center piece. For now I have to deal with what I have. So your sequence example would not work for me (but it is great knowing about this, for a potential future setup).

 

Coming back to the original question, since LRGB was mentioned a few times here I am now also wondering about differences of balancing between SHO and LRGB shooting. My uneducated guess is that for SHO, combining the 3 signals taken from different nights under different conditions should be less of a problem, because SHO anyway is false-color and has a bit more of an 'artistic' approach. For LRGB, it might be more problematic as imbalances due to differently affected signal are - probably - easier detectable. Any thoughts?

Edited by gerdastro, 14 November 2024 - 09:44 AM.

[jonnybravo0311]

I do both. Some nights I'll shoot all filters. Some nights I'll shoot only a single one. For example, if I'm planning an HaRGB image, I'll shoot Ha only under the bright moon and the RGB on nights when the moon isn't up. For narrowband, I will typically shoot S2 and Ha under bright moon and O3 when it isn't so bright. If I just want to see something quickly, then I'll shoot all filters in a single night and process the data to get a preliminary image and see which filters might need more time than others based on the results.

 

If I'm shooting multiple filters, I arrange my sequences such that I shoot L when the target is highest in the sky, then B, then G, and finally to R when the target is closest to the horizon. For example, if I were going to target Andromeda (which at my location crosses the meridian almost directly overhead) and could shoot the entire night (I wish I could but I have near zero visibility east of the meridian), my sequence would look something like this:

 

Wait until target is 30 degrees above the horizon

R for an hour

G for an hour

B for an hour

L for 2 hours

B for an hour

G for an hour

R for an hour

 

Depending on how long my subs are, I either dither every sub (if longer than 4 minutes) or every 4 minutes' worth of subs. So, if I'm shooting 30 second L, I dither every 8. 60 second G, I dither every 4. 300 second Ha, I dither every 1. Why four minutes? No science behind it, it was an arbitrary value I chose.

[psienide]

I typically start a target with the intent of getting an equal amount of each filter. Then i'll run it through a quick process to see where things need to go. Obviously this all depends on weather, moon, etc but the general approach is get to a quick composition and see where i'm at with the data and go from there.

[WadeH237]

Somewhere on the board it was suggested to just dither every 300s in general, so if L is 60s -> every 5 subs; if RGB is 120s -> every 2-3 subs.

That makes no sense to me.

 

The whole idea behind dithering is that you create a large number of slightly offset images.  Over the course of the session, the scope moves the objects in the field slightly, but the fixed pattern noise in the camera stays fixed.  When you register the subs with the stars, that fixes the position of the objects in the image and makes the fixed pattern noise move slightly in each image.

 

The benefit to dithering it two-fold.  First, having a very large number of dithers reduces the weight of the fixed pattern noise when you average the subs.  And second, it greatly increases the efficiency of statistical rejection of the brighter outlier pixels that comprise the fixed pattern noise.

 

Both of these benefits are tied to the number of dither positions, not the length of the subs.  One of the suggestions you mention, dithering every 7 to 9 subs for 60 second luminance exposures would almost certainly have very little benefit (unless you had many hundreds, or even thousands, of subs).

 

The suggestion to dither after some number of subs, rather than on every sub, came about a few years ago when the first CMOS cameras started to gain market traction.  These cameras had shallow wells and often 12 bit ADC's.  The would easily over-expose unless the exposures were kept very short.  When you end up with 300 subs from a single session, it's simply not practical to dither after each one, hence the trend to dither after every 'x' exposures.

 

But make no mistake, the benefits of dithering really start to shine when you have lots of dithers.

[psandelle]

You seem to be doing something very similar to what I do. I am curious though: have you ever shot all filters in one night, then added evenly over following nights - and compare this to a result you would get if you do it the way you are doing it now?

I do both (as I said above; it depends on how many nights I have). I've never actually done a side-by-side comparison, but I'm at a Bortle 4 site, so gradients aren't a big thing for me, so matching different days' worth of data isn't something I've found to be a problem. I think it's a tossup as far as final output is concerned. I tend to have a target goal for overall data collection, and if I can reach that in the clear skies and personal time I have, either way will work for me, and I find no difference in the final outcome. Heck, I came back a year later to get OIII data on one project, and it was easy-peasy to process.

Paul

[fewayne]

WRT gradients in particular, it's easiest to remove them if the subs all have similar ones. So that argues for packing each filter together as much as possible.

 

But my intuition is that there's a limit to that, if you do six hours of one filter in a night, LP gradients are going to be very different for the first and the last. Curses! Foiled again!

 

So...a middle ground? If anyone's quantitated this, I would love to hear it.

[peterc]

For good or ill I tend to shoot one or two filters a night max for SHO - usually just a single filter. For LRGB, so far, I shoot them all pretty much in order.

I remove gradients from the results before channel combination.

Following this thread as it goes for the experiences...

Edited by peterc, 14 November 2024 - 06:53 PM.

[gerdastro]

That makes no sense to me.

 

The whole idea behind dithering is that you create a large number of slightly offset images.  Over the course of the session, the scope moves the objects in the field slightly, but the fixed pattern noise in the camera stays fixed.  When you register the subs with the stars, that fixes the position of the objects in the image and makes the fixed pattern noise move slightly in each image.

 

The benefit to dithering it two-fold.  First, having a very large number of dithers reduces the weight of the fixed pattern noise when you average the subs.  And second, it greatly increases the efficiency of statistical rejection of the brighter outlier pixels that comprise the fixed pattern noise.

 

Both of these benefits are tied to the number of dither positions, not the length of the subs.  One of the suggestions you mention, dithering every 7 to 9 subs for 60 second luminance exposures would almost certainly have very little benefit (unless you had many hundreds, or even thousands, of subs).

 

The suggestion to dither after some number of subs, rather than on every sub, came about a few years ago when the first CMOS cameras started to gain market traction.  These cameras had shallow wells and often 12 bit ADC's.  The would easily over-expose unless the exposures were kept very short.  When you end up with 300 subs from a single session, it's simply not practical to dither after each one, hence the trend to dither after every 'x' exposures.

 

But make no mistake, the benefits of dithering really start to shine when you have lots of dithers.

If I recall correctly from the thread I took this rule-of-thumb from, this was exactly the background: 'dither after x sub length' is an indirect way to define 'dither after x subs'.
It is the same concept of what jonnybravo0311 has mentioned in his reply: either dither every sub (if longer than 4 minutes) or every 4 minutes' worth of subs

 

But from your explanation I get that I should probably dither more often. My LRGB is simplified at L:60s and RGB:120s. For L, I would assume I'd get at least 500 subs for a project (maybe 100-200 per night). And for RGB, probably 120-150 each for a project (30-50 each per night).

Given these numbers, what dithering interval would you recommend?

[gerdastro]

For good or ill I tend to shoot one or two filters a night max for SHO - usually just a single filter. For LRGB, so far, I shoot them all pretty much in order.

I remove gradients from the results before channel combination.

Following this thread as it goes for the experiences...

Looking at the answers so far this is also my tendency moving forward. SHO more or less concentrating on one filter per night (maybe being a little more considerate of moon phases and target elevation), and for LRGB after all try a more even approach - although that means a slightly less efficient use of imaging time. This sounds like a decent compromise. Still hope somebody could chime in to share some actual comparisons or at least experience.

[jonnybravo0311]

If I recall correctly from the thread I took this rule-of-thumb from, this was exactly the background: 'dither after x sub length' is an indirect way to define 'dither after x subs'.
It is the same concept of what jonnybravo0311 has mentioned in his reply: either dither every sub (if longer than 4 minutes) or every 4 minutes' worth of subs

 

But from your explanation I get that I should probably dither more often. My LRGB is simplified at L:60s and RGB:120s. For L, I would assume I'd get at least 500 subs for a project (maybe 100-200 per night). And for RGB, probably 120-150 each for a project (30-50 each per night).

Given these numbers, what dithering interval would you recommend?

You'll also note that in my reply I stated there was no science behind the 4 minute value I chose . Truth is I don't always follow that rule, either. For example, if I have my GT81 mounted (I use a separate guide scope and filter offsets in NINA), I setup my sequence like "LLLRGB Dither", so I actually AM dithering every sub, it's just that I only have to dither once every 6 subs. I'm also not losing any time to the guider starting/stopping on filter change. I can't do the same thing when using my 8" Edge with an OAG. With that setup, I have to stop guiding, do the focus (even if it is just using an offset) and start guiding again. So, I try to minimize filter changes. Thus, my sequences look more like "Take 60 x 60s L, dither every 4 frames, then switch to R, take 30 x 120s and dither every 2 frames".

 

My goal is to maximize the time my camera and scope are gathering photons on my target, not waiting on some process. Should I dither every frame? Ideally, yes. Am I going to when I'm shooting 30 second luminance subs? No.

[WadeH237]

Should I dither every frame? Ideally, yes. Am I going to when I'm shooting 30 second luminance subs? No.

Well, sure.

 

8 hours of 30 second luminance subs would be 960 exposures.  That is a great scenario for dithering after multiple frames.
 

[Alex McConahay]

Well, sure.

 

8 hours of 30 second luminance subs would be 960 exposures.  That is a great scenario for dithering after multiple frames.
 

 

 

But one of the questions of this thread is "Can I get 960 thirty second exposures in eight hours of imaging?"

 

Your 960 number assumes that each exposure takes 30 seconds, and the next exposure starts immediately. 

 

It does not.

 

It takes a download, perhaps a dither and settling, a focus now and then, a filter change now and then (perhaps, depending on whether you do one filter per night or not), and such.  I would be surprised if you got much more than 600 per eight hours of telescope time. The average time between one thirty second exposure and the next (not counting focusing, re-pointing with plate solve, etc.) could be more like 45 or more seconds on a thirty second exposure all things considered.  One third of the time (33 per cent)  for each exposure is taken up by overhead. 

 

My (rather old) equipment seems to like longer exposures (300 seconds L 1x1, 300 seconds LRGB 2x2, 300 seconds Narrow 1x1). So the overhead time (download, dither, etc) is a much smaller percentage of telescope time (one twentieth, 5%) than it is for those whose equipment prefers 30 second exposures. 

 

This whole thread is an example of knowing the factors involved and finding a happy compromise of everything is more important than a single answer. As a discussion of the topics that contribute to that discussion it has been helpful. As a recommendation as to what somebody should use, it is less so.   

 

Do what you can to minimize the overhead. Factor in whether your second or third night will be clear. (Nothing like having a fine collection of LRG, and then having to wait a year before the object lines up again on a clear night to get the B!)  Make your decisions, and go for it. 

 

And, remember, if you are worrying about focusing time----you should refocus now and then anyway, just to check whether the filter offsets, the tube shrinkage, the temperature compensation, and all that are all taken into account----so, a filter change in itself may take no more time in one scheme than another because of filter change. 

 

That does not bother gerdastro (the original post). ("Personally, for me that is a non-issue. If the project is interrupted by clouds - even for months - while annoying, I can live with that and just continue at a later date or even the next season.") On the other hand, it would drive me crazier. 

 

As for gerdastro's "Ending up with very unbalanced signal acquisition between filters due to changing conditions" Well, conditions are always changing.... If clouds moved in and I never got my blues, or the moon rose and the gradient changed, I would face similar challenges in processing whether it was because the change was from one night to another or from early in the evening to later in the same evening. 

 

So, learn what you can from the discussion above, but apply it to your own needs. Just be happy with what you come up with (or if you are not, just change your formula until you are as happy as can be.)

 

I would prefer to be able to go home with a usable (if not ideally long enough overall) set of data----something I can get a nice picture out of----than a set missing one filter's worth.  If I get a second night on the same target, I will frame and rotate so that I can add to the first. If I get a third and a fourth, I will do the same. Each night will contribute to the overall signal to noise improvement. But, if I never get those subsequent nights, at least I will have an image I can use.

 

Alex

[WadeH237]

Your 960 number assumes that each exposure takes 30 seconds, and the next exposure starts immediately.

Perhaps I wasn't clear, but I was talking about 8 hours of exposure time, not 8 hours of session time.

 

Given all the possible sources of overhead, I suspect that everyone will get a different amount of exposure time for a given session length.  I was hoping to avoid complication and focus on just the number of subs.  I could have probably worded that a bit better.