Exposures to get unsaturated stars

[pete_81]

Still getting ideas together

So when imaging, DSOs need minutes to get enough light into the sensor but the dynamic range of cameras just isn't sufficient (stars very blown-out and stretching leading to noisy DSOs) - what's the solution? Do short exposures to get stars good (and you obviously don't need too many of these) and (more) longer exposures to bring out DSOs (and then even longer exposures for the dimmer regions, accompanied by complex masking and stacking)?

And obviously darks associated with each of the exposures? Then flats & bias images for the session too?

Then stack darks & lights of each exposure (with master flats & bias) and blend each exposure together bringing out the 'HDR' image, or can DSS (for example) handle the saturated areas of an image and use the shorter exposures for these regions automatically? I doubt the latter but would be nice, wouldn't it?!

[vlaiv]

Yes, that is correct.

You really need only two types of exposures - regular ones and short ones to fill in saturated bits in regular exposures. You don't need more than that.

Bias and flat files can be shared between the two. You only need separate darks - and only if you don't do dark scaling (for example CCDs with set point temp cooling can use same darks scaled down to match short exposure length).

Blending of the data is ideally done in linear stage, and here is the "logic" for it:

Take all pixels in regular stack that saturate and replace them with short_stack * time_ratio from short stack. You don't need to be exact about saturated pixels - you can use simpler condition like all pixels that have value larger than 95% of max value for example.

Time_ratio is just ratio of two exposure times - so if you have 2 minute regular exposure and 10 second short exposure, you would multiply short stack pixel values with 120s / 10s = 12.

I think that DSS can do something similar automatically for you if you use "Entropy Weighted Average (High Dynamic Range)" as stacking algorithm. Here is what technical info on DSS has to say about this method:

Quote

Entropy Weighted Average (High Dynamic Range)
This method is based on the work of German, Jenkin and Lesperance (see Entropy-Based image merging - 2005) and is used to stack the picture while keeping for each pixel the best dynamic.
It is particularly useful when stacking pictures taken with different exposure times and ISO speeds, and it creates an averaged picture with the best possible dynamic. To put it simply it avoids burning galaxies and nebula centers.
Note: this method is very CPU and memory intensive.

I have tried this method of stacking and it works - but I have not tried it with different exposure times, so not sure if it will work well for that use case.

[pete_81]

Thanks @vlaiv,

So quick search for that technique also gave old thread that agreed with what you say above

https://stargazerslounge.com/topic/60705-dss-entropy-weighted-average-hdr-stacking/

So shall give it a go and see.

Guess the question is to use this method just once with all loaded images, or to do light, dark, flat & Bias stack for each exposure, then do a stack using the result from each. More reading required but certainly looks promising! 

 

[wimvb]

From your signatue I see you use dslrs for imaging. A modern cooled cmos has more dynamic range than a dslr, and you don't need to use multiple exposure times. But you do need many subs to stack. In the end it's about total integration time, which depends on, among other things, local darkness (light pollution). I image from a relatively dark site (mag 20.5), and use a camera setting of low gain to maximise dynamic range, and exposure times that keep the stars from saturating, except the very brightest. My unstretched masters look completely black with just a few white dots where the bright stars are. But after stretching, I get colour in the stars and still enough detail to reveal faint galaxies over a billion light years away. To achieve this, I try to gather at least 10 hours of data on any one target, and use a work flow that I have optimised for those targets. And I avoid noise reduction as much as possible.

Here's one example.

https://www.astrobin.com/hh7zvu/

[pete_81]

Thanks very much for that Wim. Yes, using (modded) DSLR as the astro cams really just can't be justified until I get a decent workflow and more likely location. Home is in Bortle 6, so not really the best, but the local bakery has floodlights on all night which do tend to light the garden a bit.

Did start a thread about exposures to use as a starting point, along with ISO setting which sparked the usual comments - use the native ISO of camera for max DR (hence my using ISO200 as standard).

But in short, my idea of multiple durations isn't ideal as you're recommending more shorter exposures (but ultimately the integration time is there as there, just with more images to process with), whilst ideally avoiding saturated stars.

Anyone else wanting to comment on this - "just do more, shorter exposures to avoid saturation"?

[vlaiv]

2 hours ago, pete_81 said:

Anyone else wanting to comment on this - "just do more, shorter exposures to avoid saturation"?

Dynamic range of used sensor is really not important for long exposure astrophotography. It is irrelevant metric (it is useful for daytime photography where we take single exposure).

Stack two images - you increase DR by 1 stop / 1 bit, stack 4 in total - increase DR by 2 stops / 2 bits and so on. Increase over basic DR depends on how many subs you stack.

You should determine your base exposure length so that your read noise is swamped by some other noise source. Usually it is LP noise, but if you have DSLR - it can be dark current noise as well since camera is not cooled and dark current can be considerable.

Once you determine your base exposure length - shoot as many subs as you can in your "budgeted" imaging time. If you have saturation (and most likely you will) - shoot couple of short "filler" exposures at the end.

In the end - let's think about what dynamic range is and how it is calculated? It is max signal divided with read noise. But what if we had 0 read noise camera? DR would be infinite in that case?

To me, dynamic range should represent ratio of strongest to weakest recorded signal. It can never be infinite.

[wimvb]

4 hours ago, vlaiv said:

Stack two images - you increase DR by 1 stop / 1 bit, stack 4 in total - increase DR by 2 stops / 2 bits and so on. Increase over basic DR depends on how many subs you stack.

You can increase dynamic range by stacking, sure. But that doesn’t mean that DR isn’t important. I’d rather start at 14 stops DR and  work my way up, than at 8 stops and have to handle 64 times as many subs.

[vlaiv]

2 minutes ago, wimvb said:

I’d rather start at 14 stops DR and  work my way up, than at 8 stops and have to handle 64 times as many subs.

Why would you have x64 as many subs?

You only need that if you set exposure length much shorter in case of 8 stops camera - but you should not base your exposure length on that - but rather on ratio of read noise to other dominant noise source.

Maybe both cameras have same level of read noise, only one with 14 DR stops has much larger full well capacity? In that case - you would take the same number of subs for each - and recover saturated parts in 8 stops camera with few filler subs.

[pete_81]

So ignoring the topic of DR etc, I've read a couple of things about darks with DSLRs and that the temperature fluctuates (obviously!) more than a cooled astro-cam.
I also see posts (example) saying that with DSLRs to ignore darks and concentrate on more lights and using dithering to get away from hot pixels and hoping that the random noise is 'neutralised' by the motion of the image around on the sensor.

Obviously skipping darks would allow a significant increase in light frames and I do agree with the DSLR camera temperature changing throughout an imaging session - would it be be a recommendation to apply more dithering (12-15pixel dithering every couple of frames) and ignore darks but not the bias/flats obviously?

[wimvb]

5 hours ago, pete_81 said:

apply more dithering (12-15pixel dithering every couple of frames)

That is the recommended amount of dithering. I believe that it was first suggested by Tony Hallas in a presentation he gave years ago. The presentation is still available on youtube: Tony Hallas astrophotography.

[ollypenrice]

I don't know of anybody who does a set of short exposures for stellar cores. I've certainly never done so, but it doesn't stop us from getting strong star colour. Indeed, it is very exceptional for me to do short exposures for any part of an image. The obvious exception is M42 where it's necessary for the Trapezium region. Like many others I used sets of about 10 seconds, 50 seconds and then 15 minutes with a CCD, layer masked together in Photoshop.

Exceptionally I may use my RGB as a set of shorter exposures for galaxy cores since the RGB has less signal and this doesn't apply in your case, but my point is that blending long and short exposures is an imaging rarity.

It's possible to pull star colour from the fainter outer edge into the core. The easy way is to use Noel's Actions, Increase Star Colour, (now known as Pro Digital Astronomy Tools, I think.) Alternatively you can make a copy layer, blur the bottom layer, make a star selection (instructions available in MartinB's tutorial on here in the Imaging Techniques section) and erase a feathered selection of the stellar cores from the top layer.

Increasingly popular is the use of Starnet++ or Star Xterminator to de-star a stretched image before placing the linear image on top in Blend Mode Lighten. You can then give it a gentle stretch and the stars will appear with less over exposure and more colour (which can be colour-saturated without affecting the rest of the image.)

Olly

 

[pete_81]

Thanks for this detail, and what an image!

So query again about DSLR and darks? Are darks more a waste of space (and time that more lights can be taken) with long astro exposures? 

[The Lazy Astronomer]

2 hours ago, pete_81 said:

Thanks for this detail, and what an image!

So query again about DSLR and darks? Are darks more a waste of space (and time that more lights can be taken) with long astro exposures? 

Darks are really only of use if you have control over the sensor temperature (eg cooled astro cam), else you end up potentially introducing more problems than you solve. 

For DSLRs, the usual advice is to take bias and dither between frames. 

[pete_81]

Sorted The answer I was expecting TBH, more time for lights though so that's ok!

[iantaylor2uk]

I see from your signature that your scopes are quite fast at f4.8 and f4.9. 

Personally I would try 30 sec or 1 min exposures and just take lots of subs. You don't need to take multi-minute expsosures at those focal ratios, unless you are doing narrowband or using an L-enhance or L-extreme filter. What would also help is if you can somehow cool the sensor on your DSLR. I have the book "The Art of Astrophotography" by Ian Morison, and in that book he has a DIY icepack coolbox which may be useful.  

[pete_81]

Hi Ian,
Thanks for that also - that's another topic I tried to start to see about starting points for DSLR exposures. As I understand it (from a daytime photo perspective), the brighter a subject, the faster the exposure can be taken so when you're saying about 30-60 sec exposures, what sort of magnitude object are you talking about?

A starting point for me is with the North America Nebula (NGC7000 in Cygnus) which is at mag +4, native ISO on camera (800), f/4.9 and 5min x30 light subs. My max on a stack in grey level is 210/255 so not saturating which may suggest I can go even longer.

However, longer exposures with uncooled sensor doesn't seem the wisest as even more noise would be present. Although coming around in circles again - that surely is the idea of larger dithering numbers and another discussion in this post. My typical setup was 5-pixels each dither, every 5 frames - quick calculation (aka astro_tools!) shows my guiding resolution at 3.22"/px and imaging at 1.77"/px so near enough factor of 2 - so dithering should be moving the image around by say 10px (assuming the dithering moves guiding star(s) around by the number that's input to EKOS (i.e. it doesn't compensate for imaging resolution unless using OAG where they're nearly the same ratio anyway)? The linked post above, and Wim's comment suggest that this is lower than I should be using, so hope to try later again.

Waffling, sorry. So yes, just wondering about exposure, f/ratio and magnitude of target when you're saying use 30-60sec, or have I again missed the point?!

Oh and thanks for cooing link - I've been looking into peltier cooling of the DSLR at some point as it seems quite popular - hoping the colder & longer nights might help too!

Edited October 22, 2021 by pete_81

[iantaylor2uk]

30-60 secs should be ok for most DSOs . You will still need to image for at least 1 or 2 hours (the more the better). There should be an optimum ISO for your camera that others could advise on, but I imagine you want the camera on ISO 800 or ISO 1600. You may not see much in a single sub but when you stack all of them and stretch/process the images, the DSO should come out. Below is an image I took recently with a one shot camera ZWO 071 of the North American nebula  which a stack of 40 one minute subs with a WO 81 GT IV refractor (f5.9) with just a UV/IR filter. The focus was slightly off in the image.

 

[The Lazy Astronomer]

2 hours ago, pete_81 said:

Hi Ian,
Thanks for that also - that's another topic I tried to start to see about starting points for DSLR exposures. As I understand it (from a daytime photo perspective), the brighter a subject, the faster the exposure can be taken so when you're saying about 30-60 sec exposures, what sort of magnitude object are you talking about?

A starting point for me is with the North America Nebula (NGC7000 in Cygnus) which is at mag +4, native ISO on camera (800), f/4.9 and 5min x30 light subs. My max on a stack in grey level is 210/255 so not saturating which may suggest I can go even longer.

However, longer exposures with uncooled sensor doesn't seem the wisest as even more noise would be present. Although coming around in circles again - that surely is the idea of larger dithering numbers and another discussion in this post. My typical setup was 5-pixels each dither, every 5 frames - quick calculation (aka astro_tools!) shows my guiding resolution at 3.22"/px and imaging at 1.77"/px so near enough factor of 2 - so dithering should be moving the image around by say 10px (assuming the dithering moves guiding star(s) around by the number that's input to EKOS (i.e. it doesn't compensate for imaging resolution unless using OAG where they're nearly the same ratio anyway)? The linked post above, and Wim's comment suggest that this is lower than I should be using, so hope to try later again.

Waffling, sorry. So yes, just wondering about exposure, f/ratio and magnitude of target when you're saying use 30-60sec, or have I again missed the point?!

Oh and thanks for cooing link - I've been looking into peltier cooling of the DSLR at some point as it seems quite popular - hoping the colder & longer nights might help too!

You can practically ignore everything you know about daytime photography - that's how different astrophotography is!!

As @iantaylor2uk says, you won't go far wrong with 30 - 60 seconds subs. 

[pete_81]

Great starting point, thanks @iantaylor2uk and @The Lazy Astronomer
Due to be clear later so hope to get out and try this shorter exposure out

[The Lazy Astronomer]

1 hour ago, pete_81 said:

Great starting point, thanks @iantaylor2uk and @The Lazy Astronomer
Due to be clear later so hope to get out and try this shorter exposure out

Be aware that individual subs will likely look terrible, like there's nothing there, but it will all come out once stacked (and processed).

[pete_81]

It's the stretching and post processing that I'll be bugging you about next !

[pete_81]

So yep, last night was decent in between clouds. I just set the imaging going for 100frames at 1min exposure. Got to 93 before cloud cover was too much. Here's a quick startools edit of NA Neb, no lights were ignored, so some will have a partial cloud cover, etc etc, but the aim of the imaging last night was to do more, shorter (than my usual) subs and see how things look after stacking & stretching. Not done the edit in APP as yet, but thought I'd post to say thanks again to all above for input - clearly the above answers about 1min exposure are indeed the way to go so that's a highly recommended starting point to anyone new or just making hard work of something that should be relatively simple!

Lights - 93x 1min
ISO800
f/6 (3D printed aperture for the ST102) at F=500mm
Bias (loads x 1/4000s) and Flats added, processed quickly in the trial version of StarTools

 

[wimvb]

2 hours ago, pete_81 said:

Got to 93 before cloud cover was too much. Here's a quick startools edit of NA Neb, no lights were ignored,

Just for the fun of it, compare this result with one where you are more restrictive in selecting subs. Clouds will brighten the overall image and reduce contrast. The added light due to clouds is easy to correct, but the associated noise can spoil the image. You might get a better result with fewer subs that lack clouds than with more subs that include clouds.

[pete_81]

Good idea Wim,
Very quickly, can I ask again for advice here - here's a screenshot of the subs used

I'm going to assume for selecting the ones without cloud to use the darker ones - eg the range 014-042,then 056-080? Opening 025 in Affinity and 090 have very different histograms, unsurprisingly.

Out of interest on that also, is the histogram on the Left (image 025 in the above folder screenshot) OK position wise or should it be more to the left or right (is exposure OK or not) for future reference of where it should be placed?

(Taking the very bright images (looks like 30) still leaves 64 similar ones in the folder to process)
 

Edited October 26, 2021 by pete_81

[iantaylor2uk]

If you use DSS for stacking, it rates the quality of each sub and you can tell DSS to stack the best 80% (or 70% or 90%, whatever you specify).