M51 unguided, 200pds

[Oslet]

Hi,

Just 20min of data. Still learning the ropes. 

200pds, canon 1300D, DSS

 

[Craney]

That's a good result for unguided.  How long per light sub ??   

I could only manage about 30sec with my 200pds before the galaxy started to form a 'jet' !!!

[Knight of Clear Skies]

Definitely a good result for an unguided image.

[Oslet]

1 minute ago, Craney said:

That's a good result for unguided.  How long per light sub ??   

I could only manage about 30sec with my 200pds before the galaxy started to form a 'jet' !!!

Hi and thanks. The subs are 60sec. I can usually get 60 sec with HEQ5 and good polar alingnment. Just need to get a coma corrector

[Altais]

Great work! Nice and round stars, a lot better than anything I've managed to do unguided.

Let's just hope the great weather doesn't go away anytime soon.

[cesco]

I am going unguided since last summer, annoyed by the bad performance of an AZ EQ6 with a too heavy Meade 7" ED achromatic refractor. I resorted to unguided imaging also after realizing that my once upon a time dark site has turned to a suburban to rural transition site. I find that statistical cancellation of background light by stacking many short exposures, which increases the SNR by the square root of the image number, is better than subtract background light accumulated in long exposures. To me, it makes no sense to accumulate image photons together background photons, as successive photon subtraction aimed at darkening the background will take away also image photons, leaving the ratio of image to background photons unchanged. On the contrary, stacking many short exposures will increase this ratio as SNR increases. I am getting great fun experimenting this method, and skipping annoying guiding procedures, apart from an accurate polar alignement that remains strongly needed.

M82: Meade 7" ED achromatic refractor on Gemini11 Losmandy mount; Atik 428EX monochromatic camera operated by Artemis software (Atik); RGB Baader dielectric filters mounted into a Baader filter slit, plus Neodymium filter mounted on the camera nose. 45s exposures: 54R, 45G, 35B. Images of each channel were stacked by the Dawn (from Atik) software and the three channels were colorized, combined and processed to polichromatic image by Registax 6.2

[Oslet]

9 hours ago, cesco said:

I am going unguided since last summer, annoyed by the bad performance of an AZ EQ6 with a too heavy Meade 7" ED achromatic refractor. I resorted to unguided imaging also after realizing that my once upon a time dark site has turned to a suburban to rural transition site. I find that statistical cancellation of background light by stacking many short exposures, which increases the SNR by the square root of the image number, is better than subtract background light accumulated in long exposures. To me, it makes no sense to accumulate image photons together background photons, as successive photon subtraction aimed at darkening the background will take away also image photons, leaving the ratio of image to background photons unchanged. On the contrary, stacking many short exposures will increase this ratio as SNR increases. I am getting great fun experimenting this method, and skipping annoying guiding procedures, apart from an accurate polar alignement that remains strongly needed.

M82: Meade 7" ED achromatic refractor on Gemini11 Losmandy mount; Atik 428EX monochromatic camera operated by Artemis software (Atik); RGB Baader dielectric filters mounted into a Baader filter slit, plus Neodymium filter mounted on the camera nose. 45s exposures: 54R, 45G, 35B. Images of each channel were stacked by the Dawn (from Atik) software and the three channels were colorized, combined and processed to polichromatic image by Registax 6.2

Nice one of M82. I can see that my approach might be right then. The conditions are wery bad now, but I think I'll try one more night this season. Target M101, 30 sec exposures.  I have not tried any filters yet. Maybe I'll have a shot at RGB. Just to see how much data I can pull from a bright sky.

[Altais]

10 hours ago, cesco said:

I am going unguided since last summer, annoyed by the bad performance of an AZ EQ6 with a too heavy Meade 7" ED achromatic refractor. I resorted to unguided imaging also after realizing that my once upon a time dark site has turned to a suburban to rural transition site. I find that statistical cancellation of background light by stacking many short exposures, which increases the SNR by the square root of the image number, is better than subtract background light accumulated in long exposures. To me, it makes no sense to accumulate image photons together background photons, as successive photon subtraction aimed at darkening the background will take away also image photons, leaving the ratio of image to background photons unchanged. On the contrary, stacking many short exposures will increase this ratio as SNR increases. I am getting great fun experimenting this method, and skipping annoying guiding procedures, apart from an accurate polar alignement that remains strongly needed.

M82: Meade 7" ED achromatic refractor on Gemini11 Losmandy mount; Atik 428EX monochromatic camera operated by Artemis software (Atik); RGB Baader dielectric filters mounted into a Baader filter slit, plus Neodymium filter mounted on the camera nose. 45s exposures: 54R, 45G, 35B. Images of each channel were stacked by the Dawn (from Atik) software and the three channels were colorized, combined and processed to polichromatic image by Registax 6.2

I think there is a lot of sense in what you're doing.

If our cameras had no read noise, there would in principle be no difference between the SNR of a single 10 minute exposure and stack of 20 half minute exposures. Then it would basically always be better to do many shorter exposures, since you would never over-expose parts of your target, you wouldn't need to do guiding, you can get rid of artefacts like cosmic ray hits and plane trails by stacking and the cost of a failed exposure is very low. (Of course, if taken too far the sheer number of subframes to pre-process would become prohibitive eventually.)

Unfortunately the unavoidable presence of read noise in every subframe means that the second approach will always yield more noise. But this effect is basically negligible when the read noise is much smaller than the noise in the photons coming from the direction of your target. The photon noise increases as the square root of the number of photons, so the exposure time required to make read noise negligible is proportional to the square of the read noise divided by the combined intensity of the sky and the target. (I'm ignoring dark noise here since it doesn't really affect the argument.)

In other words, low read noise and/or a bright sky background allows for shorter exposures without any penalty on the SNR. Astronomical CMOS sensors with very low read noise (e.g. the ZWO ASI1600) are starting to gain popularity, and I suspect we will see more and more of this kind of short exposure imaging in the future.

I didn't mean to turn this into an essay, sorry for that. But it's an interesting discussion though.

[Oslet]

1 hour ago, Altais said:

I think there is a lot of sense in what you're doing.

If our cameras had no read noise, there would in principle be no difference between the SNR of a single 10 minute exposure and stack of 20 half minute exposures. Then it would basically always be better to do many shorter exposures, since you would never over-expose parts of your target, you wouldn't need to do guiding, you can get rid of artefacts like cosmic ray hits and plane trails by stacking and the cost of a failed exposure is very low. (Of course, if taken too far the sheer number of subframes to pre-process would become prohibitive eventually.)

Unfortunately the unavoidable presence of read noise in every subframe means that the second approach will always yield more noise. But this effect is basically negligible when the read noise is much smaller than the noise in the photons coming from the direction of your target. The photon noise increases as the square root of the number of photons, so the exposure time required to make read noise negligible is proportional to the square of the read noise divided by the combined intensity of the sky and the target. (I'm ignoring dark noise here since it doesn't really affect the argument.)

In other words, low read noise and/or a bright sky background allows for shorter exposures without any penalty on the SNR. Astronomical CMOS sensors with very low read noise (e.g. the ZWO ASI1600) are starting to gain popularity, and I suspect we will see more and more of this kind of short exposure imaging in the future.

I didn't mean to turn this into an essay, sorry for that. But it's an interesting discussion though.

SNR and photon essays can be quite interesting ?

[cesco]

I red a treatise on the subject by Alessio Beltrame: https://www.alessiobeltrame.com/segnale-rumore-e-calibrazione/ (in Italian) and found a strict analogy with the accumulation method I used to improve the SNR of weak EPR spectra I worked with in my professional job. I was naturally attracted to the multi-short-shot method I soon understood to be a powerful weapon against light pollution. In addition, similarly to what we do in planetary imaging, though in a higher time scale, it turns out to be also useful in neutralizing setup defects and winds blows, as well as seeing. In fact, just by lowering the exposure below the time scale of these phenomena you can divide the diffuse image of a trembling star into a few firm star images a few pixels apart. Subsequent alignment and stacking yields a round star on a darker background, with the same integrated intensity.

Having entered in astrophotography only recently, I miss basic notions such as ISO sensibility, gain and their relationship and  related effects on the final captured image. So, when it comes to astrophotography technical discussions I get lost…You cited read out noise (RON). My camera’s RON is 5e- and its sensor is Sony ICX674 (Atik 428EX mono).  Gain and sensibility are fixed to don’t know what value, and exposure time only is adjustable, and temperature setting, of course.

My site sky is quite polluted by now…I don’t need any longer a flashlight by night, even the darkest (?) ones…, and this renders RON a negligible contribution to noise in comparison to the background noise, so I should be tranquil going safely with this method. In different  conditions I know I should manage gain and sensitivity better, and the dynamical range, and so on. But I know nothing about these, and, at the end of all, I can’t adjust none in my camera. The method surely has a downstream limit, in terms of darkness, I’ll never reach, also because my setup is unmovable, while I’ll take advantage from good seeing.

In case you are familiar with my camera, I would appreciate some useful advice according to my camera features.

Thank you.

Regards,

Francesco

[Altais]

Craig Stark has a really nice series of articles which I would strongly recommend for anyone interested in understanding signal and noise and how they depend on equipment and conditions. The articles can be found here: https://www.cloudynights.com/articles/cat/column/fishing-for-photons/

One thing he covers is how you can measure the gain of your sensor. I was inspired by this to write a program to do it automatically, which evolved to do a lot of other things as well. If you're interested the software can be found here: http://lars-frogner.github.io/Astrophotography-Lab/

It can be used to experiment with various imaging parameters and see how they will affect the SNR and dynamic range of the final result. As an example, here is a plot showing how the SNR of a spiral arm of M51 in a 6-hour subframe stack will vary as a function of subframe exposure time with an Atik 460EX camera (same read noise as your 428EX). The amount of skyglow roughly corresponds what you would have during a full moon.

[cesco]

Thank you! Very nice stuff...I'm happy to get material to study and experiment. Does your tool also gauge skyglow? This would be very useful for comparing results.

Another topic I would like to discuss is the effect of the passband of filters. I use RGB dielectric filters from Baader Planetarium, with 100-nm passband, plus a Neodymium skyglow filter also from Baader. I guess they will affect the background light differently...any problems?

[Oslet]

Sooooooooo....maybe ISO 400 maybe? I've always used 800. But after reading Altais link I'll try 400. Hoefully the data will stick to th RAW files.  I'ts getting very technical very fast. I've got a lot of reading to do. Thanks for all the good info.

[Oslet]

Ok: take a look at screenshot of the RAW files. You can clearly see the noise here.  Different exposures at ISO 800 30sec/60sec, with a fairly bright sky. This noise is also quite evident in the darker subs I've taken. My goal now is to get as much data as possible and even out the noise. My take on this is short exposure at ISO400 - maybe 30  - 50 sec. Whats the experts take on this?   Anyway I'll try tonight, and the sky will be bright. So I I'll post my results as soon as I can

[Altais]

2 hours ago, cesco said:

Thank you! Very nice stuff...I'm happy to get material to study and experiment. Does your tool also gauge skyglow? This would be very useful for comparing results.

Another topic I would like to discuss is the effect of the passband of filters. I use RGB dielectric filters from Baader Planetarium, with 100-nm passband, plus a Neodymium skyglow filter also from Baader. I guess they will affect the background light differently...any problems?

Yes, you can obtain an estimate of the brightness in mag/arcsec^2 from a light frame. But there are a lot of uncertainties in this calculation, so you should take it as a rough guideline rather than a precise measurement.

The use of filters doesn't really affect the reasoning around signal to noise. But you should keep in mind that the narrower the filter, the longer you need to expose in order for the photon noise to dominate the read noise,  since photons are hitting your sensor at a lower rate. This is why narrowband imaging tends to require much longer exposures than RGB imaging.

[cesco]

I see... by reasoning in terms of photons, that's straightforward. So it's better not to try a Hubble palette, right?

[Altais]

23 minutes ago, cesco said:

I see... by reasoning in terms of photons, that's straightforward. So it's better not to try a Hubble palette, right?

Well, even though short exposures through narrowband filters will probably be read noise dominated, that is just a result of reduced photon noise from the (broadbanded) background. The total noise level will actually be lower, since you have the same read noise but lower photon noise. So for targets with significant narrowband emission, you should get better SNR using narrowband filters than RGB filters. But the fact that the subframes are dominated by read noise means that you in principle could improve SNR further for a given total integration time by increasing the duration of individual exposures.

[Altais]

Oslet, I would actually suggest trying a higher ISO (e.g. 1600) if you aim to use short exposures.

The ISO value is basically a tradeoff between read noise and dynamic range. With higher ISO comes lower read noise, which is always nice, but this comes at the cost of lower dynamic range. Actually, ISO is just a way of describing the concept of gain, which is the number of electrons (think photons) that must be accumulated in a pixel in order for its digital value to increase by one. High ISO gives low gain, meaning that the pixel value will increase more per electron and hence you get a "high sensitivity". In this case, fewer electrons are required for reaching the maximum pixel value (this value is typically around 2^14 or 2^16, depending on the camera), and hence the pixels are more quickly saturated. Faster saturation gives a smaller difference between the minimum and maximum number of photons that can be registered, corresponding to a lower dynamic range.

A sufficiently high dynamic range is required to avoid saturation of important parts of your target (like the core of a galaxy) during a single exposure. So if saturation is a problem, you should lower your ISO. But when you do short exposures, you are not likely to reach saturation even with a relatively high ISO. So then it might be better to use higher ISO so that you can benefit from lower read noise.

Anyway, this is my (non-expert) take on it. The best thing is probably to experiment to see what works best for your equipment and conditions in practice.

[Oslet]

Ok. I guess I'll have to wait until fall for more experimenting. Anyway this is the result with ISO 400. I would say that the grainy noise is evened out. This is the worst possible condition in nautical twilight, and you can see the moonglow from the top of the picture. I have to try high ISO to see how that works out in the final image in the fall. The curse of living above LAT 60

[cesco]

I'm afraid I can't take advantage of higher ISO for short exposures since my camera, Atik 428EX has no ISO, as well as gain, regulation. I don't even know what its fixed ISO/gain values are to get at least an idea wether they are in the high/medium/low range...I searched in the Atik manual and site to find out those values, without success...

Here is one of my first short shot experiments, IC432, 45" shots, 100  R, 100 G, 120 B, under strong moonlight. Meade 7" ED achromatic refractor.

Looks terrible!

IC342 45 dwnhis1 rgb RGXRGB4 MOPM1 rj.bmp

Edited April 20, 2019 by cesco
adding a picture

[Altais]

1 hour ago, cesco said:

I'm afraid I can't take advantage of higher ISO for short exposures since my camera, Atik 428EX has no ISO, as well as gain, regulation. I don't even know what its fixed ISO/gain values are to get at least an idea wether they are in the high/medium/low range...I searched in the Atik manual and site to find out those values, without success...

Here is one of my first short shot experiments, IC432, 45" shots, 100  R, 100 G, 120 B, under strong moonlight. Meade 7" ED achromatic refractor.

Looks terrible!

IC342 45 dwnhis1 rgb RGXRGB4 MOPM1 rj.bmp 514.78 kB · 1 download

I don't think CCD manufacturers typically state the gain of their sensors. But you might find numbers from others who have measured it. Or you can measure it yourself using the Astrophotography Lab program.