Jump to content

Stargazers Lounge Uses Cookies

Like most websites, SGL uses cookies in order to deliver a secure, personalised service, to provide social media functions and to analyse our traffic. Continued use of SGL indicates your acceptance of our cookie policy.

Sign in to follow this  
Edward Dewolf

Bortle 5: gain and exposure length ?

Recommended Posts

Hi,

I started imaging with a ASI294MC-PRO from a Bortle class 5 zone.

Can anyone help me how to find out what the best combination of gain and exposure length would be to image galaxies and nebulaes?

How does one figure that out?

Thx for helping

Edward

Share this post


Link to post
Share on other sites

If you haven't watched this already it should answer all your questions. 

 

  • Like 1

Share this post


Link to post
Share on other sites

My method is to expose until the sky background noise swamps the camera read noise, such that the sky background 16 bit ADU value is 10 x the square of the read noise.

Unity gain (1 electron incoming = 1 ADU) is always a good starting gain which is gain 117 for the ASI294. However this camera uses ZWOs HSC (High Conversion Gain) mode at gain 120 and higher which claims to give the same dynamic range as using gain 0 for a much lower read noise and is only slightly above unity gain.

Untitled-1.png.eb0a883d15b49287790c7ce7255424f3.png

So at gain 120, default offset which I believe is 30, you need to expose until the median 16 bit ADU value of the sky background is around 280 ADU. There is little point in exposing longer as you just end up clipping more stars for little gain in signal to noise. Start another sub when 280 ADU sky background is reached as the noise due to read noise at that point is negligible and won't significantly contribute to the overall noise.

So start off with an exposure of say 2 mins on a plain star field (no large nebula areas), and see what the sky background ADU value is, either by hovering the mouse cursor over the preview image background which should give you an ADU value, or examine the image statistics and note the median value. It should be fairly close to the mean value anyway. If it's more than 280 try again with a lower exposure and if it's less try a longer exposure. As it's a OSC camera the mouse values will depend on what colour bayer pixel your over (assuming the preview is not debayered) so the image statistics median value is probably a better bet for OSC.

For the same sky bortle darkness use this exposure value for all your ASI224 RGB images. It doesn't have to be exactly 280 ADU, between 270 and 290 will be fine. 🙂

That's just my suggestion, others may disagree. 😄

Alan

Edited by symmetal
  • Like 1

Share this post


Link to post
Share on other sites
15 hours ago, symmetal said:

My method is to expose until the sky background noise swamps the camera read noise, such that the sky background 16 bit ADU value is 10 x the square of the read noise.

Unity gain (1 electron incoming = 1 ADU) is always a good starting gain which is gain 117 for the ASI294. However this camera uses ZWOs HSC (High Conversion Gain) mode at gain 120 and higher which claims to give the same dynamic range as using gain 0 for a much lower read noise and is only slightly above unity gain.

Untitled-1.png.eb0a883d15b49287790c7ce7255424f3.png

So at gain 120, default offset which I believe is 30, you need to expose until the median 16 bit ADU value of the sky background is around 280 ADU. There is little point in exposing longer as you just end up clipping more stars for little gain in signal to noise. Start another sub when 280 ADU sky background is reached as the noise due to read noise at that point is negligible and won't significantly contribute to the overall noise.

So start off with an exposure of say 2 mins on a plain star field (no large nebula areas), and see what the sky background ADU value is, either by hovering the mouse cursor over the preview image background which should give you an ADU value, or examine the image statistics and note the median value. It should be fairly close to the mean value anyway. If it's more than 280 try again with a lower exposure and if it's less try a longer exposure. As it's a OSC camera the mouse values will depend on what colour bayer pixel your over (assuming the preview is not debayered) so the image statistics median value is probably a better bet for OSC.

For the same sky bortle darkness use this exposure value for all your ASI224 RGB images. It doesn't have to be exactly 280 ADU, between 270 and 290 will be fine. 🙂

That's just my suggestion, others may disagree. 😄

Alan

Hi Alan, I'm interested in this concept and might have an experiment and give it a try, just wondered though, where did you get the ADU value of 280 from?

Share this post


Link to post
Share on other sites
1 hour ago, Adam1234 said:

Hi Alan, I'm interested in this concept and might have an experiment and give it a try, just wondered though, where did you get the ADU value of 280 from?

Hi Adam,

There were a mass of topics on this on the CN forums after the ASI1600 came out. It applies to all cameras, but especially CMOS cameras with variable gain and offset.

This one I think got me started. The replies by Jon Rista, (who seems to be their version of vlaiv 😄) from post 3 gives the theory.

Also this post and this one add further discussion and info if you have a couple of spare evenings. 😀

A read noise swamping background ADU is calculated where the background ADU is 10*RN^2. Others use 5*RN^2 or just 20*RN but the first one is most often used.

You need to ensure that consistant units are used in the calculations. The gain values used in the calculations are those read from the GAIN in electrons/ADU from the graphs shown above.

If you chose the unity gain value of 117 (in 0.1db units), the value you enter on the camera driver, always a good starting point, the gain in e-/ADU gain is 1.0   The read noise at this gain value from the graphs is about 5.7 e-

The ASI294 HSC mode starts at gain 120 which should theoretically give better noise results so I chose gain 120 which is about 0.9 e-/ADU. Slightly higher than unity gain. The read noise at this value is only about 1.8 e-

10*RN^2 = 10 * 1.8 * 1.8 = 32.4 e-.  To convert to ADU divide it by the gain in e-/ADU. So 32.4/0.9 = 36 ADU

The offset value (default 30 for the ASI294 I believe) effectively just adds a fixed offset ADU value to every pixel so you need to add this to the above figure so it is comparable to the sky background ADU values.  Therefore add 30 ADU to the 10*RN^2 ADU value, 36 + 30 = 66 ADU.

As this is a 14 bit camera the ADU values quoted above are 14 bit values. To display as 16 bit values, as used on most capture programs, you need to convert it to 16 bit by adding 2 zero value LSB (least significant bits). This effectively multiplies the 14 bit ADU values by 4 (ie. 2^2)

So a 14 bit sky background value of 66 ADU becomes a 16 bit value of 66 * 4 = 264 ADU

Whoops! it's not 280. I must have read the read noise and gain graphs a bit differently when I came up with 280. I read them rather quickly when I posted yesterday. 😉

If you want to get the sky background ADU for different gain and or offset values, just read the relevant values from the graphs and do the maths. 🙂

For the offset I assumed that an increase of 1 offset increase the ADU by 1. This is true for the ASI1600 (12 bit) and ASI071 (14 bit) cameras I have. However my ASI6200 (16 bit) increases the ADU by 10 for every 1 increase in offset. This threw me when I first calculated the sky background values assuming ADU/offset was 1 to 1. You can easily check what it is by reading the peak value of the noise in a bias frame, and increasing the offset by say 10, and taking another bias frame. If the noise peak increases by 10 (at native bit depth) it's 1 to 1. Remember, If displaying at 16 bit then it will show an increase of 40 (for a 14 bit camera) or 160 for a 12 bit camera.

Anyway, that's the theory Adam. I hope it helped and hasn't addled your brain. 😁

Alan

Edited by symmetal
  • Like 1

Share this post


Link to post
Share on other sites

Thank you Alan, a lot of info there but very helpful! I will try calculating what ADU I should aim for. 

Another question then, how do I measure the sky background ADU? FYI I use APT for image for image capture

Share this post


Link to post
Share on other sites
2 hours ago, Adam1234 said:

Thank you Alan, a lot of info there but very helpful! I will try calculating what ADU I should aim for. 

Another question then, how do I measure the sky background ADU? FYI I use APT for image for image capture

Look at the image statistics, (I believe most image capture programs have this option) and the median ADU value is the sky background. Also, I don't know if APT has it but normally moving the mouse over the preview image gives a readout of the ADU level under the cursor. Move the cursor around a piece of sky background and get an average value of it. It's best to use an area of sky without a lot of nebula for testing, as they can possibly affect the median value, which doesn't then reflect the true sky background.

Of course the exposure to reach the sky background depends on the filter used. L will be shortest and R, G and B around 3 times longer, assuming you use the same gain setting. For R, G and B it's best to end up chosing the same exposure for each, even though that may not be optimum for some, just to avoid multiple dark exposures to match. If you do use a different gain, you need to recalculate the sky background ADU for that gain setting.

For OSC the mean value may be better to use rather than the median depending on whether it is debayered on preview. They should be fairly close anyway if you chose an area away from the milky way to do the test. 🙂

The swamping sky background value is only really useful for LRGB or OSC imaging. For NB you will unlikely be able to expose long enough to reach the ADU limit anyway unless you expose for an hour or more or you're at a very light polluted site. If you use much higher gain for NB you may well reach it in a shorter time but remember you have to recalculate the limiting ADU value if the gain is changed.

Alan 

Edited by symmetal
  • Like 2

Share this post


Link to post
Share on other sites

Thanks very much Alan. No clear sky tonight, so will see if I can find these image statistics on APT for the 5min Ha, Oiii and Sii subs I've captured this week on the Elephant's Trunk.

I'm in a Bortle 8 sky, so quite a bit of LP so these calculations will probably be very useful.

@Edward Dewolfsorry to have hijacked your thread, but probably been useful for the both of us!

 

Adam

  • Like 2

Share this post


Link to post
Share on other sites
3 hours ago, Adam1234 said:

Thanks very much Alan. No clear sky tonight, so will see if I can find these image statistics on APT for the 5min Ha, Oiii and Sii subs I've captured this week on the Elephant's Trunk.

I'm in a Bortle 8 sky, so quite a bit of LP so these calculations will probably be very useful.

It's best to wait an hour or so after astro dark begins before doing the tests to get a more realistic sky background value as the sky gets darker for a while after astro dark 'officially' begins. 🙂

If you tell me what camera you're using Adam, and the gain and offset values you've used, I can calculate the sky background ADUs for you if you don't fancy doing the maths. 😁

Same for you Edward, if you want to use a gain other than 120, but that looks a good value to use, at least for starting out.

For narrowband there is some merit in using higher gain, twice unity is often used though I tend to use unity gain all the time. Just being lazy to avoid having to have multiple sets of darks. 😁

Alan

Share this post


Link to post
Share on other sites
4 hours ago, symmetal said:

It's best to wait an hour or so after astro dark begins before doing the tests to get a more realistic sky background value as the sky gets darker for a while after astro dark 'officially' begins. 🙂

If you tell me what camera you're using Adam, and the gain and offset values you've used, I can calculate the sky background ADUs for you if you don't fancy doing the maths. 😁

Same for you Edward, if you want to use a gain other than 120, but that looks a good value to use, at least for starting out.

For narrowband there is some merit in using higher gain, twice unity is often used though I tend to use unity gain all the time. Just being lazy to avoid having to have multiple sets of darks. 😁

Alan

HI Alan, I'm using the ASI1600mm pro. 

For RGB I'm using gain 75 and NB gain 200. I think I've got it set to use the default offset, which I think was 25 (although on my filenames it says offset N/A so not sure whether it actually uses the offset?)

I did some rough calculations myself (assuming it doesn't use the offset) and came up with 896 for NB and 320 for RGB. Or 1296 (NB) and 720 (RGB).

Would be useful to know if I'm in the right ballpark figure.

Share this post


Link to post
Share on other sites
1 hour ago, Adam1234 said:

HI Alan, I'm using the ASI1600mm pro. 

For RGB I'm using gain 75 and NB gain 200. I think I've got it set to use the default offset, which I think was 25 (although on my filenames it says offset N/A so not sure whether it actually uses the offset?)

I did some rough calculations myself (assuming it doesn't use the offset) and came up with 896 for NB and 320 for RGB. Or 1296 (NB) and 720 (RGB).

Would be useful to know if I'm in the right ballpark figure.

Hi Adam,

I assume you're using the Ascom driver for the 1600 as your filenaming didn't know what it was. It's a limitation of Ascom that offset is not a parameter that is available to the capture program using the driver. If you use the native Zwo driver then the offset parameter is available to the capture program to be read and written to. If using the Ascom driver then it can only be set in the driver control panel itself. The capture program will give you the option to access the Ascom driver and set the offset on the camera, but the capture program has no way of knowing what you set it to. To view the Ascom offset value you have to click the 'Advanced' box on the driver shown here

1791425665_AscomDriver.png.ad4e0d87c98b47236068604dd862dedf.png

When it first came out they was much debate on what offset value is optimum to use for different gain settings, but it became such a pain having to alter it all the time, and the fact that altering the offset has an insignificant effect on dynamic range, it's not worth worrying about and just set it to a high enough value where it works at all gain settings without black clipping data. In later driver updates Zwo set it to be 50 and hid it behind the Advanced box so it wasn't easily changeable by accident. I have mine set to 56 above but have now had to increase it to 64 (the same as vlaiv has his 1600 set to) as I was still getting slight black clipping on dark flats of about 1 sec duration.

You may be wondering why seemingly adding fixed ADU values using offset works and what it actually does. If it just added a fixed number of digital ADU units it wouldn't really do anything useful. But offset and gain work in the analog domain and not the digital. The analog voltage signal value read from each pixel is processed by an amplifier before it does to the A-D converter to create the digital ADU value. You want the min and max signal voltages read, to lie between the A-D converter reference analog signal voltage limits which correspond to ADU values of 0 and 4095 (for a 12 bit camera like the 1600). The A-D converter may be designed to accurately convert an analog input signal between say 0.5 and 1.5 volts for example. If the A-D converter is powered just from say 5V, you don't want the input signal too close to the 0V power rail as things may get non-linear so a value above 0 volts is taken as the minimum value into the converter.

You can now see what the offset value does. It adds a fixed DC voltage to the analog input signal such that the lowest signal value input to the A-D converter is say 0.5V, using the example above. A covered sensor pixel may give an output between plus or minus a few microvolts, depending on the noise present. You want all those values to be correctly converted so add a fixed positive DC value (corresponding to the offset) to all the pixel readings so they will always be within the range of the converter.

The gain just amplifies the signal so more digital units can be used to represent the smaller analogue value, so getting a higher digital resolution. Of course, if the gain is too high you get excess white clipping and loss of signal information. Bright stars will always end up getting white clipped so gain setting is a compromise. Similarly, too low an offset will give black clipping and loss of signal or noise information.

Back to your post, 😁 your ADU values are a bit higher than mine. You have to plot a curve to fit the graphs so values between the plotted points are a bit lower that those shown by the straight lines. 

At zero offset and camera gain 75, I got 2.00 read noise and 2.15 gain giving ADU 298. At camera gain 200 I got 1.45 read noise and 0.50 gain giving 673 ADU.

Your adding of 400 to each value for offset 25 is correct.

As it's most likely you are using default offset 50, your sky limit ADU values to aim for become 1098 ADU at gain 75 and 1473 ADU at gain 200.

Hope that helps Adam. 🙂

Alan

Share this post


Link to post
Share on other sites

That's brilliant thanks Alan, I was more or less in the right ballpark then, the read noise and gain values I used for my calcs were off from yours by a decimal place as I very crudely read the graphs for my camera. 

Come to think of it I think it is the default of 50 offset I'm using as I didn't change it.

Thanks for your help!

  • Like 1

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.