Settings on SGPro for ASI1600 Gain/Offset

[blinky]

Hi

Just wanted to confirm these settings set the ASI 1600 to Gain300 Offset50 - I mean do these settings and the camera selected set the camera to these settings?  Just reading about which driver to select that allows SGPro to set the Gain/Offset that all....

 

 

[han59]

For the ASI1600, in most cases  you better stick to unity gain=139, offset 21.  Especially if you imaging deepsky and want to use the full 12 bit range. All other settings give less range. Your light , flats and darks should be made with the same settings.

 

 

[symmetal]

As I mentioned in your previous post the Offset cannot be set outside the Ascom driver. As soon as you select any camera from the drop down menu the SGP Offset  setting will be greyed out reading 'NA'. You can set the gain there if you wish but as it is of limited usefulness and is easily overlooked and is not immediately viewable on the event. You should be able to add the gain into the file name using the file name pattern options, but I never got that to work when I tested it. Just use the Ascom driver itself to set the gain and offset, and put the settings into the 'suffix' field as a reminder to what the settings were for the event.

As han59 said it is much easier to leave it at unity gain/offset for everything especially when you are starting out. Vlaiv convinced me that was best anyway, and any 'benefits' of changing the gain have to be weighed against the downsides. If your alignment/guiding is not too good and you get sausage stars after a couple of minutes then there is a case for using higher gain at a shorter exposure but if you can guide well then stay at unity gain is my advice. ?

At unity gain you have a full well capacity of just over 4000 electrons which matches pretty well with the 12 bit range of 4096 levels, 1 electron = 1 ADU. At gain 300 your full well capacity is down to about 650 electrons, with 1 electron = 5 ADU. This equates to effectively just over 9 bit resolution.

When stretched this will give significant posterization in the image unless you take hundreds of subs to compensate. 

Alan

Edited February 8, 2019 by symmetal

[blinky]

From what I read it was the other way round,that was why I asked,I'm sure I  read that the offset cannot be set in the ascom driver and that was why there is a specific ZWO camera selection to allow this to be set per filter 

[blinky]

Oh and sorry I must be going senile, forgot I had already posted that other thread!

 

[blinky]

This is from the sgp forum and it's by one of the revs and it seems to say the opposite of symmetal. This is why I'm confused 

With the native you can set gan and offset per event within SGP. With ascom
you cannot. Essentially just more control over the gain and offset with the
native. But if ascom is working just fine then no real reason to change
either.

 

[symmetal]

Sorry blinky, just seen your reply and I have to apologise to you. It seems SGP now supports the ZWO cameras using the native ZWO drivers. It looks like it was included in SGP V3. ? In the past SGP devs said they don't want to use native drivers as it means possibly having to update SGP each time the manufacturer updates their drivers. Looks like too many people clamoured for offset control and they had to give in. The 'ZWO Camera' is the native driver while 'ASI Camera (1)' or (2) are the Ascom drivers

So to answer your original question, yes, with the 'Zwo Camera' driver, setting the gain and offset settings per event are available where you've shown. 

Setting Priority Level 1 The gain and offset can be set using the wrench icon next to the camera driver selector.

Setting Priority Level 2 The gain can be set for different binning settings in the 'Control Panel/Camera' tab.

Setting Priority Level 3 The gain and offset can be set using the 'Sequencer/Event/Options' as in your screen shot above.

Level 3 has the highest priority so will over ride any lower level settings. If set to 'Not Set' the settings from the level below will be used.

Hope that helps blinky. ?

Alan

Edited February 8, 2019 by symmetal

[symmetal]

15 hours ago, blinky said:

I'm sure I  read that the offset cannot be set in the ascom driver

This should read 'The offset cannot be set via SGP using the Ascom driver' as per my first post above. 

14 hours ago, blinky said:

This is from the sgp forum and it's by one of the revs and it seems to say the opposite of symmetal. This is why I'm confused 

With the native you can set gan and offset per event within SGP. With ascom
you cannot. Essentially just more control over the gain and offset with the
native. But if ascom is working just fine then no real reason to change
either.

The relevant words are 'within SGP' When you click on the driver settings using the wrench icon you are effectively leaving SGP and manipulating the driver yourself. It's a subtle difference but may make the above statements clearer. ?

Alan

[blinky]

No problem, thanks for clearing this up for me

[alcol620]

Just downloaded the latest update of SGPro and got the message that ZWO native driver no longer supported by SGPro. I ignored this and carried on as I want to be able to set gain and offsets for each activity. Can some one explain what "not supported" means, SGPro still worked fine.

[Starflyer]

12 minutes ago, alcol620 said:

Just downloaded the latest update of SGPro and got the message that ZWO native driver no longer supported by SGPro. I ignored this and carried on as I want to be able to set gain and offsets for each activity. Can some one explain what "not supported" means, SGPro still worked fine.

They won't give you support of you have problems with the native driver and AFAIK from now on they won't implement any native driver updates.

[alcol620]

5 minutes ago, Starflyer said:

They won't give you support of you have problems with the native driver and AFAIK from now on they won't implement any native driver updates.

Presumably, providing I have no issues I can carry one. I read a link some months ago setting out how importent it was to get the right Gain/offset balance for the ASI1600 ZWO cameras. The latest SGPro update takes away this ability.

[blinky]

Offset is not important - as long as you have enough offset to not get 0 signal pixels you are fine.  This is why its never been exposed in Ascom!  The Ascom driver has presets for highest dynamic range, lowest read noise and unity gain - you can also set gain manually to whatever you want.

[alcol620]

1 hour ago, blinky said:

Offset is not important - as long as you have enough offset to not get 0 signal pixels you are fine.  This is why its never been exposed in Ascom!  The Ascom driver has presets for highest dynamic range, lowest read noise and unity gain - you can also set gain manually to whatever you want.

I have to admit, although I have read lots about gain and offset, I don't fully understand what I am reading.  How would you set up to do an evening of imaging both NB and RGB, Currently I would set up with 139g and 21 offset for the RGB images and 200g 50 offset  the narrowband. Without the option to set these for each element of the SGPro imaging run how would you set up for such an imaging run? Sorry to be a pain.

[Adreneline]

18 minutes ago, alcol620 said:

How would you set up to do an evening of imaging both NB and RGB

I don't wish to introduce a red herring but the ASIair does not allow any adjustment of offset - it gets set to 50 and that's it; you can adjust gain to suit. Prior to using the ASIair I used SGPro and set offet to 56 based on advice from a post by @vlaiv. It worked for me but I don't know/understand enough to do otherwise. It will be interesting to see what the ASIair Pro allows.  

 

[symmetal]

Offset is just a fixed DC voltage added to the analogue value read from the sensor pixel to avoid an analogue value of zero being fed to the ADC (Analogue-Digital Converter). Depending on the gain value applied to the signal, by the analogue amplifier also before the ADC the minimum offset required to avoid a zero value also changes. 

Having too high an offset limits the dynamic range available from the ADC but in reality this change in dynamic range is negligible. It's much more simple to just use an offset that avoids zero value ADC outputs at any gain setting you use as blinky says. I use a value of offset 56 on my ASI1600 at gain 139 (unity) just to be doubly sure of avoiding zero values. You using a value of 50 for all your gain settings should be fine.

Alan

Edited February 8, 2020 by symmetal

[vlaiv]

2 hours ago, alcol620 said:

Just downloaded the latest update of SGPro and got the message that ZWO native driver no longer supported by SGPro. I ignored this and carried on as I want to be able to set gain and offsets for each activity. Can some one explain what "not supported" means, SGPro still worked fine.

I advocate use of ASCOM drivers over native for long exposure imaging. If you use ASCOM drivers, you should be fine and not affected by this change. If not - make an effort and switch to ASCOM drivers. You will need to redo your calibration subs (at least darks - you should be fine with existing flats if you have permanent setup and reuse flats).

1 hour ago, blinky said:

Offset is not important - as long as you have enough offset to not get 0 signal pixels you are fine.  This is why its never been exposed in Ascom!  The Ascom driver has presets for highest dynamic range, lowest read noise and unity gain - you can also set gain manually to whatever you want.

For this model, it is actually exposed and needs to be set properly. From what I've seen - it is exposed for other ZWO camera models as well.

Fact that you have offset setting and that it was set to 21 at one point by default - meant that there was some clipping with darks and as a consequence there was more noise in subs and sometimes flat calibration failed because of wrong offset settings. This happened for people with low offset setting (quite a bit of clipping and changed mean dark sub value) that were imaging in higher LP (LP acted as offset from lights and there was no clipping for those - and when you subtracted darks that clipped - there was wrong mean signal level that caused issues with flats).

[symmetal]

The easiest way to check on whether you have zero value ADC outputs is to take a bias frame and load it into fits liberator or your capture software to display the histogram. You may need to apply a log stretch to show the black end more clearly. The image 'hump' in your histogram corresponding to your bias frame data should not be clipped off on the left hand side, that is the full data distribution of the hump should be above a value of zero.

Dead/bad pixels may give a zero value so ignore spurious spike values well away from the bias hump.

Alan

[vlaiv]

5 minutes ago, symmetal said:

The easiest way to check on whether you have zero value ADC outputs is to take a bias frame and load it into fits liberator or your capture software to display the histogram. You may need to apply a log stretch to show the black end more clearly. The image 'hump' in your histogram corresponding to your bias frame data should not be clipped off on the left hand side, that is the full data distribution of the hump should be above a value of zero.

Dead/bad pixels may give a zero value so ignore spurious spike values well away from the bias hump.

Alan

Actually - I recommend leaving some headroom and completely avoiding any minimum values in multiple subs.

Your advice is sound - looking at the histogram and making sure it is not clipped is what it is all about, but with CMOS sensors, you really need some head room just to be safe.

If you want to be safe - determine minimum value that your camera outputs. It is often either 16 or 4 for 12 or 14 bit ADC cameras - in case of ASI1600 it is 16 for example. Take number of bias subs, stack them with minimum method (instead of average - just plain minimum stacking no sigma stuff) and inspect that in software like Fits Liberator. In fact - Fits liberator will give you stats panel and besides checking histogram out - you can check stats panel and observe minimum pixel value - it needs to be higher than min value camera is producing (only way to be certain there is no clipping). If you have bad pixels - you should remove those somehow (set to 1000 or something via pixel math) - but I think these are rare.

Why do I recommend higher offset than otherwise sound advice given above? - to make sure histogram is not clipping.

Many CMOS sensors suffer from what is called telegraph type noise that happens because electrons "leak" - either out of pixel wells or into pixel wells or between pixel wells - and that creates either higher or lower values in such pixels than would be expected by statistics of bias sub. Take a look at this example:

It is animated gif made out of 10 darks - 300s darks from new ZWO camera (6200 model). You can clearly see dark and bright pixels popping in and out "of existence" (I love that quantum mechanics jargon).

These pixels can get clipped even if your histogram is fine as they can be much lower than surrounding mean value.

With ASI1600 this feature is really prominent - it happens on adjacent pairs of pixels - look at std dev stack of darks from my ASI1600:

With this stack - brighter the pixel - more variation there in set of dark subs. It is interesting that higher values clump together in particular way - two adjacent pixels diagonally ordered seem to share "defect" - and leak electrons between themselves.

We can further observe this phenomena like this:

Here is histogram of single pixel values across 64 dark subs:

This is from a well behaved pixel that does not display telegraph type noise. It is as we would expect - resembling bell shaped curve that read noise produces.

Now look at what histogram of Telegraph type noise affected pixel looks like:

 

It has not one but multiple bell shaped peaks - it has base read noise but also depending if pixel leaked electrons or electrons leaked into pixel - it has other bell shaped curves offset from primary one.

This means that even if we capture primary bell shape with our offset - which is in my case around 62-63e (because I use offset 64 and unity gain) - there is chance that some pixels in some subs will be clipped. I know that this is very low number of pixels - roughly less than fraction of one percent of pixels is displaying such behavior - but this is advice for those that really want to take care of even tiny sources of noise.

[alcol620]

Hi guys, I want something simple.

I have a ASI1600MM camera.

What are the recommended gain settings for NB and what are recommended settings for LRGB. And what are the suggested subs exposures, in seconds. This PHD dueling is too much for me.

[vlaiv]

4 minutes ago, alcol620 said:

Hi guys, I want something simple.

I have a ASI1600MM camera.

What are the recommended gain settings for NB and what are recommended settings for LRGB. And what are the suggested subs exposures, in seconds. This PHD dueling is too much for me.

You do know that there is no correct and simple answer to that question?

I can give you simple answer. Use gain 139 and offset 56 (or even higher - I use 64) for both LRGB and NB. Use 1-2 minute exposures for LRGB and 4-5 minute exposures for NB.

Use at least 64+ calibration subs. Use darks, flats, flat darks. No need to use bias. If you get clipping (and you will on bright targets) - use filler exposures of 10 or so seconds (calibrate those with same master flat - but their own master dark of same duration).

Are above recommendations best for you? That highly depends on quality of your guiding and level of your LP. If your guiding can handle it and you have low light pollution levels - feel free to go with longer subs for both LRGB and NB then above given

[Adreneline]

2 minutes ago, vlaiv said:

Use gain 139 and offset 56 (or even higher - I use 64) for both LRGB and NB. Use 1-2 minute exposures for LRGB and 4-5 minute exposures for NB.

+1 for that recommendation although I tend to go for 60s/120s RGB and 120s/180s for NB as I get fewer wasted subs from aircraft arriving and departing East Midlands Airport!

[symmetal]

Yes, vlaivs settings are good general values. If you want simple then unity gain (and a fixed offset) for everything. The gains/disadvantages of using different gain values are relatively small. I use unity gain (139) for all, offset 56, 60s for L and 180s for RG and B, along with as long as convenient for NB, normally 480s or 600s. I expose until the sky background swamps (significantly exceeds) the read noise and these are good exposures for my Bortle 3 skies. If you have higher LP then shorter exposures are more beneficial.

Alan