Jump to content

sgl_imaging_challenge_2021_annual.thumb.jpg.3fc34f695a81b16210333189a3162ac7.jpg

Offset, erm offsetting what?


cjdawson
 Share

Recommended Posts

Hi all.

I've got a question about the settings that are available in the ZWO camera drivers.   There's two settings, the gain and the offset.

I get that the gain is e-/ADU, or to make it simple with a ZWO 1600-MM-Pro set this value to 139 and it's at unity gain.  lowering the gain makes the image darker, and raising the gain makes it brighter - but more noisy.   I don't get why I'd want to change this value from the unity value.  that's something for another day.   I'm happy enough with what the gain is to know how to set it.

 

however, and this is the big question..... offset?

What is the offset value for?   What is it offsetting?  and how do I determine what the values should be?

Link to comment
Share on other sites

When converting from analog signal of pixel value measurement to digital value that one reads of the sensor - offset is added to analog signal.

This is important part, because analog signal has some noise in it, and if we were to take just absolute value we would get clipping. For example, just reading bias frame (no light or thermal signal in there), due to noise present - value "oscillates" around 0. Some pixel values are less than zero, some are greater than zero. Digital value does not have a sign - it's a positive integer value (zero included). This means that all values below zero get converted to 0.

This is a bad thing, as shape of distribution of such signal is disturbed and average value of such signal is no longer 0, but some higher value.

To avoid this, offset is added, let's say that signal is shifted to higher value around 10 for example. Now negative values are no longer negative values, but rather values less than 10 (and positive values are above 10). This avoids clipping of negative values.

This offset is contained in bias file. It is actually contained in every file you take - bias, dark, light, flat. Proper calibration deals with this in the same way it deals with bias signal. So offset is just part of bias signal - or more precisely - it is DC / constant part of bias signal, that you can change to suit your needs.

Question is then of course, what offset value should one use? Proper offset value is one that makes sure that you don't get 0 values in your bias file. If you get 0 value - you can't be certain if it is true 0 value (negative value with offset added - like -10 with 10 offset added to be 0) or if it is clipped negative value (like -11 with 10 offset to give -1 but clipped to 0).

Best way to do this is to set gain value to what you want it to be - like unity gain - 139 for ASI1600 and then set offset to some low value - 20 for example. Take bunch of bias frames (or darks of short exposure, because bias is not very stable with ASI1600 - it behaves strangely) and examine each one. If you find 0 pixel value in any of them - you need to raise offset. You can quickly examine bunch of such frames by stacking them with minimum method, instead of average, and then running stats on resulting stack and checking minimum value of that - it it's 0 - raise offset.

You can "over do" offset, but that will only make your full well capacity somewhat smaller. I think I'm using offset of 64 on my ASI1600. It's a bit larger than needed, but I went for it just to make sure.

  • Like 1
  • Thanks 5
Link to comment
Share on other sites

Vlaiv has pretty much said it all. :smile:

If you look at the histogram curve of your image frame, (lights, flats, darks or bias), adding offset shifts the whole curve to the right, while reducing offset shifts it to the left. On the settings panel of some cameras 'offset' is labelled as 'brightness' but it's the same thing. Astrophotography processing generally involves stretching the dark areas of the images quite a lot and this is where the majority of the noise also occurs in the image. The darks and bias are used in calibration to help reduce the amount of noise present. If any frames have pixel values of zero it's highly likely that image data has been clipped, and then the calibration will not be applied correctly as the noise distribution in the frames is not representative of the true noise present. Adding a constant offset value to your frames avoids this situation happening.

Like vlaiv I found the offset values often quoted for the ASI1600 are not high enough and you get some black level clipping. I've found an offset of 56 at unity gain just avoids this. It's worth mentioning that the ASI1600 (and possibly others) actually clip to a value of 1 and not 0, (well mine does anyway). This means that the minimum output value is 16 and not zero (when converted from 12 to 16 bit resolution). So values of 16 are most likely clipped values. At an offset of 56 the minimum value I get is 48 (16 bit) which is a value of 3 from the 12 bit A-D converter. So that's just OK. :smile:

Alan

  • Thanks 2
Link to comment
Share on other sites

Thanks guys.     That makes alot of sense now.

I'm going to do some experiments with my camera to see if I can figure out what the offset value should be.   So glad that I took the time to find out what this value actually means rather than ending up with clipped images and not knowing it.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

  • 2 weeks later...

I managed to do some experiments by taking flats, darks for flats and bias frames.  The result I ended up with was that offset 52 is the minimum value for my camera.    However, I manged to mess up my settings on my last imaging run and had an offset of 21 which produced a good image, however. Something went wrong with the flats, whilst it sorted out the dust bunny rings, it didn't correct the vignetting.

I'm going to re-do this and try again.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

  • 1 month later...
  • 2 weeks later...
Guest chaz2b

I’ll speak for probably a lot of others who have inadvertently dropped in to this topic, that just went over my head!

so I’ll point, shoot and hopefully registax will correct things...I said hopefully.

p.s. I enjoyed the read though.

chaznonethewiser.

Link to comment
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
 Share

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.