Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

Offset


Recommended Posts

For starters this topic:

And here is practical example why offset is important, or rather why you don't want to have 0 values.

Imagine having a bias file (or dark with very low dark current). Read noise in it is random, Gaussian distributed (well, it better be with exception of some FPN/TN in cmos sensors) with some mean value close to 0. Let's take that it's mean value is 1. You take some subs and actual pixel values will be 1, 0, -1, 3, 1, -2, 1, -1, ... when you average those out you get 1.

Now imagine that when you write your subs, you can only write non negative values (only 0 and positive numbers get read out of the camera), and all the negative values are clipped to 0. Now above sequence looks like 1, 0, 0, 3, 1, 0, 1, 0 ... and clearly now mean is no longer 1 but some value higher than 1.

When you calibrate with such file you will not be subtracting proper bias, you will be subtracting higher value than you should be, and your calibration fails.

Similar thing happens to your lights as they also contain bias signal. In places where light is so weak, or there is no light at all, you will have some positive value instead of 0 (no light), so you mess up a bit your lights as well (this can hardly ever happen with today's LP levels, but you are offsetting your measured values anyway and getting the wrong reading).

This simply means that you are adding some nonrandom noise to your image - a bad thing to do, especially if you can easily avoid it by adjusting offset.

Just to add, when you read 0 in your subs - you can't be certain if it is "proper 0" or negative value that has been clipped to 0 (it's random after all) - so it's best to be sure and have all values greater than 0.

Edited by vlaiv
  • Like 2
  • Thanks 1
Link to comment
Share on other sites

Simplest one that I've found is to just test it. You can use bias subs for this, but I think with CMOS sensors, better way to do it is to get darks of shortest duration that you are going to use - possibly for "filler" subs - like 10-15s (those subs you take if you want to avoid star core clipping and over exposure in the strongest parts of the image).

Take about a dozen of them - stack them using minimum method, no fancy sigma clip or anything, just plain old minimum and do stats on resulting stack - look at lowest pixel value in result - if it's bigger than 0 you are good to go. If you find that it is 0 (or smallest value reported by your camera - some cameras like ASI1600 don't actually give you 0, but rather 16 as lowest value) - raise offset and repeat.

It's better to go higher in offset than lower - only thing that you "loose" with higher offset is a bit of full well capacity.

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

Been racking my brain to try to remember what I've used in the past.  Wondering if I've mentioned it in any of my threads but I don't recall doing so.  Thank you for the information.

Link to comment
Share on other sites

21 hours ago, vlaiv said:

Simplest one that I've found is to just test it. You can use bias subs for this, but I think with CMOS sensors, better way to do it is to get darks of shortest duration that you are going to use - possibly for "filler" subs - like 10-15s (those subs you take if you want to avoid star core clipping and over exposure in the strongest parts of the image).

Take about a dozen of them - stack them using minimum method, no fancy sigma clip or anything, just plain old minimum and do stats on resulting stack - look at lowest pixel value in result - if it's bigger than 0 you are good to go. If you find that it is 0 (or smallest value reported by your camera - some cameras like ASI1600 don't actually give you 0, but rather 16 as lowest value) - raise offset and repeat.

It's better to go higher in offset than lower - only thing that you "loose" with higher offset is a bit of full well capacity.

thanks, where do you find the pixel values for an image?

Link to comment
Share on other sites

25 minutes ago, Anthonyexmouth said:

thanks, where do you find the pixel values for an image?

What software do you use for stacking and analysis?

I believe PI has some sort of statistics window - it should provide such information. I use ImageJ and it has Analyze feature that provides you with min, max, average, median, stddev and such.

Link to comment
Share on other sites

A very steep learning curve if you ask me.  I'm still getting to grips with it.  I think it's superb and probably nothing it doesn't do (except make the tea) well worth the money IMO.

Link to comment
Share on other sites

I have found @vlaiv insight and understanding on this invaluable. Really interesting you used :

10 minutes ago, david_taurus83 said:

5 offset

 I followed vlaiv's instructions to the letter except I used dark-flats rather than short duration darks and integrated them using 'minimum' in PI. The minimum I achieved was zero with an offset of 50, gain 139. I've decided to up my offset to 56 just to ensure I stay above zero but to date I have not exceeded 240 s exposures and most have been at 120 s. After two plus years of using a CCD this CMOS stuff is all very new to me.

 

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
  • 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.