Superbias with Partial Bad Columns

[Astrodoc]

The threads about Superbias don't seem to have had any posts for a few years. After superbias was initially released it was said that it would be updated to handle master bias frames with partial bad columns. Has this been done?

I've attached a screen shot showing that the superbias differs from the master bias in the partial bad columns.

Left: master bias made with 200 subs
Right: Superbias made from the master bias

Both are displayed with 24-bit LUT, and identical screen stretches.

Clear skies,
Ron

[Niall Saunders]

Hi Ron,

Whilst I don't personally use SuperBias frames, as far as I can see your SuperBias frame seems to be an accurate transformation of your MasterBias frame - the SuperBias has successfully dealt with nearly all of the small-scale noise, without introducing any artefacts.

As far as your 'bad column' issues are concerned, I am trying to think what the MasterBias and SuperBias frames should depict for these issues. Yes, the MasterBias shows exactly 'where' the faulty columns are, and the SuperBias - rather than 'eliminating' them, has removed any low-scale noise associated with them, leaving them standing out quite prominently.

Of course, when these SuperBias 'bad columns' are processed during ubsequent calibration stages, then the likes of the MasterDark would no longer show the bad columns - which is what you would be hoping for.

Maybe I am missing something here?

[Astrodoc]

Thanks for the information. In the past discussions, when the tool was introduced, it explicitly said it did not work on partial columns, and that a fix was in the works. I just want to know if the fix was implemented.

Note that my master bias was made from 200 individual bias frames, so it's relatively clean. I just want to close this loose end in my knowledge and, if it now works on bias with partial bad columns, I will evaluate it.

Clear skies,
Ron

[Niall Saunders]

Well Ron, like I said, I don't work with SuperBias frames, so I can't really pass further comment.

Hopefully though, someone will pick up on this and might be able to give us an insight on how, or if, the Process is being, r has been, developed.

[daimoniro]

Hi all,
I'm doing very first pre-processing with PI.

I have a big difference between bias and superbias.
It is normal?

I have a CANON EOS 1100D, bias done with 40 subframes with "Linear Fit Clipping" for Rejection Algorithm in ImageIntegration process.

Lorenzo J.

[Niall Saunders]

Hi all,

If you are looking at your MasterBias, or SuperBias, frames - and trying to decide whether you have 'issues' - you should really start by examining the Max, Min, Avg and StDev figures for  each of these frames. Remember, you are not looking at your camera's response to incoming photons. Instead, you are looking to see how the electronics in your camera respond in the absence of photons (or, at least in the absence of photons coming down the OTA from your target).

In a 'perfect' imager, you should expect to have a perfectly uniform, 'grey' background - where all pixels report the same ADU, with that ADU figure being very small, but (most importantly) that small ADU being non-zero. You need a minimal, non-zero, ADU value to allow the electronics in the camera to do their job, and to allow basic pre-processing stages to identify what that 'bais level' is - for your imager (making it easy to then 'remove' the bias level from all images during the calibration stage).

However, your camera (everyone's camera) will deviate from 'uniform grey' - showing either random noise (which is great) or some type of 'pattern' (which is still not really an issue - providing it has been faithfully preserved in the MasterBias frame). [In fact, if you have a camera whose Bias is very close to 'uniform grey' you can actually eliminate the use of a real master or super bias frame altogether, and just replace it with a frame where all ADU values, for revery pixel, are actually perfectly equal to the 'uniform grey' value].

Now, as I understand things (I have so little 'random noise' in my MasterBias frames that I do not - currently - bother with SuperBias frames - so I may not have enough experience here to be 100% accurate) the point of converting a MasterBias frame into a SuperBias frame is to eliminate the random noise, without removing the major artefacts, such as bad columns, leaving the SuperBias faithful to the original performance of the imager electronics, whilst improving SNR prior to subsequent processing (i.e. image calibration).

The attachments in the other posts on this thread clearly show that the SuperBias image is 'smoother' (i.e. it has better SNR) but that the 'bad pixel' patterns have not been deleted, or diminished in any way. In fact this is exactly what I would want - that the non-random pattern(s) in the bias frames is actually enhanced, simply because the SuperBias frame has a better SNR.

Does this make sense?

[drmikevt]

I discussed this exact topic with Vincent not too long ago and, while I certainly can not speak for him, what I had gleaned from the conversation was that Superbias can not handle chips with partial column defects (I used to have a chip with one and he showed me how it was introducing artifacts compared to a stacked bias).  Apparently, the process was originally intended for CMOS cameras which handle columns differently.  His recommendation was not to use it with CCD's, but if you don't have a partial column issue I think it can work well. 

Again, this is my understanding of what was said.

Mike

[Astrodoc]

Thank you, Mike.