Super Bias (examples)

[steven_usa]

Based on a suggestion at another thread ( http://pixinsight.com/forum/index.php?topic=7434.0 )    it was suggested to improve my master-bias by using Super Bias.   Indeed, my images did look better after that, and it even allowed some of the stacks to align that didn't previously.

And now that I have access to my camera again, I decided to see how more bias RAWs would compare to my original set and the super bias.   

For reference, my original bias set included 20 RAWs, and they are available here:  https://app.box.com/s/s2e23tquxxq8i4oj7eos
The master version of all 3 combinations (the 20 count, super bias, and the 202 count) are available here:  https://app.box.com/s/hjs0mt9i7b90xvzaa5jv

Attached below are screenshots of the 24-bit AutoSTF stretched in PI, for comparison...


I have not yet re-calibrated my images with this new 202-count master BIAS, but I will over the next few days.  Any prediction on if I'll notice any difference?   Is 202 still not enough, or too many?    Is the artificial super-bias pretty good compared to the "natural" 202-count?    What are the white dots in my bias (moreso in bottom left quadrant)?

[steven_usa]

Note, while I'm at it... I went ahead and stacked 402 total bias RAWs at ISO400 from my 450D (and this time set the high/low sigma to 10.0 each while using BPP to script a stack).

To me (as expected), I saw barely any difference between the 402 count stack and the 202 count stack in the resulting master bias.  That's expected since I've read that not much past 50 offers really noticeable improvement.  That being said, is it possible to have too many?  Other than you'll just mostly end up wasting integration time.   


THEN, just for fun, I created a Superbias using this new master-bias created by the 402 count of bias RAWs (1/4000 exposure at ISO400).   I did it twice, once with the Multiscale Layers defaulted to 7, and again with the max value of 10  (I haven't messed with any of the other parameters of superbias).  The only real difference I see visually with the Multiscale 10 version of the super-bias from the 402 count stack is that it has less "haze" along the top.

And not that it matters too much, but another observation is that the superbias files are less than half the filesize as the regular masters.

The additional master-bias combinations are available at the share:  https://app.box.com/s/hjs0mt9i7b90xvzaa5jv


Again, the 402 count stacked bias (with 24-bit AutoSTF) didn't look much different than the 202 count stack.  But attached here is a side-by-side comparison of the Multiscale 10 (right side) versus 7 (left side).  I would suspect a superbias of the 202 stack wouldn't look much different either.


Question is: any point in using a superbias of such a large number of stacked biases already?

[mads0100]

Based on the original thread on Superbias, no, it doesn't make a difference if you're willing to take the time (and shutter counts) to get 2-300 bias frames.  It's just mechanical wear and tear at that point.

The big benefits for a super bias are when you're using like 30-50 bias frames. 

Here is the link:
http://pixinsight.com/forum/index.php?topic=7312