Hello, I have just read the interesting article on using synthetic bias. I have taken 5 bias with different offsets and entered the info into a spread sheet. This is where I get lost. I have a 16 bit camera and in the article it says you cannot determine the setting to use with a 16 bit cam but that the setting for the asi zwo2600 is 10. My camera is the qhy268c which uses the same chip. How do I determine the setting to use for my camera. Attached is the spreadsheet I created and am not sure I did that correct. I multiplied the median by 65535 and I think that is what I was supposed to do but dont know to proceed past that Thanks for any help
Cheman
offset.ods (11.7 KB)
Hi Cheman,
you are almost there! I took the data in your spreadsheet and plotted Median Level in ADU vs Offset setting. And then added a linear trendline forcing the intercept to be zero. And I get this:
So you can see that the Median Level is 16 x Setting. You can drop the .02, we are looking for an integer value. And so you can also note that despiste the fact that it is the same chip than ZWO 2600, QHY have made a different choice regarding how offset setting and Median Level in ADU relate.
The point that was being made about the 16b ADC is that the multiplier (16 in your case) could be any integer value. Whereas, if you have a 14b camera, you would know for sure that the multiplier must be a factor of 4. Turns out in your case, 16 is also a factor of 4 which is making my explanation even more confusing I realize
Hope it is clearer anyway.
So now, in Siril, when you calibrate your flats, you can either pass for the bias (instead of a masterbias):
=16*$OFFSET
or if you are using always the same offset setting of, say, 30, you can pass:
=480
Here you go, easy!
Cheers,
Cecile
Thank you very much for your evaluation. I will give that setting a try. I have one more question. I do not use darks. Can my lights also be bias corrected using this setting as well in the equation of
L-O/F-O
Thanks
Cheman
I would be a bit more cautious about not using darks, for mainly two reasons:
- ampglow which can be important to substract from your lights. That being said, this sensor is known for having almost no amp glow so probably a good candidate for such test.
- hot pixels suppression. What siril does when you give a dark is first the subtraction of said masterdark. But it also detects deviant pixels in the masterdark, and apply cosmetic correction to the corresponding pixels in each calibrated light (except in manual mode where you can choose or not to apply this correction).
So basically, if you have lots of subs, sufficient signal and you have dithered, the hot pixels should be caught during stacking at the rejection step. Otherwise, you may see some occasional dots and that would advocate for using a masterdark. Anyway, with a cooled camera, you can shoot a bunch for each exposure, prepare a masterdark library and be done with it for a while.
But yes, if you want to give a try, use the offset level and a masterflat to prepocess your lights, the field should be very very close to flat.
And pls do report back, very interested to see your results and conclusions.
C.
thanks once again for your support on this and your project of Siril. I have modified a script to do all prior to stacking which I will attach along with my result that does not seem to do quite as good a job as using standard bias. I could be my script has an error or something else but the script does complete. I tried to attached resulting stacked file unstretched in fits format but its not supported in the forum. Let me know if there is another format you would like and if you want stretched or unstretched data.
Thanks again
Cheman
OSC_Preprocessing_withbackgroundextraction_NoDarkNoStack synthetic bias new.ssf (1.4 KB)
as an alternative I can send you a dropbox link if that helps
Here is a link to a folder with two fits files. Both used the same script except the synthetic bias one used the script I posted that was altered for the syn bias. Dropbox - Siril - Simplify your life
Hi again,
just had a look at your ssf, thanks for sharing. I think the light calibration line is not quite alright. And I think you found a tiny bug that we’ll need to correct because it should not have run sucessfully.
Your lights have not been calibrated with the masterflat, which most probably why you see a wrong correction.
Could you try again, this time with this line (line 60 of your modded ssf)?
preprocess light -bias="=16*$OFFSET" -flat=pp_flat_stacked -cfa -equalize_cfa -debayer
And I think if you want to get a final answer on that, your should also run the same case but using a masterdark. So that if the stack with (L-O)/(F-O) does not come as good, we know for sure it is because of the synthetic offset rather than because of skipping as well the masterdark part.
C.
Hi
I ran the script with the corrected line. I’ve attached the result link and the log for you. I have not as yet had the chance to compare the result. Depending on results I will try running with darks as well but not sure when I can get to that as I will have to take those frames. I have been very happy with my results not using darks however. Thanks once again.
Hi,
just went through the log, now the masterflat has been correctly applied. I’ll fix the small bug just now so that the faulty line is caught. Let me know when you’ve had time to compare the synth bias approach with your normal workflow. Can’t fetch your fits in a reasonnable amount of time, my Internet connection is really slow…
C.
I have run the scritps with (1)regular bias and no darks (2) synthetic bias with no darks and (3) synthetic bias with darks. All three give very good results as far as I can tell visually before and after post processing. If there is much/any difference statistically I really dont know how to tell. If there is a way for me to look into that, I can try with some guidance 
thanks
Che
