Generation of calibration frames and use of Bias frames for CMOS

Dear Björn @barnold84,

Indeed, Bias are not recommended with the new CMOS camera's. Technically, if you shoot bias frames with the shortest exposure time possible, then things will behave badly with these new CMOS sensors. If you cheat, by shooting bias with 0,1 second exposure time, then all should be okay though in my experience. Fundamentally, it is not a real bias then, but it will work!

If you load both MasterBias and MasterDark, APP will automatically subtract the MasterBias from the MasterDark when calibrating your data to make sure that all works as needed.

The MasterDark in it self will never be bias subtracted in APP when we create the Master. This ensure that the MasterDark can always be used

1. When you only load the MasterDark without the MasterBias
2. When you load both MasterBias and MasterDark, then APP will automatically and dynamically subtract that MasterBias from the MasterDark when calibrating your lights.

Please let me know if this answers your question 😉

Mabula

Actually I don't reckon the rule holds any more.

Master bias frames with my old ASI294MC pro made the results of an integration worse so I deleted the ones I shot and did not bother again, however they appear quite effective with my newer ASI2600MC Pro. I still use master darks created without a master bias but I shoot variable length sky flats and calibrate them with a master bias.

I did do an integration once with a master bias and no master dark as I shot something at a gain I had not used before, so while I could shoot bias frames quickly I could not do the same for darks, that came out quite nicely and the difference later when I shot darks to produce a master dark for that gain was very minor, and that was with 600 second exposures.

Dear Mabula, @mabula-admin

Thanks for the explanation. Regarding the capturing of the Bias I prefer not to cheat but rather to understand what's going on in order to make a better informed decision which approach to take. Until now, I always shot the Bias frames with lowest possible exposure time which I think was 1/8000s.

Another question from my side would be what is a "modern" CMOS? I have an Altair 183M with a Sony IMX183 sensor. I guess there's a certain property to distinguish between the "modern" and non-modern CMOS, and not the date of market introduction? 

Just for a better understanding: in APP, the Master Darks are basically the average of the Darks. The role of the Master Bias is to determine the BadPixelMap, otherwise they're doing not much in that process?

I'm actually wondering why I would need a MasterBias when calibrating. The MasterDark, when created without subtracting the MasterBias, contains read and thermal noise which is also contained in the Light frame. If one subtracts this, one gets the signal (roughly speaking, there's statistical variation of course). Why would I subtract the read noise somewhere again, given one isn't using Master Dark scaling?

CS!

Björn

Hi @the_bluester

Generally, I started the approach to use the lowest gain on my camera and use longer exposures instead (300s+). For narrowband (which I will start soon), I will need to find the right gain setting to keep the total exposure time limited (my concern would be the long term guiding accuracy).

I also have an Atik 428EXm, which is a CCD. I've used it with darks but according to Atik one could also try going without darks as the dark current seems to be very low if cooled appropriately. Bias should suffice with this camera.

Björn

With my ASI2600 I have shot exclusively on the lowest gain until recently when I started shooting through narrowband filters with it (Yes it is a Colour cam) With the HA and Oiii filters I have at the moment I found I was able to shoot at gain 100 (The highest regular gain for that camera) without saturating anything but the brightest stars so I used it for the read noise advantage.

For your question in the previous post. If you create a master dark without using a master bias frame then you don't want to use a master bias in the integration as the darks will already subtract the read noise.

I use a master bias doing my master flats but not my master darks, so I load the master bias and the flats for a session, create the master flat with them then unload the flats and master bias before going on to load the master dark and the light frames. So my master flat is calibrated with the master bias but the lights are not as the bias information is included in the master dark.

One of these days I might shoot new darks and try calibrating a master dark with a master bias, that should allow me to just load flats, master dark, master bias, lights and go, given that the ASI2600 seems to produce good bias frames.

Posted by: @barnold84

Dear Mabula, @mabula-admin

Thanks for the explanation. Regarding the capturing of the Bias I prefer not to cheat but rather to understand what's going on in order to make a better informed decision which approach to take. Until now, I always shot the Bias frames with lowest possible exposure time which I think was 1/8000s.

Another question from my side would be what is a "modern" CMOS? I have an Altair 183M with a Sony IMX183 sensor. I guess there's a certain property to distinguish between the "modern" and non-modern CMOS, and not the date of market introduction? 

Just for a better understanding: in APP, the Master Darks are basically the average of the Darks. The role of the Master Bias is to determine the BadPixelMap, otherwise they're doing not much in that process?

I'm actually wondering why I would need a MasterBias when calibrating. The MasterDark, when created without subtracting the MasterBias, contains read and thermal noise which is also contained in the Light frame. If one subtracts this, one gets the signal (roughly speaking, there's statistical variation of course). Why would I subtract the read noise somewhere again, given one isn't using Master Dark scaling?

CS!

Björn

Hi Björn @barnold84,

Well, it is hard to say which sensors are modern versus non-modern... because it is not only the sensor... it is also the electronics around the sensor and the build quality involved which will have an effect on this problem I think. So the CMOS bias issue is not that black and white... You need to know and learn the specifics of your camera, which I think is a good thing to get the best out of it.

"The role of the Master Bias is to determine the BadPixelMap, otherwise they're doing not much in that process?"

No, the masterdark is needed to create the BadPixelMap for detection of hot pixels. MasterFlat is used for cold/defect pixels. MasterBias is not involved in BPM creation.

"I'm actually wondering why I would need a MasterBias when calibrating. The MasterDark, when created without subtracting the MasterBias, contains read and thermal noise which is also contained in the Light frame. If one subtracts this, one gets the signal (roughly speaking, there's statistical variation of course). Why would I subtract the read noise somewhere again, given one isn't using Master Dark scaling?

Very valid point! Indeed, the bias is in every frame, so it is also in a dark 🙂 So a single MasterDark contains the bias as well. There is no need to create additional Bias then. But... realize that if you use flats, you need to make sure that flats are calibrated with bias or darkflats, otherwise, flat-field calibration can never work as expected. The Sensor offset/pedestal needs to be subtracted from both lights and flats to make flat-field calibration work properly.

So if you don't use bias, but do use flats, you need for proper data calibration including flat-field calibration: darks, flats, darkflats and your lights...

Mabula

Hi Mabula, @mabula-admin

Thank you for the explanation. I didn't have the need for dark flats yet but I guess with narrow band I'll have exposure times that justify the use of dark flats.

Just once more for the Bad Pixel map. This IMX183 sensor has a strong "amp glow" and when I create a BPM with darks with long(er) exposure times (180s+) it seems that the star burst nearly fully goes into the BPM. I'm wondering if it should or not. I've used shorter exposures for the BPM then (30s).

Björn