5 replies to this topic

[jlecomte]

Hi,

 

I am a total beginner when it comes to astrophotography, so these questions will probably make you roll your eyes . Please, bear with me.

 

First, a little bit of context:

 

- At a high level, I understand the concept of image calibration and the need for darks, flats, flat darks, etc., although it's been only theoretical at this point... I learned about calibration by reading the excellent "The Deep Sky Imaging Primer" book.

- I have a ZWO ASI 533MC Pro on order. It should arrive (hopefully) before the end of the year.

 

Now, here are a few questions regarding calibration frames:

 

1) Bias frames and modern CMOS cameras:

 

I read somewhere that many people forego bias frames all together with these new CMOS cameras due to their extremely low read noise. Is this accurate? Or am I misunderstanding something? Not having to worry about yet another type of calibration frame would sure help to simplify the entire processing workflow!

 

2) Dark frames:

 

If I understand correctly, one of the benefits of a cooled dedicated astronomy camera over a DSLR is that because these cameras are temperature controlled, I can use them at any time of the day to take dark frames, as long as I use the same temperature, gain, offset, and exposure time for the darks as I do for the light frames (e.g, -10C, gain 100, offset 70, 30 second exposure). This is great because doing so does not take away any time that could be spent capturing the night sky. Also, if I understand correctly, one has to take a large number of these dark frames (e.g., 100) and stack them in software like PixInsight for example, and I suppose the result is called a "master dark". Assuming these statements are not completely incorrect (feel free to provide clarification if I am way off), here are a few questions:

 

a) How do you take these dark frames? I am familiar with N.I.N.A., but I don't see an easy way to do this with that software. I guess one could use the sequencer, but that sequencer relies on an object to be added as a sequence target. What do you use as a target object when you take darks in the middle of the day from within the house? Or do you just use a bogus object since it does not matter, and set the type of the frame to DARK? I am just guessing here... Or is there a better way to do it with N.I.N.A. or other (hopefully free) software?

 

b) How many dark frames does one usually stack to create a master dark? 10, 100, 1000?

 

c) How often do you need to recreate a master dark? Every 6 months? Once a year? More/less often?

 

d) I think I remember reading somewhere that it was possible to "scale" a master dark, i.e. create a master dark from 30-second dark frames, and create a version of that master dark to use with 2-minute light frames by, I guess, multiplying the value of each pixel by 4 (just a guess...) Is this correct? Can you do that in PixInsight? (Note: I don't have PixInsight, but will probably buy a license eventually since it seems to be the de-facto standard)

 

Again, I apologize if these are stupid questions. But any pointer would be greatly appreciated. Thank you so much!

 

Julien

San Jose, CA

Edited by jlecomte, 03 December 2020 - 12:55 AM.

[dhaval]

Hi and welcome to CN! 

 

Let me see if I can help - 

 

1) Think of bias as "offset" that is already present in  your camera. What it does is, present you with a base-level of readout noise when you take the shorted possible exposure (ensuring no light hits the sensor). Typically, bias frames "are" required. However, with CMOS cameras, people seem to take "dark-flats" (or flat-darks). You need either bias frames or dark-flats. Dark-flats are basically "darks" that match the exposure time of your flat frames. And they are used to calibrate flats and not light frames. (BTW, if that is confusing, I suggest you ask more questions, be happy to help).

 

2) Dark frames - you do not need 100s of dark frames - about 15 or 20 are good enough. With CMOS cameras, you do need to ensure that you match gain/temperature of the dark frames to your light frames. 

 

Now, with regards to your specific questions about NINA - I don't use NINA, so I can't help you there, but the idea is simple - you take a series of dark frames the same way you would take light frames, just ensure no light hits the sensor. To take dark frames, you don't need the camera to be connected to the scope - you can put it in a "dark box", ensure you cover it so that no light seeps in, and then take your dark frames. Light leak from your darks is a major hassle and will mess up calibration. 

 

I typically create a master dark once every year. Unless the noise profile of your camera changes, there is no real need to do it any more frequently.

 

You can scale darks, but from what I hear, it is a no-go with CMOS cameras (it was done with CCDs, not recommended with CMOS). The noise profile, it seems is not linear for most CMOS cameras and hence the software will not be able to do a good job at scaling the master dark with the light frame. I don't use the 533 camera, so I am not quite sure of its specifications. Hopefully, someone who uses that camera or knows its specs can help you here.

 

CS!

[fewayne]

Welcome Julien!

 

Bias frames don't have much to do with read noise. They are accounting for the minimum current the sensor elements produce even with no signal at all applied. Particularly important for flat-frame calibration; some cameras that don't give very reproducible results at very short exposures (e.g., ASI 1600 series) do better with "flat darks" instead.

 

Don't know NINA, but usually if you select "dark", the control software figures it out. I'd just go ahead and try it.

 

The idea with taking lots of dark frames, as with any frames, is to increase the signal/noise via averaging. More frames means less noise so your dark calibration will be more accurate, but it's a diminishing-returns game, since the SNR improves only as the square root of the number of frames. I do 100 but that's just because it's automated and the computer has nothing better to do when I'm not using it.

 

One good guide is to use master darks until calibration artifacts start to show up, and then 10 days less than that. :-) I just had to redo a set I'd had around for about a year because I got terrible amp-glow artifacts. Redid the darks and the problem went away.

 

Scaling darks: It's possible, but can introduce problems since things aren't precisely linear. Another set of darks is pretty cheap to produce, and you only have to spend disk space on keeping the masters. I consider the best practice to just match whatever gain, temperature, and exposure time used for the lights. This motivates me to limit the number of combinations I shoot, which is also a good thing. I mostly use one combo for RGB and one for narrowband, and not worry too much if the individual lights have absolutely optimal exposure. That'll all come out in the wash of stacking, and keeping it simple reduces the number of errors. 

[jlecomte]

Thanks guys, this is very helpful! I am going to ask the question specifically related to N.I.N.A. in their discord channel to find out how people use that software to do this task. I'm sure there is a simple, albeit not immediately obvious to me solution. Cheers!

[dhaval]

I guess, why do you need to use NINA? Can't you use the software that came with the camera to take darks? Either SharpCap or if this is a QHY camera, then QHY has their own software as well.

 

CS!

[jlecomte]

I already received a response to my inquiry in the N.I.N.A. Discord channel. The sequencer actually does not require a target. Simply click the "+" icon, specify the type of frame as "DARK", set exposure, gain, offset, etc. and start the sequence. Simple enough!