Time to settle the score regarding BIAS VS. DARK FLATS!

[Steve Coates]

I've been using DeepSkyStacker since I've started this about 2 years ago. I've learned how to acquire Lights, Darks, Flats, Bias and Dark Flats. I found a site, Jim Solomon's Astrophotography Cookbook, which explains:

Signal=Light - (Dark + Offset)/Flat Light - (Flat Dark + Flat Offset)

Here, "Dark" refers to the thermal noise signal of the imaging camera; i.e., the noise signal that varies in proportion to temperature, ISO, and exposure length. Note, however, that any exposure we take with a digital camera contains the Offset, and "Darks" are no exception. So, if we define Dark' to be an exposure of some length with the body cap in place, then Dark' = Dark + Offset, and, similarly, Flat Dark' = Flat Dark + Offset. Plugging these values into Equation 4 yields the following simplified form:

Signal = Light - Dark'/Flat Light - Flat Dark'

And just to make things even simpler, let's drop the prime indicators (the apostrophes) that we stuck on "Dark" and "Flat Dark", and just remember that by "Dark" and "Flat Dark" we mean frames captured with the body cap in place but with the same ISO and exposure length as the Lights and Flat Lights, respectively.

Based on this I have only been using Light, Dark, Flats and Dark Flats. Does anyone have any further insight on this? Should we be using Bias only, Bias AND Dark Flats or Dark Flats only? The DSS tutorial suggest that we should be using Bias frames?
I'd like to know what you all are using?
I hope this stimulates some conversation on this topic and most importantly steer us all in the direction of maximizing our efforts.

[Steve Coates]

Hey Hopp! I know that Bias frames are easy. My question is based from the Jim Solomon web site in that, are they needed? I'm starting to second guess my ways in that I should have been taking Bias/Offset all along in addition to Dark Flat frames. I have had a difficult time finding an answer to this. I'm not a math student but I find his explanation reasonable and I have used it all along. This question was posted on the Cloudy Nights site and the answers suggested to use Bias/Offsets. So this is a little confusing to me...

[Steve Coates]

I may just start to use both bias and dark flats in my images. I was told once that doing so was overkill, which got me reading and now second guessing myself. I don't have an issue using both dark flats and bias, so long as it does not degrade the images. I'd like to know what everyone else is doing?

[Marc Basti]

I'm glad this discussion came up, it's timely for me. Up until recently, I was going w/the premise Steve mentioned, bias+flat darks are overkill and I just used bias frames. W/my dslr@ 1/whatever sec. it not going to generate any noise to speak of, just the camera noise-bias. The same w/my ccd camera, even though I'm shooting 2-4sec. exposures for flats (longer exposure to avoid shutter blur/shadow) the noise would be negligible and just the bias should suffice. However, recently I got a Flatman and I was using it to shoot flats w/my 7nm Ha filter and to get to around 25K adu's I needed to expose about 25 secs. Previously (pre-Flatman) I would just crank up my light source (had it on a dimmer switch) to stay within the 2-4sec. range, but the Flatman only goes so bright. My point is now I'm dealing w/25sec exps. which could (I guess) have exposure noise in them. So for now I'm just taking more exposures (normally 10exps up to 25 w/Ha filt.) and count (hope) on the s/n ratio to clean up any x-tra noise. I'm thinking I might be a candidate to shoot flat darks ?? Marc

[Steve Coates]

I was thinking, based from the original post, that Bias frames are overkill, and I have only been using Dark Flats. Let's break down the definitions:

The Dark Frames are used to remove the dark signal from the Light frames (or the Flat frames for the Dark Flat frames).http://deepskystacker.free.fr/english/faq.htm

So as we all know we remove the dark signal from the Light frames by taking Dark Frames of the same exposure time, temperature and ISO as the Light frames (only with the cap on). Am I reading the part in parenthesis as the Dark Flat frames are used to remove dark signal from the Flat frames? This leads to the question, based from Jim Solomons Cookbook, that Offsets are not needed because they are already contained in the Dark Frames that we take.

Note, however, that any exposure we take with a digital camera contains the Offset, and "Darks" are no exception.http://www.saratogaskies.com/articles/cookbook/index.html#Background

Bias/Offests defined:

The Bias/Offset Frames are used to remove the CCD or CMOS chip readout signal from the light frames. Each CCD or CMOS chip is generating a readout signal which is a signal created by the electronic just by reading the content of the chip.

http://deepskystacker.free.fr/english/faq.htm
So Marc, based on the above you are correct that when you take your Flat frames with a longer exposure time you now need to take Dark Flats to remove the dark signal from your Flat frames.

[Marc Basti]

Steve agreed, my problem (I use CCDStack) is I don't see a spot to plug that (a stacked flat dark frame) into the calibration flow. I might have to trick it and put into the bias spot (thinking out loud), like I said my problem.--- I looked in DSS under the calibration process (flow chart) seems like DSS doesn't care if you have one (bias) or the other (dark flat) or both, as long as you have at least one. I guess it's whatever works best for you. One thing I did, like Bob's club member, was to redo a bias stack a couple times a year. First we have big temp. swings (you know) up here between wint.+sum. and second (from what I had read) just wear from the use of your cameras electronics will alter the bias. But if you're using dark flats, you got that covered. Marc

[Steve Coates]

Ok pretty good stuff. This was posted on the CN website and was explained that it depends on the stacking program used. Apparently there are programs like PixInsight that do Dark Frame scaling. So Un-scaled Dark Frames will have the Bias/Offest contained within it. Here is the link.

[Greg Marshall]

I think you should always use bias (offset) frames. The question is whether you need "flat dark" frames. The answer is that it depends on the camera and exposure duration. I like to take flats with a bright source to keep the exposure down to a few seconds. I also (usually) do it with the camera cooled. Under these conditions I don't use "flat dark" frames and don't think they would help significantly.

[Steve Coates]

What stacking program do you use? From what I've read (and linked) in the above post, it is only neccessary to use Offsets/Bias when using scaled Darks. I don't scale my images so the Offset/Bias are already comtined within the Dark Frames. But, if you DO scale then you would need to add Bias/Offsets.

[Greg Marshall]

I use MaxIm DL for calibration and stacking. I believe it supports scaled darks, but I always use exposure times and temperatures for which I have darks in my library.

And I don't think that use of bias has anything to do with scaling. Note that the originally quoted formula does not "add" the offset, it subtracts it (along with the dark) from the light. I'm not sure that the given formula is entirely correct. I've been up all night, so I don't trust my brain now, but I'll post again when I've done some more research.

[Greg Marshall]

OK, here goes: Bias (or offset) is simply a value that is added to every pixel in every exposure. You can think of it as an offset that is intentionally applied to ensure that the "black" level is always above zero despite random variations. This bias value does not vary with the exposure duration. In fact, in a well-behaved camera the bias is so constant that you can simply use a constant for the bias of each pixel. This can actually improve image quality by eliminating the read noise of the bias frames.

I was wrong about bias and scaling. Since bias is present in every frame, subtracting a dark frame from a light frame removes both the dark current and the bias. However, if the dark frame is going to be scaled you can't apply the scaling factor to the "raw" dark frame because you would then be scaling the bias as well. So the bias must be removed before scaling, but it still needs to be removed from the light frames as well.

Ignoring flat frames for the moment, the simple (non-scaled) case is just

CalFrame = LightFrame - DarkFrame

The bias subtraction happens together with the dark current subtraction. Scaling the dark frame can be done in several different ways, but is easily understood from this formula:

CalFrame = (LightFrame - BiasFrame) - ((DarkFrame - BiasFrame) * ScaleFactor)

That is, both bias and dark current are subtracted from the light frame, but they are done separately so that the dark frame can be scaled.

To do flat calibration the frames calculated above are divided by the flat frame, which must first be calibrated in the same way to remove bias and dark current. However, with a short exposure (and good camera) it would not be necessary to subtract the dark current, but only bias. So in the simplest calibration model we do not scale the dark and do not dark subtract the flat:

CalFrame = (LightFrame - DarkFrame) / (FlatFrame - BiasFrame)

Note that we still need the bias in this case! If the dark must be scaled the equation is

CalFrame = ((LightFrame - BiasFrame) - ((DarkFrame - BiasFrame) * ScaleFactor)) / (FlatFrame - BiasFrame)

I think you can see how the other cases would work.

The only case that does not need bias frames at all is when there is no dark scaling and both light and flat frames are dark subtracted. I don't like dark scaling because it relies on the "linearity" of dark current over temperature and time. I prefer to restrict myself to 2 temperatures (the colder one whenever possible and the other when it's too warm out) and a handful of exposure times. My library of calibration frames includes all combinations of these temperatures and exposures. Regarding flats, I prefer to keep the exposures short (and shoot them at low temperature) to avoid the need for flat dark frames. So I always have a bias frame in the library and make it a good one (average of many many frames) and re-do it every few months.

[Steve Coates]

The only case that does not need bias frames at all is when there is no dark scaling and both light and flat frames are dark subtracted.

Agreed! Therefore since I use DSS with NO scaling I just use Flat, Dark Flat, Lights and Darks. I've got a brain cramp now. Thanks for contributing to this I know it is helping me further understand this concept.