Understand "DARKFLAT" image type in WBPP/BPP

dghent

Member
Feb 12, 2019
16
3
daleghent.com
I just stumbled in drenched from power washing my patio all day and saw this. This is amazing, Juan, and all the flavors, too! Thank you for considering this and it should make lives easier. I should hope that other acquisition software that offers a flats wizard might pick up on this.
 

ngc1535

Well-known member
Feb 1, 2014
467
45
Robxy, Bernd and others have heard my take on this. Just in case others are interested I summarize a bit of it here.

Biases and Darks are named not in terms of how they are acquired but instead by what they measure. Dark frames measure dark current. They measure a signal that changes linearly in time and can be scaled. They are not "dark" because of the lack of light or due to length of exposure. Biases measure a per pixel offset. This is an electronic signature that does not have dark current. As described elsewhere, the bias can have multiple components of in terms of a fixed bulk offset, a pixel specific offset (fixed pattern), and I am including reproducible offsets caused by amp glows and other electronic behavior.

If you can see your way through to the above- where dark frames measure a dark current and biases are an electronic signature- then it follows that "flat darks" , as is being used by most, are not darks at all. There are not taken to measure a dark current- therefore they are not dark frames. They are created to measure an electronic signature. The fact that these purported "flat darks" have variable exposure times is of no consequence. All that matters is what they measure.

Now some would complain that having multiple biases with variable times doesn't make sense. But this is already the case for dark frames. Dark current is not linear with changes in temperature. So you have multiple darks at the same exposure time but different temperatures that must be matched to data. In the same way, you would have multiple biases at different exposure times that also would have to be matched to data (for certain sensors).

This does not mean that "flat darks" do not exist. They do! Flats for which the exposure time is long enough to generate significant dark current require a dark frame that also has the dark current. But again, if the "flat dark" has no dark current, (and was not being measured or used in that way) it belongs in the bias column or set of things.

Dark frame scaling is the second use of biases. Under the definition above a dark frame generated from a camera that has variable biases (that is variable/non-linear electronic signals with time) would not be scaling their darks from a single bias frame. They would be generating unique darks that include the unique bias/electronic signature that it contains. (This is already done with CCD sensors that do have variable amp glow, typically time matched darks are taken). Or, even worse, it would be necessary to capture those unique electronic signatures in order to subtract from the dark frame and scale the dark current.

I may have not everything nailed down in the definition- but I think it approaches a consistent way to refer to the frame type without adding a new name. With regards to flats, both biases and darks with the bias are subtracted from flats. So there isn't a difference in terms of math operation. The difference in label comes from what they measure. If a dark has no dark current- it is simply the electronic signature of the senor, it is a bias.

So this may be one of those times where I am crushed because I missed some crucial element, and so this could be a real learning experience for me. This definition, I think, maintains the status quo in terms of image type.

-adam

P.S. One more way to think about a "flat dark." Darks in general are scalable creatures once you subtract the bias (electronic signature). "Flat Darks" as being used with CMOS cameras are not scaled ever- because as you know, there isn't anything to scale... it basically is a "bias."
 
Last edited: