Wrong Temperature/Duration Darks

[Mark de Regt]

I'm only now processing a 9-frame mosaic I took back in 2012 and 2013.  For some reason, I don't seem to have taken darks at all the right temperatues for all the right durations.  For instance, I have some images at 15 seconds and 90 seconds duration, taken at -15C, but I don't seem to have matching darks.

I have longer darks for those temperatures taken then, and I have the right duration at -5C and -10C.

Does PI scale darks well if I use a longer dark (with a bias, of course) than the images I'm calibrating, but at the right temperature?

Thanks!

Mark

[pfile]

yes - actually it scales based on noise minimization in the calibrated frame. so yes it should work. just make sure to check "optimize" under the dark filename in ImageCalibration (i think this is the default.)

rob

[Arnie]

Hi Mark,

I am trying to find an answer to the 'dark rescaling' question myself also. I have been making a library of darks and bias and then master frames.

What I found from PI main page was this:

++++

IC is optimized to work with calibration libraries: don't worry about differing temperatures and exposure times between dark and light frames. IC will always rescale the dark noise to match every light frame.

++++

So I assume that it works both ways as it pretty cleary in my opinion says so. Ie if you have shorter or longer they can both be used as PI rescales them. And temperature will not play any role either. So basicly the darks could be taken indoors.

What I did was I made myself a Super Bias out of 200 frames and a Master Dark out of 399 dark frames each being 10 sec exposure. I would like to get someone more experienced to tell me if my assumptions are correct or have I failed understanding the PI rescaling.

There is the topic: 'Dark Frame Optimization Algorithm' here on the forums 'general' area, which I believe has very detailed description of how things are but unfortunately I am not as experienced that I would understand it.

-Arnie

PS. ROB answered while I was typing. Please Rob do you think it works both ways?

[pfile]

well the 'common wisdom' is that dark durations should exceed the light duration when scaling. i have always done it that way (for instance, 1800s darks used to calibrate 1200s and 600s lights) and i've never done the experiment of making a short dark. i'm sure PI would scale up properly though.

i think one place where this is broken is with DSLRs - since canon generally "cooks" the raw files sometimes dark scaling does not work properly. and in fact i think there are threads here where people experimented with normalization of DSLR dark subs while integrating the master dark, and had better results. normally you would not do that when making a master dark. i stopped using a DSLR a long time ago so i'm not up on the state of the art...

rob

[Arnie]

Thanks Rob for answering. Yes I started AP only last winter and using Nikon D7100 as my DSLR.

Gathering information as much as I find and have also recently learned that the darks with CMOS' are not as straight forward as I first thought they would be. Right now when writing this my computer is stacking my last set of light frames with only Super Bias as aid. Going to compare the result with the one that had both Bias and Dark as aid.

-Arnie

[Physicist13]

problem is that the dark noise is function of both duration and temperature..and non-linear to boot! the scaling is doing, essentially, a best fit routine (correct me if wron please Juan!!). The closer in both time and temp the calib dark is the better. PLus there is the general global minimisation problem..but that another matter ;// . Temperature tends to be the main power dependancy so closer to light frame temp the better...duration tends to scale better (more linear) .
Patrick