Quick DSLR calibration questions (BPP)

[steven_usa]

Couple quick questions, I hope... In regards to DSLR image processing:

(1) Does PI (or specifically the BPP script) not use Dark Flats?  Just wanted to make sure I'm not missing something.  I did collect Dark Flats along with my other calibration files over the summer, and I have used them in other stacking workflows (other s/w).  But it seems BPP doesn't need them?  Does this mean it assumes you have "enough" bias files?  or as I learned recently, if you don't have enough bias files (apparently 20 wasn't enough) you can now use the Super Bias tool.


(2) What's are opinions here, with using BPP, on the temperature range on DSLR darks (relative to the Lights)?  I've heard PI can compensate for being off a few degrees, but I don't see anything explicit in BPP on if it is doing this automagically or not.  Clearly we do the best we can to collect enough darks at all the temperatures.  But say I had 8 Lights at 16degC and 300".   At what point do matching 300" darks at the same ISO are deemed to far off in temperature to be suitable?  1deg, 2, 3, 4, 5deg?  Like I've had some at 17deg, a few more at 19degC, and a few more at 20degC.  I know there is no exact answer here for all situations.  So I guess I'm asking two questions: how does the BPP handle darks that are off a few degrees from the Lights?  (especially considering it is creating a master dark first)  What are the general guidelines on BPP for how far off to go in temperature when selecting darks?   1 dark at 1deg off probably isn't as good as 20 darks that are 5deg off, right? (or is neither of these sufficient?)  Is there anything wrong with going with cooler darks, or should we always go with warmer darks? (again, relative to the Lights being stacked at the time)


Thanks,
Steve

[Don]

Hi Steve,

BPP will use dark flats if you provide them.  Just add them to the Darks list - BPP groups the darks by exposure time, using the exposure tolerance you specify.  If you provide dark flats that differ in exposure time from the regular darks by more than the exposure tolerance, BPP will make a separate master dark using the dark flat subframes, and use it for calibrating the flats if the exposure time matches.  In any case BPP does perform dark subtraction on the flat subframes - it will use the regular master dark if there isn't a better exposure time match.

BPP has a checkbox labeled "Optimize dark frames", and ImageIntegration has an "Optimize" checkbox in the Master Dark section.  The purpose of optimizing dark frames is not specifically to compensate for temperature differences but to minimize noise introduced by dark subtraction.  However it will have the effect of matching the noise level in the master dark to the noise level in light subframes acquired at different termperatures, at least up to a point.  Note that hot pixels are scaled too, so dark optimization may result in the hot pixels not being completely removed from the lights (but dithering and/or cosmetic correction should accomplish the removal of remaining hot pixels).

You are likely to get different and perhaps contradictory opinions on temperature matching of darks and lights.  Personally I don't keep a dark library, but acquire new darks during each imaging session and match the ISO and exposure times to the lights.  I do enable dark optimization though, because with an un-cooled DSLR the dark current noise won't be consistent across all frames in a given night, even if the ambient temperature is stable.

Don

[steven_usa]

Ah, thanks    Well, time to try BPP on all my data AGAIN for the dark flats.

I see the "Optimize dark frames" for BPP, I assume for "Optimize" checkbox you meant ImageCalibration.  Looks like mine was checked (is that the default?).  Although, I also I assume BPP is running a variant of ImageCalibration automatically?  Normally after BPP I just do StarAligment and ImageIntegration.

[steven_usa]

While on the subject of BPP... Any guidelines for the sigma low/high on the Dark or Flats fed into BPP?   Should it be the same Sigma low/high used later for the Lights, or are these totally unrelated?   I guess if you manually calibrate the darks, flats, then you have a chance to observe the high/low before proceeding?   But I assume the high 3, low 4 default is generally sufficient (for darks, flats).  For Lights, I've had my High down to 1.1 and Low to 5.0.

[Don]

I see the "Optimize dark frames" for BPP, I assume for "Optimize" checkbox you meant ImageCalibration.  Looks like mine was checked (is that the default?).

Yes and yes.  Sorry for the mis-type.

  Although, I also I assume BPP is running a variant of ImageCalibration automatically?  Normally after BPP I just do StarAligment and ImageIntegration.

Yes, BPP always runs the ImageCalibration process.  It will also run the StarAlignment process if you specify by unchecking the "Calibrate only" box on the Lights tab.  And it will perform an ImageIntegration if you check the Apply box in the Image Integration section after unchecking "Calibrate only".

Any guidelines for the sigma low/high on the Dark or Flats fed into BPP?

You should try different values and see what works best for your data.  It may be difficult to discern any difference in the final result.  When I calibrate manually I use no pixel rejection for the darks and flats, but if you select this option in BPP, an error occurs and stops the script - looks like BPP is trying to close the high and low rejection map image windows, but they don't exist.  So I use the most permissive value (10) for the high and low sigma values for flats and darks in BPP.  This is not based on any empirical data, but just on my belief that there shouldn't be any need for pixel rejection in the darks and flats ... 

Don

[steven_usa]

Thanks again for the tips on BPP!

This is not based on any empirical data, but just on my belief that there shouldn't be any need for pixel rejection in the darks and flats ... 

Interesting idea.  I'm nearly done stacking all the data I have with the default values for high/low on Dark/Flats (and bias) in BPP.  I think I could go back through and try the 10/10 value like you suggest to see if I even notice a difference or not.

While doing ImageIntegration manually, I've been experimenting with Sigma high/low on the Lights.  Since I've gone back and added the Dark Flats, I am seeing less noise in the images (and I judge this by noting I don't have to use as small values for sigma high/low in ImageIntegration).  In fact for some images, I can raise those values on up to say 9 low, 8 high on lights.  The resulting STF image is more "red" cast, but I don't see any obvious noise (e.g. odd dots of red, blue here and there).  But just because I can, I'm not sure if I should... In general, should I raise Sigma low/high (in ImageIntegration) as high as I can get away with?   Or is it better to back it off a bit, maybe sacrificing a little data but ensuring you're not corrupting with too much noise?