question and suggestion for the batch preprocessing script

[whwang]

Hi,

I have a question and a suggestion for BPP.

Question: do the master dark and master flat generated by BPP contain bias?  Or is bias subtracted already?

Suggestion: On color CFA images, I recently found that the resultant image color depends on the color of the flat field image.  This can be quite annoying.  For example, even if the light frames were identical and the WB setting does not change, using twilight flats or flats with an LED panel will produce calibrated light frames of different colors, because the flat images have different colors.  I suggest to first normalize the intensity of RGB in the flat, and then do the flat field correction.  This way, the image color will be always the same as what the WB intends to produce.

Cheers,
Wei-Hao

[pfile]

the flats are bias subtracted, the darks are not.

i assume it was done like this to support DSLRs where your flats may have had a different ISO than the lights. therefore the master flats are "self-contained" and do not depend on the calibration frames for the lights. of course you would have had to have made your flats with a separate run of BPP if they were at a different ISO.

i agree about the flats, but i think what you're going to hear is that since you have to do color calibration anyway, the color cast in the calibrated light does not matter. however for some specialty work or non-astro work it would indeed be very nice to preserve the original WB.

rob

[whwang]

Hi Rob,

Thank you for telling me about the master flat.  It is useful to know.

It is true that we almost always need to re-adjust the color of a calibrated/stacked image.  This is why I had never paid attention to the color of flat.  I only noticed this issue recently, when I need to further stack two groups of stacked images taken in different years.  Because my flat changes, the two stacked images have dramatically different colors.  The very different brightness of each of the RGB layers from different years means that when they are stacked, they will be given artificially wrong weighting.  I don't quite like this.  Of course, there are ways to work around this problem.  But if the flat has a neutral color, then this problem will be gone.

Another reason to preserve the original WB is for non-deep-sky photos, as you pointed out.  Last month, I used PI to calibrate my images of solar corona.  However, because the flat is not white, there is a color shift after the images are calibrated.  I ended up with using Photoshop's raw conversion to restore the WB of my PI-calibrated images. It will be much nicer if the PI-calibrated image can have the right color right out of the box.  I am not sure re-normalizing the RGB brightness in the flat can solve this though.

Cheers,
Wei-Hao

[pfile]

in my case i let dcraw debayer my eclipse reference image with camera white balance, and then used linear fit to fit the stacked master to the properly white-balanced image. it seems to have worked OK.

rob