Bad Pixel Mask - combine data with different slopes

[bhilbert4]

In order to maximize the SNR when looking for bad pixels, it would be nice to be able to combine data from different filters, or data with different mean signal rates.

What's the best way to do this? It seems like there are several options:

1. Run the bad pixel mask code separately on data from each filter / data with a given mean slope. Then combine the output bad pixel masks from each of these runs. For example, run the code on all your F356W data. Then again on all F444W data. Then again on all F277W data, etc. Stack together the output bad pixel masks from these runs. Set some threshold: a pixel must be flagged as dead in N% of the output bad pixel masks in order to be dead in the final bad pixel mask.

2. Combine data with different slope values within the bad pixel mask code when creating the mean slope image.

   2a. Can we just normalize each slope image by its median, and then combine all the normalized slope images into a final mean normalized slope image?

   2b. Or should we have the code examine the header of the input data and create a mean slope image for each filter, and then normalize each of these before combining into a final mean slope image? This means you still could not combine data from a single filter that has different signal rates.

@jemorrison what do you think?