Wait, wait... you really can't use a badpixels list as a substitute for a dark frame.
Take a long dark frame and then push the histogram up so that the hot pixels get clipped, and you'll notice that there are patterns in that frame aside from the hot pixels themselves: you'll likely see, as I said, a
blue glow in the top left corner, and also other less obvious artifacts that are at brightness levels close to random noise, but clearly
aren't random noise.
Why can't you use a badpixels list to fix those? Well, two main reasons I can think of:
- both the original badpixels code and mine work, to an extent, by interpolating the bad pixel using the nearest neighbors. But if the neighbors are bad, too, then hmm...
- we could increase the 4096-characters limit to something higher, or even allocate memory dynamically to remove the limit altogether. But still, the bad pixels are stored in a linked list, each item being 96 bytes wide, so your 9634 pixels would take up 900 kilobytes, and that's more memory than my camera
has.
All in all, if you want to obtain the effects of dark frame subtraction without taking a dark frame after each exposure, then just take one dark frame for each shutter speed, and then use those all the time.
It's also theoretically possible to take just one dark frame and
simulate ones for other exposures using a bias frame.
The badpixels list just isn't a viable approach in comparison with the above methods.