Hello Stan,

If I take a stack of say 6 images with a FWHM ranging from the following.

2.7
3.2
3.5
3.8
4.3
4.5

do a poission rejection (like I normally do) and a MEAN combine my FWHM ends up at 4.4

If I combine with Median on the same stack and same rejection my FWHM ends up being 3.5

I can blink the two and see the difference. Should I not get an average with Mean??

S/N number for the Mean = 98.07
S/N number for the Medain = 103.95

It is almost like the functions are mixed up?

Ken Crawford

Poisson rejection is complicated for stacks with disparate FWHM because the algorithm can too easily reject the bright star cores from the better resolved images. Likewise the wings of lesser resolved images may get clipped. You can see when these things happen by blinking the rejected stack (while in Data Reject mode) and noting red centers or peripheries of stars. These effects can result in weird PSFs that are not accurately characterized by FWHM.

Median combine of disparate images is less complicated and you are more likely to get a result that is near the middle, which is what you observed.

For stacks with disparate FWHM, Standard Deviation might do a better rejection.

Stan

Thanks Stan,

I have always liked Poisson as it was agressive with small stacks getting ride of my RBI stars.

I wonder if I can get as agressive with SD..

Ken

quote:

I wonder if I can get as agressive with SD.

Standard Deviation is usually not as robust as Poisson, especially for short stacks. One problem with SD is that decreasing the sigma-factor (increase the rejection rate) results in rejections of natural (OK) pixels along with outliers, but that has only a minor effect on the final image.

The real problem with that stack is the presence of some truly awful frames. I would be very reluctant to include frames with less than half the resolution of the better frames. I usually throw away frames with such poor resolution (or don’t even bother to save them from the camera)...

Stan

I agree, I normally throw out the bad ones but I had so few frames this time.

One other quick question... with the Poisson rejection, just because the FWHM is larger does that mean the fine details in galaxies are larger also??

Fine, high contrast features might be softened. Look at the rejection patterns (red tinted pixels) to see what's being rejected.

Setting iterations = 1 can contain that effect.

Stan