Pixel math problem

cmarcus

Well-known member
I am trying to combine two images using PM to add a comet to star field (having followed the procedure for comet processing as per Keller...).
I am quite happy with the comet and star pictures per se, but when I combine them I lose a lot of the stars.
I have used the formula C + S (ie just adding the two files - C= comet, S= stars) I have also tried C + (2*S) to see if the latter would give me more stars, but without success.
Attached are screenshots. "masters" shows the comet and the star field to be combined. "combined" is the result, which seems to have killed off a lot of the stars.
TIA for any input

Attachments

• 72.2 KB Views: 15
• 117.7 KB Views: 16

Juan Conejero

PTeam Member
As a first approximation, try with a maximum operation:

max( C, S )

cmarcus

Well-known member
You mean entering that as the PM expression?
It doesn't change much, still not many stars.
I have tried C + (6*S) and that has given a fairly good result...

Juan Conejero

PTeam Member
That happens because both images, C and S, are not well mutually adapted in the statistical sense. Try the following expression:

C - med( C ) + S

This expression will effectively replace the mean background level of the comet image with the mean background level of the stars image. Naturally, the other option is:

S - med( S ) + C

which will preserve the mean background level in the comet image.

cmarcus

Well-known member
Hi, That's still not giving me many stars. Again adding multiples of S does improve the problem.

Many thanks again.
C

fredvanner

Member
I'm a complete beginner, so I may be missing something, but...
In the comet integration (Keller step 3) the noise background will be integrated (with whatever ImageIntegration Combination and Normalisation settings are selected). This averaged background will no longer have any correlation with the individual star images. When this is subtracted from the aligned star images (step 4) it is quite likely that stars with brightness comparable to the noise background (which would otherwise be revealed by the stacked integration) will simply be subtracted out. I suggest the solution would be to use a mask to zero out the background of the integrated comet master before step 4.