The issue is not so much gradient but rather lack of contrast. A moon free night with very good atmospheric transparency compared to a 25% growing moon with average transparency. Don't get me wrong there's just lower signal-to-noise data in there. So we stack to increase SNR, does that mean an hour of perfect conditions is the same as 2 hours in slightly non-perfect conditions?
The results worked well. Here's the final:
http://www.iceinspace.com.au/forum/s...d.php?t=109766
Attached is a screenshot of two pictures. No DBE applied, just a simple and identical histogram curve and some debeyering. Now obviously the data on the left 5 min exposure is better than the data on the right one. But the right one is still fine.
In this case I had about 1 hour worth of excellent data, and 3 hours worth of ok data. The stack of only the excellent data did not have the same quality as the stack of all 4 hours.
But obviously there is a point where this no longer applies. If I take data in a full moon during a dust storm I imagine the results would be worse when combined with good data then just having the good data alone.