Last month, I stacked Cat's Eye without any RGB channels for stars. It came out fine (I thought it was Beta 22, but may have been Beta 20).
This month, I captured my RGB subs but decided to re-stack everything with Beta 22. The OIII channel which emerged, which was comprised of the same-or-very-similar set of subs as last time, seems over-calibrated. My individual subs show no visible vignetting, but the stacked result shows reverse-vignetting where my corners are bright.
I examined all the individual subs for clouds, dew, light pollution, etc. I tossed none. I varied the number that were "selected" in the stack. I turned LNC off. Nothing seemed to reduce the problem. And of course I'm on a permanent pier in an observatory, so I'm using the same master flats, dark flats and darks as I did last time.
Sorry for this: On a whim, or out of desperation, I decided to try PI's WBPP process (it was painful compared to me being used to APP)... but it stacked the OIII channel just fine, using the same subs.
Any ideas?
PI on the left, APP on the right.
Could you upload the data to our server? I can then have a better look as to what may be happening.
Hi @mountainair,
Perhaps you can share your data so we can have a look ourselves ?
Usually this overcorrection of the vignetting will occur if the lights and flats or flats and darkflats are not fully compatible in terms of the sensor offset. Did you check if the are matching between all data? If you use old masters, then you run the risk of this happening when you shoot the lights possibly with a different sensor offset. If your capture software had an update between shooting the calibration data and the lights, it could have caused this, this happens a lot.
Mabula
@mabula-admin Mystery solved: It was sensor temperature! One night I had some trouble with my imaging suite mid-sequence, so I left the cooler off while I worked on it. I remember now that I had forgotten to re-enable it until partway into the night's sequence. When I exclude those 24°C images and only stack the -15°C images, the vignetting goes away. You can't see vignetting in the subs, but there appears to be far less signal -- the nebulosity around Cat's Eye was much dimmer. Maybe PI excluded this via Subframe Selector. I am now adding sensor temp to my already-very-lengthy filenames.
Suggestion: The sensor temp is in the FITS headers. If there is a large range across the selected subs, perhaps you should call it out via a warning dialog for the user. And/or, you could add a column for sensor temp (but then you should probably have one for offset, too).
/AstroPixelProcessorUpload/MountainAir_VignetteIssue can be deleted now. If you're curious, check out subs 25-30 compared to the others.
Hi @mountainair,
Excellent, glad that you have found the cause of the issue. Yes that is 40 degree temperature difference, for most sensors, each 6 degree change will double the amount of dark current signal, so in this case the dark current in those 24 degree lights is roughly 2^6,7 (40/6 = 6,7) = 100 more ! than in the darks... that meanse that there is additional offset in those lights, and you really need matching darks for those then as well..
Yes, both sensor offset and temperature should be reported in the frame list going forward, I agree. It has been added to our Todo list 😉
Mabula
