Drizzling introducing artifacts?

[cmarcus]

Hi Juan
thanks for taking time to look at my data.
I do indeed have Enable Pixel Rejection checked. I have even redone integration using the exact same settings as per your screen shot and I am still getting the artifacts....
At this point could the fault be upstream in the processing - i.e. alignment in this case?

[Juan Conejero]

I don't think so, since all stars appear correctly aligned in your screenshots. If it isn't a mistake in process settings, then I have no idea. Another possibility is that you have selected the wrong .drz files, that is, files that have not been updated by ImageIntegration with rejection data.

[cmarcus]

Hi Juan
Thanks for your reply. I'm pretty sure that is not happening - I am getting the right drz files. I have tried several times over on several sets of data now. Is there any point in you taking a look at my drz files, themselves?
In case it is any help, this issue is happening on two separate machines, with different sets of data. Though it has not always happened, as I have said.
Clifford

[pfile]

i think i found the bug: if there is an ROI defined when ImageIntegration is run, somehow the drizzle files no longer contain the rejection information. Juan, can you try re-running this data set with an ROI defined in II? i think you will see the problem.

rob

[Juan Conejero]

Hi Rob,

Yes, this happens with the current version. This problem has already been solved in version 1.8.5 (drizzle files are not updated with a ROI enabled in II, and a warning message is written to the console). For now, just remember to disable ROI when performing the final integration, so the drizzle files are correctly updated. Your guess is plausible, this would explain Clifford's reported issue.

[cmarcus]

That sounds plausible, as I do do ROI for drizzling.
Will test this out a little later today and report back.
many thanks,
Clifford

[cmarcus]

Many thanks, that has indeed solved the problem
Clifford

[Juan Conejero]

Hi Clifford,

Glad to know it's all sorted out. All the merit goes to Rob for discovering the cause of this. I didn't figure this out, probably because I already have fixed this issue in the version I am working on right now, so it is not in my "stack of pending problems" anymore.

[cmarcus]

Indeed, Kudos to Rob!
Good to be aware of this because doing an ROI seems a logical thing to do when drizzling - to decrease file size and as often as not one is drizzling on a small target so would crop anyway.
Glad to have drizzling back anyway!
Regards
Clifford
PS when refering to 1.8.5 - presumably that is an upcoming version?

[pfile]

thanks guys but it was somewhat random. for some reason i had scheduled a bunch of Ha of the cats' eye nebula with my 500mm setup... so i decided to try to drizzle 4x just to see what it would look like. because the nebula is so tiny, i decided to ROI the original ImageIntegration just to speed things up, and then i saw the hot pixels in the DrizzleIntegration run.

rob