New memory and sequence filtering management
Improved stacking memory management (#331 (closed)), started but postponed the finalization of caching the up-scaled sequence on stacking (#333), sequence filtering for stacking with arguments for the stack
command (#275 (closed)), multiple sequence filtering for stacking from command or GUI (#261 (closed)), stacking result file name change (#345 (closed)) and new command setmem (#340 (closed)). Fixed shifts and reference image in up-scale stacking.
Remaining work:
-
stacking produces completely wrong images when used from the command (seen withFIXEDstack pp_light_ rej 3 3 -norm=addscale -filter-fwhm=70%
) -
stacking produces an image with a black band when rejection stacking is used on an up-scaled sequenceFIXED -
caching the up-scaled sequence does not remove files that are excluded, so it always builds a seqfile with all available files. Two possible solutions: delete the excluded files or create the seqfile manually.postponed, not working, removed from release -
caching the up-scaled sequence is controlled by a switch in the settings which is not yet exposed in the settings GUI nor configuration file.postponed, not working, removed from release -
#331 (closed) depends on #337 (closed) which is not yet doneDONE (#337 (closed) is actually done in this branch only for the resize wrapper function, and it is working) -
#261 (closed) GUI does not display when data is not available, and the new version doesn't have the minimum value displayed any more (like 'max accepted FWHM: 4.72'), only the number of images retained. Not sure if it's useful.DONE -
Display the information that the reference image is not in the retained images for stacking, or make sure it's not a problem.DONE the message is red in the console. Normalization is skipped when the reference image is not stacked. -
Update stackall command to match stack new argumentsDONE -
Verify that sequence export has not been broken, with or without filtersDONE, export is working fine -
Shifts were not properly rounded in up-scaled stacking. It was fixed in rejection stacking, but not everywhere else. Maybe we should just round the shifts in the up-scale rather that at use, I don't remember why we didn't do that.REVIEWED, working elsewhere and unchanged -
Look at the changes made in this branch (Changes tab below), probably some temporary code remains somewhere and is not to be merged.DONE, discussion below pending
Edited by Vincent Hourdin