How can I treat illuminations independently on Zeiss Z.1 LS

[max-brambach]

How can I treat illuminations independently on Zeiss Z.1 LS

[m-albert]

Hi @max-brambach,

as discussed elsewhere, this section of the configuration in the template script can be used to specify how to treat the illuminations and views.

MVRegFus/bin/mvregfus_run.py

Lines 45 to 78 in 53b959c

	# list of pairwise view indices to perform registration on
	# registration_pairs = [[0,1]]
	registration_pairs = None
	registration_pairss = [registration_pairs] *len(filepaths)
	# optionally, specify the meanings of indices
	# occuring in the list of pairs
	# this can be used to fuse illuminations independently
	# to do so, use view_dict, which is a dictionary containing
	# the indices as keys and dictionaries as items, such as:
	# >>> view_dict[0] = {'view': 0 # view 0 within the file
	# 'ill' : 1 # illumination 1 within that view
	# }
	# another example:
	# >>> view_dict[0] = {'view': 0 # view 0 within the file
	# 'ill' : None # like this, both ills of this view are fused
	# }
	# observation: - illumination 0 comes from left
	# - illumination 1 comes from right
	# - rotating in positive direction (in angles)
	# brings left to the front
	# so it makes sense to register like this: (view, ill)
	# (0,1),(0,0)
	# (0,0),(1,1)
	# (1,1),(1,0)
	# (1,0),(2,1)
	# etc.
	# four view example:
	# view_dict = {i:{'view':i,'ill':None} for i in [0,1,4,6,7]}
	# if ills of all views should be averaged, set view_dict to None:
	view_dict = None

In my experience, treating illuminations as separate views only slightly improves the image quality of the output, and fusing them into single views (i.e. without specifying view_dict or 'ill' = None) before multi-view fusion is much more performant when dealing with many views.