[daip] scimg question

[Hanno Spreeuw]

Hello, I have a question about scimg.
The scimg helpfile says that one should first transform the spectral 
line data into a "continuum" data set using split, uvcop, etc.
However, if you run scimg on your line data you seem to end up with a 
datacube which you could then average using sumim.
Is this ok?
In this way one could avoid bandwidth smearing.
I must add that I probably have anough s/n per channel for selfcal (it 
is P-band VLA data centered on SgrA).
Thanks,
Hanno.