ft converts a source model or a components list into model visibilities that is inserted into the MODEL_DATA column of the MS. Alternatively (usescratch=False) it can be stored in the header of the MS to be served on the fly when requested. The function of ft is needed to use more complicated sources, or to attach models to be subtracted later from the visibilities with uvsub

The source model should be in units of Jy/pixel, usually a *.model image that tclean saves as a result of deconvolution. 

ft only uses a standard gridder (a prolate spheroidal) but supports multi-term images (see tclean help). Other gridders, like mosaic or widefield, are currently not supported. If you need one of the non-standard gridders, attach the model to the MS with the task tclean, setting savemodel=True, and gridder to the requested algorithm. 

The default is to replace any current MODEL in the MS by the model specified in the model or complist parameters. The parameter incremental=True will add the specified model to the current MODEL in the MS. This can be useful, e.g., when new point sources are identified that need to be added to the model in the MS. 

When both, a model and a component list (in complist) are specified, ft will only use the model image in the model parameter.

ft is strict in its frequency interpretation. Only frequencies that the model covers are inserted into the MS. If the model has a different frequency range, it needs to be modified to span the MS spws that are to be associated. Otherwise the MODEL will be filled with zeros for the frequencies that the model does not specify.  

In addition to tclean, another alternative to ft is setjy, which attaches a model and also scales the flux of the model to a flux value calculated from a standard or to a provided value. 

NOTE: Model images for standard VLA calibrators 3C48, 3C138, 3C147, and 3C286 for the VLA bands are provided with the release. The location inside the package is dependent on the operating system; setjy(listmodels=True) will show all available models.