I do it this way:
First, I do as much as I can directly in the MIDI instrument tracks before going to the main CuBase Mixer and external plugin effects. When it’s time to mix down, or get fancy with complex effect chains that require extra real time ‘automation’:
Set up a virtual MIDI port (I’m on Windows so I use loopMIDI and name the new virtual port(s) “Automation_x” [with x being a number or letter]).
Build an appropriate Generic Remote Device Map (actually I stack several of them) for the whole DAW.
Set the input of my remote Map(s) to the loopMIDI virtual MIDI port.
Make a MIDI (or multiple tracks if desired) with my MIDI controller(s) as the input, and the output directed to my “Automation” virtual ports.
This setup routes everything through a MIDI track before forwarding it into my Generic Remote map(s). This allows me to easily arm/disarm when and if my remote maps are active. It also allows me to record in real time any and all controller movements I like in a MIDI format.
There are still a number of things I’d rather just use quick controls and VST automation lanes directly; however, there are quite a few things one can automate in CuBase through generic remote maps that do not have native ‘track types’ or ‘automation lanes’ (Launching Macros, stopping the transport, arming/disarming tracks for recording/monitoring, manipulating cycle points, etc). This setup allows me to automate all that good stuff directly from a MIDI track. In short, you can ‘remote control’ nearly every feature, mouse click, or key-combo via assigned MIDI events! By routing it all through MIDI Tracks first, you can ‘record/edit’ all those remote events to your heart’s content.
Other interesting benefits to running controller movements through a MIDI track like this include:
MIDI tracks support multiple track versions, and quite a number of options in automating how one handles ‘cycled’ recording sessions. I.E. On Cycle: Replace/over-write existing part, merge with existing part, start a new part in the same track, start a whole new track, etc. One can also take advantage of introspective MIDI recording. This gives you something kin to unlimited ‘undo-redo’ abilities when recording automation passes live. Make a ton of them, and then mix and match the best of all your automation takes. Introspective MIDI recording is one of my all time favorite CuBase features…so why not take advantage of it when ‘automating the entire DAW’?
MIDI events go into ‘parts’, which are easy to cut, paste, splice, drag around in sections, etc. The track automation lane editors aren’t quite as flexible (though the new 9.5 release does give us some new ‘range’ features).
It’s easy to edit this stuff in the MIDI Key Editor, and you’ll also get the bonus of the Logical MIDI Editor.
It makes it possible to automate things in VST effect plugins that don’t normally have a way to control them via MIDI with your MIDI Track insert effects (I.E. MIDI LFO).
Using track transformers, it’s easy to set up multiple tracks and arm/disarm remotes as needed (with record and monitor buttons).
When it’s time to freeze things into the VST automation lanes (where applicable), I can convert various CC lanes in the KeyEditor into track automation lanes, and from there they are easy to ‘cut and paste’ right onto VST/i automation lanes. Or, I can just activate write for the relevant VST/i automation lanes, play a pass, and it’s ‘frozen in’ as true VST/i meta events. Last, I disable my original MIDI automation tracks, turn off the write switches, turn on the read switches, stash the MIDI automation tracks in an out of the way folder of the project, and make a back-up copy of my remote maps with readme files in the project directory case I ever need them again in the distant future. At that point I’m ready to take care of any ultra fine edits/bumps on the time-line that might be necessary (due to latency on my system when trying to mix with remote controllers).
Here are some threads with posts I’ve made over the years on the technique: