Interesting thread!
For me, top-down works better than bottom-up. Here is what that means:
In a certain sense, I used my existing vanilla synths and fx as a modular system (on a different level) so far. When stacking synths, that´s similar to combining OSCs. Likewise, building processing chains like (Synth A open filter) -> (delay) -> (Synth B audio in) -> (pedals) and so on. I´d call this quarter-modular in contrast to semi-modular. Or quadrant-modular if you want to be a smart ass :D.
Next stop: Semi-modulars. Opening up the box-internal signal chains at some points, granting more access. A good point to get started and learn, as you don´t have to do everything from scratch, but can gradually grow into things. Just inserting a filter or adding that 2nd and 3rd LFO is pretty easy this way. Buying and learning on an "as needed" basis is more fun to me, as I get instant results I can use in my tracks.
And then of course, the next step is full modular, a.k.a. crack effect.
-- Icon_Detach
I like semi modulars and synth voices. I think everyone should have at least one and they don't have to be expensive. When you have some creative inspiration in modulation, not having to take the time to set up a patch for a basic sound is liberating. When you're in the heat of creativity and just want something to throw down a working bassline, arp, etc... I can get it done and then focus on the details. I like VCOs from Noise Engineering and the Braids line of modules because they can stand on their own with just a trigger and a 1V/Octave source... yet still treat them as you would any traditional module and follow them with filters and VCAs for more wiggling fun.