I started with R*S modules (not a full panel), and recently switched almost entirely to LW (one voice panel based on Sing to Me; one sequencer). I still have two R*S modules in a boat with two open spots, and I suspect I will keep this around to fill in gaps in the LW panels as I discover them.
I would recommend starting with R*S modules or panels. As I and others have mentioned elsewhere in the thread, the LW panels can be really powerful, but I would argue that they are not as ergonomic, and as one is getting to know the format, the constraint offered by R*S is possibly valuable (ymmv).
As far as differences, one thing I’ve found myself missing is the DUSG versus the dual transient generator. I don’t think the latter re-trigs without “spending” both envelopes. (I could be super wrong about this, as it wouldn’t be the first time that there was functionality hidden in plain sight. Happy for someone to tell me I’m wrong!)
One other thing I’ll put out there for you or anyone else considering a Sing to Me or some variation on it: I sure wish mine had one more Mix Pro. While the one is sufficient if you want to combine both oscillators into a single mono voice, I’ve found that a single oscillator combined with its output from the waveshaper module (and some noise) can work well as a mono voice, which means that you can effectively get two mono voices out of this*. It’s hard without some mix control dedicated per oscillator, though**.
*= Sing to Me has a single stereo out with also support the idea that it’s a single mono voice, but I just run outs from the VCA myself, as my specific Sing to Me variation does not have the stereo out module
**= What I’m doing is using the two inputs on the VQVCF, but for me it’s not ideal, since it ties noise volume to the volume of another of the oscillator inputs, which isn’t always what I want. May work fine for you.
Anyway, just thought some of what I’ve learned from recently getting to know my Sing to Me-ish might help someone!