I've had a few proxy captures on the EVH 5150iii 50w and 6505+ where the gain/eq sound right but its lost a bit of the body, like a 8MP photo vs a 20MP photo or something, hard to explain. I'd A/B the tone and going from the amp to proxy the amp just felt more detailed even though at a glance they sounded the same. For whatever reason that didnt happen here 3x so maybe the JVM is just a good amp for it, no idea.
Was it possible to "add a little body" back with a tweak?
This is 100% the case, and is true for other capture/cloning solutions as well. Where Proxy might shine, others might falter (for example, some fuzzes). And vice versa. The goal is to ensure it shines in as many cases as possible and if/when it doesn't, users will send us UUIDs so we can fully understand why. Another big advantage of Proxy being server-based.
Also, having a bunch of actual amp captures to work with should also provide Line 6 with ways to improve the modeling to more accurately simulate the amp behavior. Having the capture analyzed on the server also provides a butt ton more processing capabilities than could ever be achieved on a local piece of hardware; however, server resources are NOT free. It' really easy to build up an AWS/AZURE bill that completely eats up any profit you make on devices over time.
Then, as you point out, it provides a great ability to figure out why a clone didn't work great .... and fix it.
Gotta say that in general I'm pretty unimpressed by how little CPU juice there is to be had in most hardware modelers. I mean, I can run 3-4 maxed out instances of HX Native on a single thread of a Macbook Air M3. And over the years, companies never got tired of telling us how much of a benefit dedicated CPUs and OSes were. I'm still wondering why that would be. They're not cheaper, not more stable and not performing better.
I don't know. I think if I were coming out of the gate with Line 6, I would fail on the side of effectiveness and worry about efficiency once I got the effectiveness down cold. Seems normal to me.
I'm surprised by how much DSP clones eat up, but then they sound better than anything else (aside from NAM, but it's about the same IMO). Been surprised at how good clones sound and feel. Incredible job by Line 6.
I guess my question would be, if it gets the job done, who cares how much DSP it eats up?
Yes, I can see where those who want multiple amps, and multiple cloned pedals and multiple reverbs all in one, it would be frustrating ..... but I have to ask, are there other options to getting the same thing done without doing it all-in-one?
Kemper manages this by using "performance mode" and allows "spill over" between switches.... assuming that the reason one would put multiple amps and multiple verbs, etc into a single setup was to achieve an instant change between tones live. If it is to run 2 amps in parallel, that is something different.