First and foremost, congratulations, this is a simply amazing solution that you guys are providing — one assumes that most of the work is being handled by Amazon Luna's own infrastructure, but I have peeked at the JavaScript 'made by LL' and it looks like you really have done a great effort to solve a lot of issues already! That said, I'm aware that my request will be anything but easy to implement. It might even be impossible! Here's the biggest catch with the Zero viewer: whatever viewer preferences are set in our 'main' viewers will be lost on Zero — which is obvious, since all viewer preferences are, well, stored on our viewer installed on our hardware. Zero presents a 'bare bones', pristine environment — how could it be otherwise, since we're basically streaming interactive videos from a (virtual) machine that is powered-up on demand? — which is more than fine for the casual user. But for us tinkerers, well, things are annoying. Buttons and our own shortcuts are, of course, gone; but so are things like camera settings (the most annoying thing for me), or, well, display graphics settings. Like most SL residents who have been around for a while, we profusely hate the weird 'view from above' that is the 'standard' camera position; and we have several settings for different environments — e.g., lowering settings on a region with way too much lag, or going to full Ultra Plus Extra Pro to take some amazing pictures. All of these required tinkering and tweaking over the years. And, since they are mostly configuration settings written to disk, they will survive viewer upgrades, and can even be used on different viewers as well (or copied to different computers, so that you have the exact same viewer experience on all computers you use). Now, since, naturally enough, we will not have any disk-based access to the virtual machine running the remote Zero viewer, but are merely using a single application launched there, there is no way we can upload our favourite configurations to Zero — and that's really, really annoying. I'm also aware that a naïve approach won't work. I thought about two possibilities. None are implementable in this specific scenario, but maybe they're worth being thought about. I'm posting them as comments due to space limitations here.