Smallish update: I have the texture loader working. No pictures as the materials are still ugly as hell
Reading textures was really slow in the Python version (roughly 1 minute). Now it takes 10 milliseconds for the whole neighborhood
Originally, loading the Neighborhood took 1min 30 seconds in the Python version (Uru can load it in 3 seconds only).
With the new importer, it only takes 16 seconds. 3 of those are required for HSPlasma to read the files, and 13 are spent building meshes. It's still a bit high but it's largely acceptable.
EDIT: nevermind that, some bogus code was slowing down loading for no reason. The importer needs only 1 second on top of HSPlasma's 3 to load the whole neighborhood (mesh, colliders, textures and sounds), for a total of 4 seconds. Nice !
Aloys wrote:In many regards Plasma is a dead-end, even though Cyan open sourced part of it; the community is just too small now.
My opinion exactly. Just let the big game company develop most of your engine for free. This also means you support pretty much every modern platform, including Mac, Linux, any VR headset, whatever crazy peripherals, you name it. All of it while staying backward-compatible with existing tools.
I would really love for this version of the engine to completely replace Plasma at some point, but right now that's too far away to make any prediction.
Aloys wrote:Obviously, getting the whole game to run in Unity would be a pretty scary amount of work
It is, but at the same time Unity does all the complex stuff already. What needs to be done is the wiring between the logic bricks, so to speak. And we have all the sources and the PRP format figured out. I believe it's doable, although it will take time.
Aloys wrote:I'm a bit out of the loop here, but I suppose the server part of the code still hasn't been open sourced, right? Same for the avatars sources?
Dunno. But I don't think I'll need either of those.
Uru's netcode isn't top-notch from what I heard, and reimplementing it would be too cumbersome. I'll just use Unity's networking instead, will be enough to support decently-sized multiplayer coop as a first step. The rest of the server stuff is mimicked by the offline client, so it won't be too hard to reimplement. Would have been sweet connecting to MOULa using a Unity client though...
As for avatar stuff (and "global pages" like UIs), I'll probably completely replace those with something better. Those avatars certainly didn't age well...