Hey Folks!
We're working on using Spine on our newest game and was wondering if we could get a bit of guidance on some of the performance issues we're running into.
We're building a real time strategy RPG where players will battle large armies of heroes that are built and rendered in spine.
Our heroes each have 30 bones and a dozen or so different animations - our goal is to be able to render a 25v25 (50 heroes total) on screen on an ipad mini gen 1 at 30 fps. Our animations are generally linear transforms, and we are not doing any FFD. We're currently able to only get to about half of that - 15v15 before things start to tip. The spine runtime is the biggest culprit here - which is to be expected as its really the bulk of the scene complexity.
https://www.dropbox.com/s/r5xomqwssgkb5to/Screenshot%202015-06-02%2009.26.32.png?dl=0
We're on a pretty old branch of the tk2d based runtime at this point - so upgrading to the latest and cutting over to Mecanim would be a fair amount of work, wondering if I could get some qualitative feedback from Mitch or other folks who have moved over to baked clips - did you see a perf benefit? Am I barking up the right tree here - or should I try and triage our performance in the Spine runtime first?
Went ahead and started kicking the tires on what this upgrade would look like/do.
Captured on a higher end mac
- On our old (outdated) runtime - 60 heroes @ 30 fps
- Latest spine runtime - 140 heroes @ 30 fps
- Latest spine runtime w/ mecanim baking - 400 heroes @ 30 fps
so definitely seems like it may be worth it
