So, Trepan... I may have something worth using it for, finally.
I have objects in the 2K+ tricount range, that I'd like to reduce to, say, about 500 at a distance. They have 80+ Pieces that need to be tracked accurately, for position and rotation.
I was thinking about doing this with a global signal via LuaCOB, that would tell the objects to show / hide a bunch of the less-important details if the players' cameras were all > somenumber Y, but I thought I'd investigate this option as well. What's more efficient? And, if I do this through Lua, where's some source to look at? Again, we're not talking a fast-swap of one giant mesh, but a whole bunch of sub-operations- it's going to involve a fairly substantial amount of code. Is there any elegant way to handle this with sufficient speed? Or do we need some third solution, like embedded LODs within something like S3O, before this is more than a pipe dream?
About the LOD Gadget.
Moderator: Moderators
Re: About the LOD Gadget.
This sounds like a really bad idea since if you think about it, it's unsynced to synced communication needed, in realtime.
Re: About the LOD Gadget.
That's what I was figuring, too. Oh well, I've gotten around most of the performance problems with a hack-around for now.
Re: About the LOD Gadget.
The solution is already coded into the engine.
I've only released one example file, limited audience.
I plan on getting around to making a better example
file and maybe a conversion tool at some point. But
given the amount of support I got from the 'community'
when I asked for decimated models, it may be a while
before I get back to it.
I've only released one example file, limited audience.
I plan on getting around to making a better example
file and maybe a conversion tool at some point. But
given the amount of support I got from the 'community'
when I asked for decimated models, it may be a while
before I get back to it.
Re: About the LOD Gadget.
Well, I finally have a practical use for this, cutting visible tricount for the World Builder, that would make use of Spring's typical LOD very effectively. Instead of a model at 10K, 5K, 2.5K, 1.25K, I'm talking 2.5K --> 500. One step LOD, not 4+.
I see no real reason to go higher, and absolutely no reason to spend the time needed to produce further LODs, until I have seen what your LOD code can do, and have tested it. LODs are a pain, man. I do not want to waste a week for something that can be done in a simple way.
All of this is ultimately about FPS. The World Builder demo is now at the point where I can get fairly good FPS at all times, I want to improve the quality of detail and keep the performance.
I see no real reason to go higher, and absolutely no reason to spend the time needed to produce further LODs, until I have seen what your LOD code can do, and have tested it. LODs are a pain, man. I do not want to waste a week for something that can be done in a simple way.
All of this is ultimately about FPS. The World Builder demo is now at the point where I can get fairly good FPS at all times, I want to improve the quality of detail and keep the performance.