Recently I’ve been working on a map that has quite a bit of complex brush based geometry. I’ve used brushes to create a few large complex objects. Now of course In order to make vvis run I need to do something with them; normally I just func_detail them, however I’ve also been considering turning them into models with proper and just bspziping them in to the map afterwards.
My first question is: which is more efficient: detailing complex geometry or converting it to models using proper?
My real concern here is the vrad process. We all know that detailing a brush makes the vvis ignore it but the vrad still has to deal with it. I’d really like to have hdr in my level however it keeps crashing before it can complete the hdr lighting. Now I’ve read several threads asking about this and no one seems to be able to come up with a solution.
My second question: which would the vrad process find it easier to calculate light around: complex detailed geometry or a static prop model?
You could save brush count but work up ent count with models- and they'd work about the same.
I think turning them into models would be better because you could do what I do, make a [B]huge[/B] detailed model, then size it down to my preferred size (off grid) and propper it.
If you're only doing it because VRAD is crashing, then optimise your lightmaps instead.
Yeah I’ll have to deal with the lightmaps eventually; hopefully I can get them set appropriately, I’ll just have read up a bit more on it, (all I’ve ever read about it was the 10 sentences on the subject in the Half-Life 2 Map Editing Optimization Guide).
In terms of general performance I suspect that models would be preferable since I don’t have any static props in my level as it is (save for the ones I’m proposing to convert).
Ultimately I’m going to conduct an experiment where I compile one version of the map with the objects detailed and then compile an identical version of the map with the same objects as models and record the difference in compile times and in in-game performance.
I was all set up do perform the experiment last night but then my video card died so it will be at least the end of the week before I can replace it.
Thanks for the help.
Sorry, you need to Log In to post a reply to this thread.
