hi,
im already in a discussion with soem friend.
topic: lightmapscale & value of 0.1
does it influence only the lenght of the compile process or does it influence also the level performance ?
gerbils opinion:
I’m afraid they are wrong.
Firstly, having 80+ lightmaps eats up LOTS of VRAM - anyone with smaller amounts of VRAM will chug into tons of problems as the lightmaps get swapped in and out of VRAM each and every frame, this will most likely give people <10 fps on old cards. Even on new cards, with every lightmap in memory, it increases both the CPU and rendering overhead because polygons are processed in batches according to their texture + lightmaps. With (many) more lightmaps, you’ll end up with lots more permutations of textures + lightmaps, involving vastly more state swapping while rendering, and using much more graphics card bandwidth.
Let me put it like this - on my relatively powerful card (ti4200 x8 agp with 256 megs of VRAM) I went from <40 fps to >80 fps. You can’t really argue with that You may not be seeing the performance difference yourself because some other part of the map (maybe CPU usage processing the geometry) is the bottleneck, but it certainly will affect plenty of people, some to the point of not really being able to run your map
yndar: now its on yours :moo:
ciao
fidel

