At Burningthumb we have made a game that involves flying a plane around a large city which has pretty good performance on an iPhone, however in order to further raise it’s frame rate we have looked into different fulling methods.
Unfortunately however the documentation for Unity 3 has not yet been released with the beta and the benefits of Umbra occlusion have not been made clear.
Please answer the following questions involving culling:
How does Umbra differ from the occlusion culling in Unity iphone 1.7?
Is Umbra occlusion culling beneficial in open spaces, or does camera frustum culling and distance culling do the job?
Can distance culling be reversed such that objects with farther culling ranges are not culled, while the far clip applies normally – instead of the current system where items can be culled at closer distances, even though the far clip distance must still contain them?
Does occlusion culling work with additive loading?
Is there still a limit to marking objects as static, if so could meshes not be combined according to there view cell, rather then into one large mesh, or is that what occlusion does already?
Umbra is similar but more powerfull.
the occlusion culling in iphone 1.7 was position based only, it didn’t care for the orientation for example, while umbra will occlude including your orientation.
also it can do it all in automagic mode.
for open environments occlusion culling is never an option as you lack what it bases on normally: occluders
no occluders = no occlusion culling, independent which system you use.
if the environment is pseudo open, then you can naturally use occlusion culling. pseudo open means city environments, mountain - valley environments, very dense “plane woods” and such things.
but wide flat primarily open environments can and will never work nor need occlusion culling, there frustum culling is all you need.
and objects have no culling range, the camera has culling and the camera only.
I would assume it works with additive loading, yes. At least if you load whole scenes additively because occlusion culling is scene wide, not object based
neither happens, the static objects are combined on the fly to the minimum amount of draw calls. neither “a single massive” nor “per cell” would be a good idea in any way
Additive loading doesn’t work with occlusion, at least in Unity 3 f1. I’m uncertain if this is a limitation of Umbra, or just an oversight on Unity’s part. Additively loaded Occlusion Areas simply aren’t recognized at run-time. Hopefully there’ll be a fix for this - we’re building a massive environment that will need to take advantage of both culling and additive loading.