Offical crap load of questions thread

[Siliconmaster]

Could you clarify this? Do you mean that lightmap UV's scale themselves to better fit the BSP UV's?

Well, if you're redoing the lightmap uvs, then it is my understanding that you also must be doing custom lightmaps in 3ds max, since running radiosity in sapien resets all lightmap uvs upon starting. Therefore, while shrinking a section of uvs will lose detail, 3ds max will still rerender it at the new size on the bitmap- therefore, keeping the sizes of the uvs that sapien made is not necessary at all, and those sizes are usually pretty screwy to begin with.

A better-worded version of my original sentence would be more like :
The thing to remember is that when 3ds max renders out new uvs, they match whatever you set your lightmap uvs to, so conserving space by way of shape is not as important as conserving by size.

Shrinking the cliffs down would limit the amount of detail that can be in its UV space. I assume you just mean that every poly in a single lightmap mesh should fit within the 0-1 UV space without overlapping?

Yeah, basically. Then just up the size of the overall map to something far higher than 32x32 or 64x64. I had a few 1024x1024 maps in Precipice.

So basically, seams do matter for lightmaps. The method you just described is what I thought would leave a lot of wasted UV space, but from what you are saying it seems like the benefits would outweigh the cost. Jesus though, I'm not looking forward to this.

It sucks, but it's worth it in the long run. Just make sure not to modify your bsp after you start, or it'll all go to hell.

I do have one more question about lightmap meshes though. How exactly does Sapien decide how to separate the BSP into sections? Some of my lightmap models cover large areas of the map, some cover very small areas, and one only covers a single rope (where the other rope belongs to a larger lightmap mesh). Is there any logic to this? It seems pretty wasteful and disjointed.

Sapien is retarded when it comes to doing that. My guess is that it tries to do it by portal areas- if you have a single room portalled off, it's likely you'll get most of those faces on a single lightmap uv- not guaranteed at all though. When doing Precipice's lightmap uvs, I found that one side would be perfectly split up while another would have, say, every other vertical surface on one map and every third horizontal on another. It's totally bizarre, and unfortunately there's not much you can do about it.