I have started to make some basic game modular assets to try and get my head around the concept. So when it comes to UDK it turns out each model needs a second ID to be used for Light Maps. I was wondering when it comes to set pieces that can be snapped together does each light map need to be similar to work correctly?
Keep in mind I have barely used Light Maps before this. Also I was wondering when it comes to engines like UDK and Unity are Light Maps required? Because I noticed in the Crytek engine it advertises dynamic lighting and natural lighting, does that mean games made in that engine do not use Light Maps?
Replies
http://www.3dmotive.com/training/free/creating-a-lightmap-uv-channel/?follow=true
It`s the same, a lightmap is lighting baked in to a texture, you culd bake lighting in to you difusse texture too, with an external program or in photoshop, but most of the times it`s not practical. So a good rule for static meshes, modular or unique in udk is: make you uv channel 1 for diffuse and normals, overlapping and mirroring faces if needed etc. then make a second uv channel without overlapping faces for baked lighting (aka lightmaps)
The basic steps for light mapping environment pieces in UDK are explained on the UDN site http://udn.epicgames.com/Three/LightMapUnwrapping.html
It's generally accepted that (while using normal forward rendering) all static assets should have a lightmap. If you don't give assets light maps they will attempt to bake lights on the vertex level which becomes drastically more expensive in the memory.
If an engine is using all dynamic lights, it is probably using a differed lighting solution and light maps aren't required. UDK has the option for differed lighting but only in DX11 mode.