I have been asked by several content creators recently about what they will need to do to make content for Trainz: A New Era and our new graphics engine, currently codenamed 'E2'. While we have no E2 specific tools or techniques to discuss at the moment, there are things content creators can do when making content for current versions to get the most out of the backwards compatibility that E2 offers with current version content. In particular:
LOD
Make good use of LOD. Look at the posts here for a good description of LOD and how to use it. In particular, note that optimising the lower LOD levels for reduced polycount is a lot more important for performance than optimising the polycount of the highest LODs.
Texturing and Material use
Use the best material you can. If you don't need alphatransparency or masking, that will mean using the .m.tbumpenv material. This provides four separate channels across two texture files - an RGB diffuse, a per-pixel environment reflection strength, a normal map, and a per-pixel specular highlight strength. Make good maps for all of them. In particular, objects with good normal maps will benefit far more from improved lighting conditions than objects which do not have normal maps.
Avoid cardboard cutout objects
Two dimensional mesh planes with pictures of three dimensional objects textured on will look a lot worse in E2 than they do in Trainz now. None of the lighting and shadowing will work properly without real three dimensional objects to work with, and any remaining two dimensional objects will be clearly shown up as two dimensional.
Avoid modelling shortcuts
It is tempting to cut a few corners during asset creation. As an example, maybe you've decided that the inconsistencies with lighting from heavy use of 'double sided' texturing are tolerable versus the extra effort required to work out why some of the small parts on your model export inside-out and to fix them so they export properly. With improved lighting and shadowing from E2 you can expect this kind of problem to become a lot more obvious. You should instead fix the face normals properly in the source mesh file, so the object is not inside out when exported and doesn't need double sided texturing.
Don't push the envelope too hard
The community occasionally surprises us with content that pushes hard at the boundaries of what is possible in Trainz to achieve something that isn't officially supported. Unfortunately, Trainz history shows that some of these assets turn out to be incompatible with future Trainz versions. So as tempting as it is to push the boundaries and try to get the absolute most functionality possible, if your aim is to have as smooth a transition to E2 as possible, you might want to build content with a more conservative feature set, and avoid some of the more advanced community-developed features.
LOD
Make good use of LOD. Look at the posts here for a good description of LOD and how to use it. In particular, note that optimising the lower LOD levels for reduced polycount is a lot more important for performance than optimising the polycount of the highest LODs.
Texturing and Material use
Use the best material you can. If you don't need alphatransparency or masking, that will mean using the .m.tbumpenv material. This provides four separate channels across two texture files - an RGB diffuse, a per-pixel environment reflection strength, a normal map, and a per-pixel specular highlight strength. Make good maps for all of them. In particular, objects with good normal maps will benefit far more from improved lighting conditions than objects which do not have normal maps.
Avoid cardboard cutout objects
Two dimensional mesh planes with pictures of three dimensional objects textured on will look a lot worse in E2 than they do in Trainz now. None of the lighting and shadowing will work properly without real three dimensional objects to work with, and any remaining two dimensional objects will be clearly shown up as two dimensional.
Avoid modelling shortcuts
It is tempting to cut a few corners during asset creation. As an example, maybe you've decided that the inconsistencies with lighting from heavy use of 'double sided' texturing are tolerable versus the extra effort required to work out why some of the small parts on your model export inside-out and to fix them so they export properly. With improved lighting and shadowing from E2 you can expect this kind of problem to become a lot more obvious. You should instead fix the face normals properly in the source mesh file, so the object is not inside out when exported and doesn't need double sided texturing.
Don't push the envelope too hard
The community occasionally surprises us with content that pushes hard at the boundaries of what is possible in Trainz to achieve something that isn't officially supported. Unfortunately, Trainz history shows that some of these assets turn out to be incompatible with future Trainz versions. So as tempting as it is to push the boundaries and try to get the absolute most functionality possible, if your aim is to have as smooth a transition to E2 as possible, you might want to build content with a more conservative feature set, and avoid some of the more advanced community-developed features.