Understanding UE5 and workflow
polycounter lvl 9
Ive been using UE4 for some years now and just about to start a new project in UE5. The one thing Im confused about is people say "take models right from ZBrush into UE5" - but what about the whole texturing step? Am I really UVing a 2 million polygon mesh from ZBrush and texturing it in Substance? I guess Im just confused about how and when to use Nanite. Am I still baking normal maps for my meshes? Nanite just has me confused about my workflow now.
Replies
thats a good question 😆 was wondering bout that too
I imagine the objective of the showcase/advertising projects from Epic Games is to push the extreme, not to create a shipping game. For a UE5 project you can create assets pretty much the same as you would for a UE4 project.
For your own projects, I would ensure good performance when creating the content by setting the specs of your assets accordingly. If you want ditch normal maps for shading objects, use enough geometry to get a well defined silhouette with nice shading (decimate geometry on flat surfaces). With the increased geometry you can bake masks into vertex color and then combine with tiling textures in the shader. Should you have to unwrap dense meshes, could use subdivision levels.
I would just start a UE5 project and experiment. Definitely recommend reading the documentation to learn about the intention and current limitations. Iirc Nanite is ideally used for all static geometry - unless you can't because of its current limitations (opaque material only, no displacement/wp offset in shader, no vertex paint in editor, nanite meshes don't render to runtime virtual texture, ...). Plenty of solutions still missing. Here is a video from the Coalition about their experience with UE5 test project (surely there are more recent examples).
In the end, much depends on your project. Game (genre, platform?) or an art showcase with rendered video and images?
nanite is good for these reasons
1: it instances extremely efficiently
2: geometry render cost is far more closely tied to screen resolution than the amount of geometry in the model. This means you don't need lods and you don't have to optimise to the degree you would when making a mesh for a conventional system.
You can still model, UV and texture in the traditional way and in most cases that is far more efficient in terms of actually building stuff.
@poopipe thanks for the answer. still a bit confused as to when to use nanite or not? and when using nanite, am I just texturing a higher poly decimated mesh that allows me to forego a normal map?
in almost all cases you probably want to use nanite if you're targeting a platform that supports it (i.e current gen onwards)
you can take a standard game asset from your unreal 4 project with all it's materials, textures etc. convert it to nanite and it will work perfectly fine (provided it doesn't feature any of the unsupported features) .
you do not need to adjust the way you build things
you can save yourself work by skipping optimisation passes and lod creation
you can forego the normal map or you can choose to use one.
Using nanite is probably the only situation where the mid poly workflow isn't a shit way to build things