Workflow for VR asset creation and questions about high poly / low poly modeling
Hello I've been creating some assets recently as practice for a future VR game we're planning and I've been struggling to understand what workflow would be optimal for this.
There's high poly to low poly baking, low poly to high poly subdivision to low poly baking, low poly with bevels and weighted normals to baking, or mid poly to baking.
I currently have a handful of low poly models that range from between 5-15k tris
I'm quite happy with their appearance and would like to optimize them for substance painter, then to be imported in unity. Could they be used in their current state, or is it essential to prioritize creating a high poly object to get the normals from?
I've simply used normal box modeling to build detail until they looked how I wanted as that's what I've learned from blender courses from gamedevtv, grant abbitt and others on youtube, then I learned that game devs usually bake their normals from high poly models, I then tried to take my models and try to upscale them with the subdivision surface modifier on each separate part, but my topology doesn't seem to like upscaling as it just glitches out. So now I'm left wondering if my models are even worth keeping or if I should just redo them with high poly modeling in mind.
If I'm modeling a cup or a book or something else simple does that also need a high poly version so everything stays consistent? Is it even worth creating a ton of unique textures in substance if they eat up all the draw calls?
I'm also confused as to what's considered sufficiently low poly for VR Development. Most of the games in VR look really flat with no detail in anything so I might be way too ambitious in my modeling and should maybe stick to highly simplified low poly objects with flat shading from a texture atlas like in mobile games.
My experience is only with stylized low poly without smoothing "synty"
style, but haven't seen this in any VR games, is it because the edges
look bad?
Our game will consist of multiple small open areas linked together with
portals so there shouldn't be too much to render at once.
If anybody knows of a course dedicated to VR game assets please send a link. Creating models isn't that hard but this is all a big headache
If anybody knows of a course dedicated to VR game assets please send a link. Creating models isn't that hard but this is all a big headache
Replies
one reason a lot of VR games look flat and simple is that the the main contributor to player comfort is having a consistent high frame rate at a high resolution - when you're targeting a mobile device like quest with a pre-packaged engine like unreal or unity and don't have a bunch of experience profiling it's a pretty tall order to get the absolute best out of it and the sensible thing to do is aim low and concentrate primarily on gameplay / presence
you can throw quite a lot of geometry at a mobile device provided you have the resources and knowledge required to optimise the content but it takes time and thus money
also : there's a few threads on here regarding VR assets from the last few years so it's worth a search. I've written a few replies myself about the comfort research and some of the art direction tricks we used in the run up to the original PSVR launch - that stuff is largely valid still
so ..
The first thing I'd do is block everything out in basic geometry and put it in the game to make sure you have your scale and geometric detail levels worked out.
Doing anything else before you've confirmed all that is a huge waste of time and money.
Tricount is a shit way to measure the cost of an asset - everything has too many triangles if it's small enough on screen.
Use exactly as much geometry as you need for your view distance and make sure you're not rendering anything you don't - the same principles as apply to any modelling.
Texturing was a mixed bag on that project - It was Photoshop / Photoshop on top of Designer work for most environment stuff. We shipped in 2014 so Painter hadn't released or we'd have used that as well. I wouldn't do it the same way now because I've spent the last 10 years developing material pipelines and have learned new stuff.
It doesn't make any difference though - textures are textures.
Most of the specialist stuff to do with VR is around player comfort and performance - some of that comes from high level art direction and some from tech/design.
In terms of the art, the main thing to keep an eye on is consistent scaling of visual information - it's far more important in VR than on a flat screen that things are the 'right size' relative to everything else and you can't use some of the tricks you would outside of VR (like artificially high ceilings)