#leadmodels2024

My older way of model optimization suffers from adding a manual outline, the method called Inverted Hull (basically duplicating the mesh, flipping the normals, and sizing it up slightly with a colour change). This was to prioritize legibility on Quest since his material is flatly white with no way to tell depth on his face with his teeth and nose. Having the global decision for a manual outline was to have cohesion across all versions/platforms (and not having to export a "quest" version, which would multiply the work time whenever a change to the base model is made).

Going forwards I may rid of the Inverted Hull outline completely on any future models to use the built in one provided in Poiyomi shaders instead. This comes at the risk, though, of looking strange for fallback to strangers, or quest. As a plus, though, it means less work for making mesh adjustments and having to re-add the outline again and again, and can be customized more freely with a shader.

A huge thing that isn't thought about is the amount of free time a creative has to do ambitious stuff, especially if they've got health problems or work full-time (me). Shortcuts matter to hell and back. Optimization and tri count still isn't on my priority list of things to watch over when a majority of my VRChat outings are private, smaller instances anyhow. Even then, I still wish to see how far I can push the amount of detail I can get out of the same polycount or lower, and cull unneeded spots.