As mentioned elsewhere, the goal of this library is not to render graphics, but maybe? Loading NWN models is for the purpose of conversion or in some other asset pipeline.
See the mudl model viewer side project for work on a NWN model viewer.
Most default NWN models can be parsed without errors.
The parser can parse at about 100mb/s and can read pretty much all ascii models in ~20s on 2015 MacBook Pro.
The beginnings of binary model parsing is in the library. It's hard to tell what's right and what's wrong until there is more rendering experience.
.. tabs:: .. code-tab:: python #! /usr/bin/env python from rollnw.model import Mdl import os mdl = Mdl.from_file("my_ascii_model.mdl") if not mdl.valid(): print(f"failed parsing: {f}") print(mdl.supermodel_name) .. code-tab:: c++ // TODO