As of current version, the loading interface looks like this:
pub fn load_obj<V: FromRawVertex<I>, T: BufRead, I>(input: T) -> ObjResult<Obj<V, I>>
so i'm like "yeah, i can plug my own vertex format too!", but when i look at the FromRawVertex trait, i see this:
pub trait FromRawVertex<I>: Sized {
/// Build vertex and index buffer from raw object data.
fn process(
vertices: Vec<(f32, f32, f32, f32)>,
normals: Vec<(f32, f32, f32)>,
tex_coords: Vec<(f32, f32, f32)>,
polygons: Vec<Polygon>,
) -> ObjResult<(Vec<Self>, Vec<I>)>;
}and the vertex from raw conversion in done in an opaque way inside the library, so if I wanted to implement my own vertex format, I would have to re-write all this conversion code myself or copy-paste it from the source.
A better way support user-defined custom vertex formats would be proving a trait like:
pub trait FromRawVertex {
type Output;
fn convert_vertex(position: [f32; 4], normal: [f32; 3], tex_coord: [f32; 3]) -> Self::Output;
}that can be used convert each vertex individually, then have a standalone conversion function like:
fn process_vertices<T: FromRawVertex, I>(raw: RawObj) -> ObjResult<(Vec<T::Output>, Vec<I>)>
Maybe i'm missing some stuff, but that's what I understood so far.
As of current version, the loading interface looks like this:
so i'm like "yeah, i can plug my own vertex format too!", but when i look at the
FromRawVertextrait, i see this:and the vertex from raw conversion in done in an opaque way inside the library, so if I wanted to implement my own vertex format, I would have to re-write all this conversion code myself or copy-paste it from the source.
A better way support user-defined custom vertex formats would be proving a trait like:
that can be used convert each vertex individually, then have a standalone conversion function like:
Maybe i'm missing some stuff, but that's what I understood so far.