Aura™ Pipeline considers 3 core metrics of a model:
The number of visible triangles. A high number of triangles leads to lower FPS (frames per second), i.e. lower rendering performance.
Drawcalls are the number of unique operations the GPU needs to perform to render the model. A high number of drawcalls leads to lower FPS (frames per second), i.e. lower rendering performance.
High rendering performance is key as it frees up the CPU on mobile devices to focus on other calculations, such as Augmented Reality positioning. Bad rendering performance negatively impacts Augmented Reality positioning.
Storage size refers to the file size of all meshes in the model. A model consists of more than meshes but usually meshes take longest to load. Smaller storage sizes lead to faster load times.
But what about textures and materials? Models with large or a high number of textures and materials will take longer to load than models with fewer textures and materials. Aura™ will also optimize those aspects of any model. However, models and materials usually can be stripped without breaking the model. So, this metric focuses on what we can strip: mesh data.
Complexity is the a score based on the overall number of unique objects and the maximum hierarchy depth.
A complex model is one with either many unique objects or a very high hierarchy depth (lot of nested objects). A simple (i.e. "non-complex") model is either one with fewer objects or a medium number of objects but rather little nesting.
Complexity does not directly influence file size or rendering performance.