What problems are you facing?

#11 introduced Orama.RHI for abstracting graphics APIs. Currently our only RHI backend is a Veldrid fork which is pretty messy and hard to make changes to ourselves. This means we're very limited in what we can do with the RHI.

Describe the solution you'd like

Slowly phase out Veldrith with our own backend implementations, starting with Vulkan since the RHI API took heavy inspiration from it and we already compile shaders into SPIR-V bytecode.

Describe alternatives you've considered

We could stick with a Veldrith implementation and suffer the consequences of a somewhat locked RHI but the more we build with the limitations the harder they'll be to eventually remove and improve on.

Additional context

This might make it into v0.1 but if Veldrith is less buggy we'll ship with that.